Luminous Landscape Forum
Raw & Post Processing, Printing => Printing: Printers, Papers and Inks => Topic started by: Mike Guilbault on June 05, 2011, 11:26:29 pm
-
I've been doing some reading on print resolution, namely the 360ppi for Epson printers. I'm generally printing directly from Lightroom and have the Print Resolution set at 360 and using an Epson 4900.
From what I understand, if the native resolution after sizing the image in LR for the page is below 480, then adding 50% to this resolution will help/improve the quality of the output (in most cases). So for example, I size an image and the native resolution comes out to 252ppi for the dimensions I want. Does this mean to change the setting in LR's Print Module, Print Resolution from my normal 360, to 378? Or do you have to send it PS first, resize it to 378, back to LR and then leave the Print Resolution setting off (or does that actually do the same thing)?
I just want to make sure I understand correctly.
And, if the resolution is below 360 even after this, do I leave it alone or do I set the Print Resolution in LR to 360ppi again?
Thanks,
-
Mike
In essence, what you should be aiming for is to reduce the number of times that the image file is resampled (up-rezzed or down-rezzed) to as few as possible. Each time it is resampled some loss of IQ will occur, inevitably. Unless you are printing the original image to a size which exactly represents 360 ppi on paper (or whatever your printer's output is) you can never avoid at least one interpolation. So in LR we should be aiming to send the file to the printer at the printer's inbuilt resolution - for most Epsons this is 360 ppi and for many printers you can output at 720 ppi as an alternative. What we don't want is for LR to resample the file to something the printer can't output at (say 293 ppi) and thus force the printer to resample it yet again before printing. So if LR is telling you that the resolution for that image size is below 360, set the resolution to 360. LR will resample it once and the printer will just output the file without further processing. Similarly, with most Epsons if LR reports that the output resolution should be above 360 ppi, set the output to 720 ppi - but make sure that you have the highest quality output enabled in the printer driver, otherwise this is pointless and it will just print at 360 anyway. For other printers the ppi will be different - HP is 300 ppi I believe.
John
-
With the complexity of the screening and dithering algorithms built into modern Epson printers the concept of "native resolution" has become irrelevant. It's been overtaken by other functions for dot placement. Avoid resampling the image altogether. Print at anything between 240 and 480 PPI. The printer will throw away information exceeding 480, and you could see some deterioration of print quality below 240. Generally, higher resolutions are more helpful for smaller printers and lower resolution for larger ones, simply because of differences in viewing distance. You can check all this for yourself by experimenting and observing.
-
Yes and I believe this 'native resolution' legend was so very thoroughly discussed a few months ago...
http://www.luminous-landscape.com/forum/index.php?topic=48894.0
-
Thanks for the reminder - and yes I stand corrected - nowadays Epson printers can work with up to 720 (I had forgotten because I never get into that stratosphere), but I can't see any difference to anything North of 360, and quite frankly looking for quality difference between 240 and 360 can also be challenging. Much South of around native 240 however one can begin to see evidence of IQ deterioration comparing a print made at native 360 or above. I say this based on some tests I once did on my 4800, and repeated on the 3800.
-
Jeff and I have just done a very thorough examination of this in our new Camera to Print and Screen video tutorial, which will appear later this summer.
In brief, needing an exact multiple of some number (360?) is a myth. The printer's dithering algorithms simply take whatever data you send it and do what it does. Too little and print resolution will be down, and too much and (with Lightroom at least) it automatically resses down to 720ppi so that the printer doesn't choke.
How much is too little and how much is too much? We'll it depends on the printer.
Jeff Schewe, who consults with Adobe, and whose sharpening algorithms are in Lightroom believes (and we demonstrate on camera) that you can res up a file in LR by up to 50%, and actually improve output appearance. Epson printers these days are 720ppi. Canon and HP printers 600ppi.
As far as how much is too little, it very much depends on the size of the print and the distance from which it will be seen. A 18oppi print might look lacking in detail on a 11X17" print held at arms length, while on a 36X42" print over the mantelpiece it could look fantastic.
Everyone should do their own tests. All it takes is a bit of time, ink and paper. Trust your own eyes.
Michael
-
And my experience with the HP and Canon printers is you can do the same at 150 ppi for very large things and achieve quite smooth dither and edge sharpness from a good file, at viewing distance, if you are working with a file size that demands that, say a 40x60 from a 5D Mark 2 as an example where you really don't have a choice.
Questions I would like to know is just what the differences are between using Lightroom to output these large files as apposed to Photoshop CS5, Q-Image, or the Canon plug-in which is also excellent and has sharpening adjustment capability (as does Q-Image). In other words how much of the dither control is happening in the printer driver and how much in the other software used to send it ( apart from sharpening algorithms) these days .
-
Trust your own eyes.
Michael
That's the problem... I don't trust my own eyes! :o My vision isn't what it used to be so I want to print the best quality I am capable of so that the viewer is satisfied, not just me. Although, I am probably more critical than the average viewer of my work.
So, based on John, Mark and Michael's comments I'm getting closer - and please correct me if I'm wrong in these summaries of what's been said for my own understanding: John's 'method' is basically what I've been doing - set the PR in LR to 360 and be done with it. Mark says use the native resolution if it falls between 240 and (now) 720 and just let the printer do it's thing. And Michael/Jeff says that if the native resolution is below 720, up-res it by 50% for some IQ improvement (based on the actual image of course).
It seems that John's method is the simplest solution, but Mark's also makes sense, as long as there is no noticeable loss in quality at the lower (below 360?) resolutions (so you have to test). And as John says, the fewer up/down res-ing we do, the better so this again makes sense. But in some cases, up-resing 50% may improve the quality if the native resolution is below 720. Again, a test may be in order here.
Does that basically sum it up? I sure hope so! ;)
-
That's the problem... I don't trust my own eyes! :o My vision isn't what it used to be so I want to print the best quality I am capable of so that the viewer is satisfied, not just me. Although, I am probably more critical than the average viewer of my work.
So, based on John, Mark and Michael's comments I'm getting closer - and please correct me if I'm wrong in these summaries of what's been said for my own understanding: John's 'method' is basically what I've been doing - set the PR in LR to 360 and be done with it. Mark says use the native resolution if it falls between 240 and (now) 720 and just let the printer do it's thing. And Michael/Jeff says that if the native resolution is below 720, up-res it by 50% for some IQ improvement (based on the actual image of course).
It seems that John's method is the simplest solution, but Mark's also makes sense, as long as there is no noticeable loss in quality at the lower (below 360?) resolutions (so you have to test). And as John says, the fewer up/down res-ing we do, the better so this again makes sense. But in some cases, up-resing 50% may improve the quality if the native resolution is below 720. Again, a test may be in order here.
Does that basically sum it up? I sure hope so! ;)
I think you've got a fair handle on what you've read here, largely non-conflicting advice. Whenever I think maybe I shouldn't trust my own eyes, I ask my better-half to just look at the prints and tell me whether she sees any difference, without explaining anything. Then, if I think she's off on a tangent, I tell her what to look for. Those two tests combined with my initial observations are usually sufficient.
-
Does that basically sum it up? I sure hope so! ;)
No, not really...here's a simpler guideline for Epson printers; if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.
For Canon and HP the numbers are 300/600PPI...
-
No, not really...here's a simpler guideline for Epson printers; if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.
For Canon and HP the numbers are 300/600PPI...
Which is exactly what I said - but perhaps I didn't say it simply enough . . .
;) John
-
Which is exactly what I said - but perhaps I didn't say it simply enough . . .
Not really...you got wrapped up in the question of resampling and interpolation–incorrectly I might add since print drivers don't really interpolate the image data they receive. The drivers simply convert whatever image resolution they receive into a dither for determining what droplets of ink to print and where.
-
Here’s the test I just did to an Epson 3880. Image was shot on a 5DMII, ISO 100 of some very fine, high frequency subject. Printing at 10.6X7.1, Photoshop reports that native resolution of the un-cropped data is 530ppi.
I printed out of LR with the Resolution check box off. Then on set as Jeff recommends in this case to 720. For my eyes, it took a loupe to see the differences. But the 720 output had a slight edge. For me, it was real pixel peeping but a difference does result.
-
Printing again but letting the check box default to 240 (on), now under the loupe, the differences are pronounced as one would expect. The 530/720 resolves fine lines much better.
FWIW, output was done on Premium Luster paper.
-
When you need to examine an enlargement with a loupe to see any difference of print quality within such a broad range of resolution settings, it tells me that within the ranges being discussed here, this isn't much of an issue with today's technologies, and therefore perhaps not worth obsessing over.
That said, a question for Jeff (other also welcome to respond): let's say your native resolution is what Andrew started with - 530 PPI. There are three options for printing it: (1) leave it alone; (2) downsize to 360, (3) upsize to 720 (Epson case). What is "theoretically best"? There would appear to be some possibly conflicting principles here, or maybe one or more just aren't true any longer: (a) the less you resample the better the IQ; (b) if you resample you do less apparent harm to IQ by downsampling rather than up-sampling; (c) if your Epson printer can print at 720, it's best to take advantage of all that resolution. ???? Different principles give different answers.
-
240 versus 530 under a loupe I would expect the differences to be visible. What about without the loupe?
-
240 versus 530 under a loupe I would expect the differences to be visible. What about without the loupe?
In this particular image, a bit difficult to see without the loupe unless you know where to look. Its a shot of some buildings where many have windows with venetian blinds that resemble at a distance very, very fine ruled lines in the print. I see the differences there. With 240dpi, they blur together more, almost like a moire where with the higher rez, they look clean and distinguishable.
But while I could be happy with either print at viewing distance or even up to my nose (this is only an 8x11), I think what I learned is not to be careless with the check box and its default. I suppose, based on this one test, if I know I have a higher rez image going to a smaller print size, just leave the check box off as in this case, default 240 and on just wasn’t as sharp. Would I open up Photoshop or pull a calculator based on all kinds of print sizes to see if the image is above 360? Not sure although I’d certainly consider doing the math for my print templates as once I build em, I can set the check box and resolution appropriately. It would be useful if LR told us what the current native resolution is when the check box was off so we could more easily decide what to do.
-
Thanks. Fair enough.
-
With my Epson 9800 my rule of thumb is to send an image of 200 to 400 for paper and 150 to 400 for canvas. I can't tell the difference between prints in those ranges.
What I find of immense importance in image quality is the degree of sharpening. All of my prints within the above ranges using NIK's output sharpening are far superior to what I was printing before I used NIK.
-
All of my prints within the above ranges using NIK's output sharpening are far superior to what I was printing before I used NIK.
What did you use before NIK?
Terry.
-
That said, a question for Jeff (other also welcome to respond): let's say your native resolution is what Andrew started with - 530 PPI. There are three options for printing it: (1) leave it alone; (2) downsize to 360, (3) upsize to 720 (Epson case). What is "theoretically best"?
It would be foolish to down sample from 530PPI to 360. Why throw away perfectly good pixels? The only two viable options are leave it at 530 or upsample to 720PPI. Depending on the image detail, the upsampling to 720PPI (for Epson) is the optimal option and since this is all done inside of Lightroom using it's unique resampling (which Photoshop can't do) and there's no real downside since you don't have to spawn off a file iteration, why not do it? The only slight downside is it takes a bit longer for LR to process the image data and send it to the printer and a bit longer to spool the file. The actual printing time seems the same.
Also, this all presupposes that you are printing to media, generally a photo paper not matte and that you are using the highest res setting in the driver.
-
OK, based on current LR technology that all makes perfectly good sense. Thanks.
-
No, not really...here's a simpler guideline for Epson printers; if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.
Which is exactly what I said - but perhaps I didn't say it simply enough . . .
;) John
That's what I gathered from your post John, and basically what I was doing. I'm glad Jeff concurs with this... just strengthens the reasoning and is what I've been leaning towards as well.
And yes, now I'm interested in the sharpening questions just posted. I usually sharpen to my liking in LR and then add low sharpening for the print.
-
I printed out of LR with the Resolution check box off. Then on set as Jeff recommends in this case to 720. For my eyes, it took a loupe to see the differences. But the 720 output had a slight edge. For me, it was real pixel peeping but a difference does result.
Hi Andrew,
I think it is important to stress that this is an LR workflow result. For those who print from e.g. Photoshop, there is more contol over the type and amount of sharpening at 720 PPI, both resulting in smoother (because interpolated) gradients and potentially sharper looking results (more sharpening at the output pixel level is possible if the subject benefits from it). Sharpening at the highest spatial frequencies also boosts sharpness at slightly lower frequencies, but with less visible artifact risk.
The control LR offers regarding sharpening at the native output resolution of the printer, is limited.
Cheers,
Bart
-
The control LR offers regarding sharpening at the native output resolution of the printer, is limited.
Intentionally so because trying to sharpen for output is not a visual activity...it's trial and error at best and believe me when I said a lot of trees died to make Lightroom's output sharpening work well. Unless you can state exactly what you think is missing from LR's output sharpening, I don't see the above statement useful. It's limited to those sharpening routines we think are useful. Facts would be better than opinion...
-
Intentionally so because trying to sharpen for output is not a visual activity...it's trial and error at best and believe me when I said a lot of trees died to make Lightroom's output sharpening work well. Unless you can state exactly what you think is missing from LR's output sharpening, I don't see the above statement useful. It's limited to those sharpening routines we think are useful. Facts would be better than opinion...
Hi Jeff,
I'm not sure, but maybe you interpreted my remark in the sense that LR sharpening is bad, or something. All I said is that the control over the sharpening is limited. You ask for facts because you value them higher than my opinon, I'll give you facts.
How about sharpening for printers like a Noritsu or Fuji frontier? Laser printing on photochemical paper is different than printing on an inkjet printer, it can benefit from a different type of sharpening, and they do not all have the same native PPI resolution.. The same goes for certain types of inkjet media, Canvas knows many different structures, and it's quite different from Glossy which is different from non-glossy paper. Perhaps we want to print on non-glossy paper, yet convert it to glossy with a Diasec treatment, or coat it with resin. Or perhaps we need to print it on self-adhesive material or ... I do know there is some limited control/choice in LR, but then I didn't say it was bad.
Another fact is that interpolation, upsampling, usually doesn't add detail (and hopefully few visible artifacts). What few people seem to understand is that the upsampling itself can be mathematically characterized by a Point Spread Function (PSF). That fact can be exploited by using deconvolution sharpening after upsampling. It can even be taken as far as skipping the capture sharpening, thus avoiding any risk of artifacts and enlarging those, and combining the capture sharpening with the output sharpening with a (somewhat timeconsuming) restoration of detail, called deconvolution at the native print resolution. This would help in producing huge magnifications e.g. for tradeshow booths. A good friend of mine shoots & produces huge (6x3 metres or larger) background images each year, often from a single 21 MP 1Ds Mark III frame, and they will be viewed from close up when people come to the stand after stopping in their tracks as they come walking down the aisle when they see the images.
I'm glad you now agree that upsampling to 720/600 PPI has it's merits, after an initial reluctance against my long time suggestions that it can produce better output (depending on subject), because it offers more control (e.g. better interpolation than the printer driver and sharpening at the native reolution). So let's now keep an open mind that while LR has improved (good enough for many), there may still be a need for more user control (perhaps outside of LR, yes there is a life outside LR ...).
Cheers,
Bart
-
I'm glad you now agree that upsampling to 720/600 PPI has it's merits, after an initial reluctance against my long time suggestions that it can produce better output (depending on subject), because it offers more control (e.g. better interpolation than the printer driver and sharpening at the native reolution). So let's now keep an open mind that while LR has improved (good enough for many), there may still be a need for more user control (perhaps outside of LR, yes there is a life outside LR ...).
Cheers,
Bart
While I use that approach too I still wonder what the aliasing effect could be if the actual optical resolution possible in the print (paper/ink/printer/software) is already available/overdone with 360/300 PPI input. When there is image content that could degrade with aliasing I think dealing with it (a good, flexible anti-aliasing filter) in the downsampling of the application (or driver but I doubt it exists there) is a better plan than upsampling and sharpening the image to 720/600 PPI. The driver will not downsample then and aliasing must happen at the printing stage. Qimage (there is life outside LR) can do that downsampling anti-aliasing but it can be a slow process on big files so I tend to set the print quality higher in the driver that it asks for 600 PPI and avoids downsampling. It may not be the best choice for all images.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
The Canon plug-in for their LF printers has a built in slider for output sharpening similar to Q-Image ( along with its 16 bit output capability). Apparently you can test and save those adjustments for the future as a preset for your Specific media - such as creating your own presets for brand and texture of canvas, type and degree of textured rag, fabrics, super gloss media or fiber gloss media apart from "satin" media etc. It looks interesting but I have no idea how effective it is for output sharpening compared to other software that's been around for awhile. It does avoid the two category approach - generic photo or matte papers only . Does anyone who has the 8300 know if it is effective in the way the Q-Image output sharpening is? If so I'm very happy to have it available on the fly the way it is.
john
-
It would be useful if LR told us what the current native resolution is when the check box was off so we could more easily decide what to do.
In the "Rulers, Grids and Guides" panel (right side of the print module), check both "Show guides" and sub-item "Dimensions".
-
It would be foolish to down sample from 530PPI to 360. Why throw away perfectly good pixels? The only two viable options are leave it at 530 or upsample to 720PPI. Depending on the image detail, the upsampling to 720PPI (for Epson) is the optimal option and since this is all done inside of Lightroom using it's unique resampling (which Photoshop can't do) and there's no real downside since you don't have to spawn off a file iteration, why not do it? The only slight downside is it takes a bit longer for LR to process the image data and send it to the printer and a bit longer to spool the file. The actual printing time seems the same.
Also, this all presupposes that you are printing to media, generally a photo paper not matte and that you are using the highest res setting in the driver.
Jeff, just reviewing this contribution - the one aspect of it that causes me a bit of a lingering concern is the proposition to upsample say from 530 to 720. I think you would agree that if there is no need to upsample for getting the print size one wants at satisfactory resolution, there's no point doing it. Regardless of how capable LR's upsampling math may be, does it still remain correct that upsampling is inventing information, therefore it is an approximation of immediately contiguous data, and therefore may not yieldt quite the same image quality as achieved without the upsampling? If the answer to that question is "Yes", then follows the next one: you need to make a very big print relative to the native resolution of your file, which for sake of discussion here is say Andrew's 530PPI. If you don't resample when you change the linear dimensions, PPI goes down to 180, which I think we may agree is usually below a safe lower limit for viewing large prints at their appropriate viewing distance (we're not talking about using a loupe here). So we'll resample - say from a comfort level of 300 to a maximum of 720. Does it remain fair to expect that (i) the upsampling from 180 to 300~360 will definitely improve IQ and (ii) beyond the range of 300~360, the more one upsamples, the less the incremental return in terms of IQ, or that it may even turn negative at some level?
-
In the "Rulers, Grids and Guides" panel (right side of the print module), check both "Show guides" and sub-item "Dimensions".
I always have that set. It does show you dimensions but at what native PPI of the image, such we can use the new “Schewe rule” for setting the Resolution field (if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.).
-
While I use that approach too I still wonder what the aliasing effect could be if the actual optical resolution possible in the print (paper/ink/printer/software) is already available/overdone with 360/300 PPI input.
Hoi Ernst,
You are correct that aliasing artifacts are lurking around the corner when there is more high frequency detail than can be resolved. However, when we upsample, that by itself will not lead to aliasing because we acually reduce the amount of detail per pixel. Another thing that helps is that aliasing manifests itself best in an ordered grid, but the (inkjet) printer uses a dithering pattern which actually reduces the risk of those artifacts showing. But there are also other output technologies, e.g. RGB laser output on photochemical paper with resolutions from 254 PPI to 600 PPI, and for those it would indeed make a lot of sense (even more than for inkjet) to tune the output resolution to that native resolution. Especially for the lower PPI machines the risk of creating downsampling/resampling artifacts in an ordered grid is very real. There is an increasing possibility of being confronted with high MP images (think high MP sensors, and/or stitching). Hence my plea for adaptability.
When there is image content that could degrade with aliasing I think dealing with it (a good, flexible anti-aliasing filter) in the downsampling of the application (or driver but I doubt it exists there) is a better plan than upsampling and sharpening the image to 720/600 PPI.
For most media, I don't think the risk of introducing aliasing related artifacts is a real concern. The most care should be given to downsampling and minor resampling of images (= regular grid) with lots of detail (sharp edges, fine lines, repetitive patterns at an angle, etc.). A proper resampling algorithm will automatically adjust it's resampling filter, a Lagrange type of windowed filter is a nice choice with some speed benefits over full blown Sinc windowed filtering.
The driver will not downsample then and aliasing must happen at the printing stage. Qimage (there is life outside LR) can do that downsampling anti-aliasing but it can be a slow process on big files so I tend to set the print quality higher in the driver that it asks for 600 PPI and avoids downsampling. It may not be the best choice for all images.
I'm not sure I follow what you're saying here.
I do remember that it took a bit of discussion before I convinced Mike Chaney that for downsampling one should do proper pe-filtering. He then deviced a specific type of AA-filtering (perhaps because of speed concerns) and added that as an (default) option to Qimage that's used when downsampling is involved. For upsampling the available algorithms offer a good trade-off between speed and quality. I still like Qimage's Hybrid SE resampling a lot, also because print file output responds well to deconvolution sharpening.
Cheers,
Bart
-
Jeff, just reviewing this contribution - the one aspect of it that causes me a bit of a lingering concern is the proposition to upsample say from 530 to 720. I think you would agree that if there is no need to upsample for getting the print size one wants at satisfactory resolution, there's no point doing it.
Hi Mark,
If you don't mind me responding to the question directed at Jeff ..., most printer drivers have dithering optimizations only at a few fixed final output resolutions (the printer driver even tells which resolution it wants, when asked, and they are fixed numbers, e.g. 720 PPI). For speed reasons that dithering will most likely not be calculated but indexed from a lookup table, hence the required preconditioning to a given PPI. There will always be resampling involved if not offered exactly that resolution in PPI. Either the printer driver upsamples (with some unknown black box method) or the application that sends the data to the driver does it (with a very likey much better method). Presumable both will add a bit of sharpening after the resample to output size, with the application offering various settings.
Cheers,
Bart
-
Hi Bart,
It's an open forum - I don't mind at all, and appreciate your contribution.
What you are saying about the printer sounds a bit like a scanner in the sense that it has fixed stages of resolution at which it operates. That in itself raises other interesting questions about how one optimizes its output between what the user can control and the printer and printing application do. As usual, the devil in the details?
-
I always have that set. It does show you dimensions but at what native PPI of the image, such we can use the new “Schewe rule” for setting the Resolution field (if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.).
This confused me too for the last few minutes until I realized that you have to uncheck the "Print Resolution" button (which I always leave checked :)) and then the native resolution figure appears next to the print dimension on the image.
Rob P
-
Bart,
What is the software that will allow me to use deconvolution sharpening after the upsampling? And under which circumstances would it be an advantage to skip capture sharpening in favor of deconvolution at the native print resolution? Well, maybe I do part of it already - I do not capture sharpen in the traditional sense of the word. I use Raw Developers deconvolution in the processing to TIF and no other capture sharpening. I have not printed yet.
> But there are also other output technologies, e.g. RGB laser output on photochemical paper with resolutions from 254 PPI to 600 PPI, and for those it would indeed make a lot of sense (even more than for inkjet) to tune the output resolution to that native resolution.
My plan is to have made prints of about 1 m on the long side on a Lambda or the like. The Lambda prints natively at 200 and 400 dpi. The Canon 5D2 files set at 200 dpi would print about 70 cm on the long side. What could I do to enlarge this to about 1 m?
A scaling factor of 1.5 applied in my editor (PhotoLine) would give 106 cm @ 200dpi. However, I have been advised to only use integers as a scaling factor and leave the rest to the print shop. Don't know what they will do.
Kind regards - Hening.
-
Every time I read Forum threads showing users into all kinds of issues over sharpening, I have to ask myself whether I'm missing something by continuing to work with the most flexible and refined sharpening application I've ever tested - Photokit Sharpener 2. Unless you're into reverse-engineering of the image, there's something for everything here, and it works.
-
How about sharpening for printers like a Noritsu or Fuji frontier? Laser printing on photochemical paper is different than printing on an inkjet printer, it can benefit from a different type of sharpening, and they do not all have the same native PPI resolution..
Have you tested the LR output sharpening when saving out an image in LR's Print to JPEG? What seems to be lacking? Sharpening strength or sharpening type? There is some traction towards improving the Print to JPEG output sharpening in the future so it would be useful to get users' feedback.
-
Jeff, just reviewing this contribution - the one aspect of it that causes me a bit of a lingering concern is the proposition to upsample say from 530 to 720. I think you would agree that if there is no need to upsample for getting the print size one wants at satisfactory resolution, there's no point doing it.
Have you actually tested this? "Thinking" about this stuff is ok...but it's best to actually test it in practice. It was through testing that the LR team determined that upsampling and then sharpening produced a better result particularly for images with a lot of high frequency textural detail. As Andrew indicates with his test, the results can be subtle and you have to know what to look for, but his test proved to him that absent a reason not to, upsampling then sharpening to 720PPI produced a better result. Until you actually do the tests yourself, you won't really know what your eyes will be telling you, right?
When Mike & I were shooting the Camera to Print & Screen video we picked an image from his GH2 camera with lot's of high frequency texture. The native rez was about 450PPI. He printed out an image at native then upsampled to 720PPI. Even with the naked eye (and also through a loupe) he could see the improvement in the resulting print–which was done on Premium Glossy paper. So, I think Mike is now convinced that taking a native rez image and upsampling can produce finer/better detail.
This whole thing came about because with my P65+ files, Lightroom was downsampling (wasting) resolution to 480PPI. So the decision was to increase the resolution cap in LR 2 to 720PPI in LR 3. Aside from slightly (and I mean slight) longer processing time and spooling time there's no real downside to doing this. Try it yourself...
-
I've previously tested a lot in the range of 180 to 480, but I have not tested LR3 at 720. It goes on the list........thanks for the heads-up.
-
I've previously tested a lot in the range of 180 to 480, but I have not tested LR3 at 720. It goes on the list........thanks for the heads-up.
I have tested the difference between 360 and 720 ppi from LR on my Epson R2400. To utilise the 720 file, you have to have "Photo RPM" selected in the 2400 print driver - for other Epsons, it will be called something different, I expect. There is a definite improvement in the rendition of fine detail using 720 ppi, as long as the "native" resolution reported by LR was above 360 in the first place.
John
-
Thanks for that John; when time permits I'll find a good high frequency image and test it on my 4900 and see what I see!
-
Mark
I'm always working in B/W, of course, so I don't see differences in colour rendition. The file I used to test was the one I attach here, and where I really noticed the difference between the two output resolutions was in the pattern of the cups on the table, the tablecloth, and the patterns in the subject's clothing. Oh yes, I forgot, the writing in the rosette on the van. It's subtle stuff like this which you are aware of as a kind of smoothness, rather than an upfont "Wow that's better" response. But as soon as you put a loupe on it you are convinced.
John
-
Understood.
The loupe is fine for pixel-peeping in the positive sense of seeing in fine detail what's going on. For more general purposes looking at the print at normal viewing distance is what really matters; so that is what I shall focus on most when I get to testing it. most likely on 13*19 inch prints.
-
. . . since print drivers don't really interpolate the image data they receive. The drivers simply convert whatever image resolution they receive into a dither for determining what droplets of ink to print and where.
Is this really true? Since this seems to be the crux of the issue. Bart seems to agree with me that whatever you send the printer will be resampled if it is not, say, 360 or 720 ppi (for Epsons). My impression was that ppi (pixels per inch as measured on the paper) and printer dpi (dithered output) are two quite separate animals, and that if you send an Epson 279 ppi it will be resampled by the printer's "black box" before it is passed to the dithering and screening department.
It would be nice to know if I am completely wrong on this, and if so, why.
John
-
This confused me too for the last few minutes until I realized that you have to uncheck the "Print Resolution" button (which I always leave checked :)) and then the native resolution figure appears next to the print dimension on the image.
Rob P
So it does. Very cool (and useful). I’m going to have to rebuild all my existing print templates with the checkbox off (easy to do, turn off, click Update settings).
On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value.
-
My impression was that ppi (pixels per inch as measured on the paper) and printer dpi (dithered output) are two quite separate animals, and that if you send an Epson 279 ppi it will be resampled by the printer's "black box" before it is passed to the dithering and screening department.
I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.
The resolution settings of 2880/1440 on the highest setting for the 79/9900 is not dots per inch, it's droplets per inch with each droplet being 3.5 pico liters in volume. Now, exactly how Epson's most recent dither works I can't tell you other than it was a joint effort between Epson and the Rochester Institute of Technology's Munsell Color Science Laboratory and supposedly isn't a standard error-diffusion process but some kind of exotic math to arrive at the dot dither.
But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.
-
On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value.
If the native rez is above 720 and shows the plus, it'll end up being down sampled to the current 720PPI max.
-
Have you tested the LR output sharpening when saving out an image in LR's Print to JPEG? What seems to be lacking? Sharpening strength or sharpening type? There is some traction towards improving the Print to JPEG output sharpening in the future so it would be useful to get users' feedback.
We are using Lightroom to output rendered jpeg files which are then printed on our Imetto printer (basically a chinese version of a Durst Lamda, laser based RA-4). We use it to print almost exclusively on Kodak Metallic Paper (very glossy).
LR seems to be hitting a sweet spot for us, results are good, better than without, better than the printers built in sharpening, don't see many artifacts (normally setting jpeg to 100 for better quality), normally using standard sharpening for glossy media type.
As far as improving it for jpeg rendering ... gonna have to leave that to experts to see if they can make it better. I'm not one of them.
-
The resolution settings of 2880/1440 on the highest setting for the 79/9900 is not dots per inch, it's droplets per inch with each droplet being 3.5 pico liters in volume. Now, exactly how Epson's most recent dither works I can't tell you other than it was a joint effort between Epson and the Rochester Institute of Technology's Munsell Color Science Laboratory and supposedly isn't a standard error-diffusion process but some kind of exotic math to arrive at the dot dither.
But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.
My only quibble with this, Jeff, is that it's not fixed at 3.5pl - at any given resolution (2880, 1440, 720, 360 etc) there are a number of dot sizes available. The choice varies depending on the resolution and multiple dot sizes are employed as part of the "exotic math". the LUT that was developed as part of all that gives choices from around 18,446,774 trillion colours (not a typo).
Otherwise, that's also my understanding.
-
My only quibble with this, Jeff, is that it's not fixed at 3.5pl - at any given resolution (2880, 1440, 720, 360 etc) there are a number of dot sizes available.
Actually, according to Epson when you select the 2880 rez option it ONLY prints with the smallest 3.5 pico liter droplet not the variable sizes...the variable droplet is only used for 1440/720.
-
Actually, according to Epson when you select the 2880 rez option it ONLY prints with the smallest 3.5 pico liter droplet not the variable sizes...the variable droplet is only used for 1440/720.
Yes, good point. 2880 mode is fixed, but all other resolutions are variable, and the choice of which variable sizes to use varies depending on other settings (media, microweave and so on). It's a very complex matrix.
-
On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value.
Yes, that's because the output resolution should be over 720 ppi for that print size, but 720 is the max that LR can output. You see this a lot with big MF images and small print sizes.
I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.
But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.
Thanks for that, Jeff. I am quite happy to be put right on this, but what I still don’t understand is – if the printer does not re-sample the incoming file to its own ppi grid, as it were, then why would you bother (as you yourself do) to specify exactly 360 or 720 ppi on the LR output side? According to your version of events, it wouldn’t make any difference if you sent the Epson 359 or 718 ppi, because the input would just be dithered, not re-sampled.
My apologies if I am just being dim ;)
John
-
The loupe is fine for pixel-peeping in the positive sense of seeing in fine detail what's going on. For more general purposes looking at the print at normal viewing distance is what really matters; so that is what I shall focus on most when I get to testing it. most likely on 13*19 inch prints.
Mark
Strangely enough, my own testing seems to show that other people often prefer prints with lesser quality rather than a technically higher one. I did an experiment on this a while back – I got some of my older MF B/W negatives scanned as reasonable quality TIFFs (3000x3000 pixels) and printed them via my Epson to the same size (about 10x8, quite small) as my existing wet darkroom prints of the same subjects. I tried to match the overall balance, contrast etc as closely as possible between the new prints and the old, so that they were a good match. The paper was as close a match as I could manage, too (Ilford Gallerie silver prints and Harman Baryta inkjet).
Almost without exception the people I showed them to preferred the inkjet prints from the scans rather than the darkroom prints. This was despite the fact that under a loupe the silver prints clearly had finer detail and smoother tonal gradations, especially in the highlight areas. It seemed to me that the viewing audience was responding to apparent higher levels of sharpness in the digital prints, a result of (quite subtle) USM. Typical comments were that the digital prints looked more “punchy”, “crisper”, or had “better detail” (when in fact they did not).
All of which was educational, but quite depressing.
John
-
I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.
Hi Jeff,
I've never heard an assumption like that before, but there's a first time for everything. We both don't have a solid proof (like an official statement by one of the major printer manufacturers) to support either position, but I think there is more circumstantial evidence to support simple interpolation. Not only is interpolation a logical simplification to speed up the dithering process, there are also moiré effects (http://www.luminous-landscape.com/forum/index.php?topic=54211.msg444078#msg444078) created that are completely consistent with (poor quality) print driver interpolation/resampling.
The printer resolution feedback from the printer driver itself when interrogated, is IMHO also a clear indicator that the fixed PPI settings are the basis for further dithering, and that interpolation is happening.
As a historical tidbit I found an old contribution by Mike Chaney (http://forums.dpreview.com/forums/read.asp?forum=1003&message=6759746) (developer of Qimage) on an other website, where he also explains in simple terms what happens and why it can help to interpolate to 720/600 PPI prior to sending the data to the printer driver.
I also seem to remember a post somewhere from a relatively credible source that hinted at improvements in some printer drivers using bicubic interpolation instead of bilinear, but I'd have to search and see if I can still find that post.
There is also a faq entry on Eric Chan's website (http://people.csail.mit.edu/ericchan/dp/Epson3800/faq.html#native_res) specifically mentioning the resampling issue in connection to the Epson driver settings.
Finally, it's important to remember that in order to avoid accidental downsampling artifacts by lower quality printer driver resampling, setting the printer driver to accept the highest possible input resolution (720/360 or 600/300 for the brand and paper choice) is the safe route to use. Substandard downsampling is a huge source for artifacts. That's why a program like Qimage adopted the use of anti-aliasing prefiltering for downsampling. The current LR version 3 also seems to have improved in that department. It's better to make sure what quality goes into the driver, than leaving it up to that black-box to find a quick and dirty solution.
Of course not all images are that critical that one will always run into resampling trouble, and not all images have enough detail to absolutely need the highest possible PPI setting. Nevertheless, I've seen no clear evidence that using 720/600 PPI input files has a detrimental effect on image quality, but 360/300 can have that effect.
But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.
May I suggest you check with Eric Chan what evidence he found that resampling is what is happening? I know he has exchanged views with Mike Chaney, but presumably also with others (not to be named individuals at Epson?). And perhaps (if he happens to read this) Eric can react here if he feels he can add something to clarify the situation.
Cheers,
Bart
-
Finally, it's important to remember that in order to avoid accidental downsampling artifacts by lower quality printer driver resampling, setting the printer driver to accept the highest possible input resolution (720/360 or 600/300 for the brand and paper choice) is the safe route to use. Substandard downsampling is a huge source for artifacts. That's why a program like Qimage adopted the use of anti-aliasing prefiltering for downsampling. The current LR version 3 also seems to have improved in that department. It's better to make sure what quality goes into the driver, than leaving it up to that black-box to find a quick and dirty solution.
Cheers,
Bart
Bart,
The rest of your message is along the lines of what I would have written. It would be a strange thing that Qimage reports requested input PPI numbers like 300 and 600 PPI for HP and Canon models and 360 and 720 PPI for Epsons including the latest models if the drivers could cope with any input resolution straight away.
I am not completely happy with the part above. The point is that if aliasing can happen with the original image resolution at print size versus a lower optical resolution of the printing process then anti-aliasing should be done. By upsampling no extra aliasing conditions are created but it disguises the fact that aliasing still can happen as the 600 or 720 PPI does not represent the optical resolution of the printing process, the last will be lower. The same low optical print resolution may already be available with 300 or 360 PPI input. The example given of 560 PPI input then could give a better print if a good anti-aliasing filter was used to bring it down to 300 or 360 PPI instead of laying down 600 to 720 PPI print quality on a paper that barely does a 300/360 PPI equivalent in print quality. No magical anti-aliasing happens then with inkjet dots, the cell size still has to represent color and detail at pixel level. Correct, it will be worse with black only halftone screening but it is not dissolved by high resolution droplet distributions, stochastic screening, diluted inks or droplet variation if the paper quality can not support that printer quality setting. Again correct that good downsampling became more an issue with higher MP images, that trend of more MP is still going on. A simple rule to extrapolate everything upwards to the next native printer resolution may not be correct for all image content. It will be correct when the image resolution at print size is worse than the print resolution, it may prove wrong with the opposite.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
Yes, good point. 2880 mode is fixed, but all other resolutions are variable, and the choice of which variable sizes to use varies depending on other settings (media, microweave and so on). It's a very complex matrix.
Set against the Canon iPF wide format models that squirt 4 picoliter droplets in all resolution settings, they do not have a variable droplet size but their droplet size is very close to the Epson minimum droplet size of 3.5 picoliter. It is interesting that Epson does not reveal what the maximum droplet size is but a conservative estimation of 3 droplet sizes and a maximum droplet of 11 picoliter should not be far off. Of the dry minilab models (Epson head technology - Fuji, Noritsu Brand) the smallest droplet is not mentioned but should be 2 picoliter or below for 4x6 prints and there are 5 droplet sizes in total.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I am not completely happy with the part above. The point is that if aliasing can happen with the original image resolution at print size versus a lower optical resolution of the printing process then anti-aliasing should be done. By upsampling no extra aliasing conditions are created but it disguises the fact that aliasing still can happen as the 600 or 720 PPI does not represent the optical resolution of the printing process, the last will be lower. The same low optical print resolution may already be available with 300 or 360 PPI input.
Hoi Ernst,
We agree that downsampling is aliasing artifact prone, unless proper anti-aliasing precautions are taken. So, e.g. letting a sub-optimal resampling routine downsample to 360 PPI is more likely to wreak havoc than upsampling to 720 (substitute 300 and 600 PPI for Canon/HP). The same potential risk is there if the image data exceeds 720 (600) PPI for the intended output size and thus downsampling is a given, so one had better use a very good downsampling algorithm to begin with (almost certainly not the printer driver, at the current state of affairs).
If I understand you well, what remains to be established is whether (upsampled) 720/600 PPI image data still has the potential of aliasing, given the fact that the dithering/droplet placement accuracy is higher. Is my assumption correct that that is what needs to be established, or is it the concern that the (paper) medium itself has a lower physical resolution?
If that is a correct assessment, then I'd say that aliasing primarily manifests itself when sampled in an ordered grid. It's the image detail being slightly out of sync that potentially results in massively lower resolution aliases. Both the (random and at a higher frequency) dithering should suppress aliasing, and the fibres and ink diffusion in the print medium should also help to break the regularity of the pixel grid. Adding to that, with the selection of non-glossy media, most printer drivers switch to a lower maximum PPI (something that e.g. Qimage automatically picks up and deals with).
To avoid opinions, and to get some practical results/facts, it could help to do a simple (but merciless) test with different PPI input file settings on different media. For matters involving aliasing, I find a 'zoneplate' target very helpful. I have one available for download (http://bvdwolf.home.xs4all.nl/main/foto/down_sample/down_sample_files/Rings1.gif) on one of my downsampling example (http://bvdwolf.home.xs4all.nl/main/foto/down_sample/down_sample.htm) webpages. As detailed on that webpage, one should first convert the image mode from indexed to RGB in Photoshop, and then the target can be tagged (resized but not resampled) at different PPIs and printed as one is accustomed, and also printed after resizing to a different PPI with resampling. Converting to RGB mode is necessary to avoid shortcomings in Photoshop, and possibly other photoeditors. Keeping the image in 8 bit/channel RGB mode should be sufficient for most workflows, although one can also try if switching to 16-b/ch before resampling and/or converting to the output colorspace, helps to avoid unforeseen artifacts.
It is possible that not only downsampling but also upsampling will still produce aliasing artifacts, despite the loss of per pixel resolution, but that's due to sub-optimal resampling. A good example would be the 259/360/361 PPI test on an Epson printer. GIGO (garbage in garbage out) still rules.
Cheers,
Bart
-
I've never heard an assumption like that before, but there's a first time for everything. We both don't have a solid proof (like an official statement by one of the major printer manufacturers) to support either position, but I think there is more circumstantial evidence to support simple interpolation. Not only is interpolation a logical simplification to speed up the dithering process, there are also moiré effects (http://www.luminous-landscape.com/forum/index.php?topic=54211.msg444078#msg444078) created that are completely consistent with (poor quality) print driver interpolation/resampling.
Actually, I didn't make this up, I'm just repeating an explanation I got from an Epson engineer who was trying to explain how the Epson dither works...moiré or edge aliasing in high contrast diagonals and circles are a result of the interference between the dither and the image edges or other high frequency textural data. Upsampling to 600/720 tends to eliminate them because at the highest image resolutions, there is far less risk of encountering moiré when the dither occurs and this follows along with the Nyquist theorem of oversampling reducing errors.
BTW, Eric Chan's 3800 page predates this latest round of testing we did regarding changing the Lightroom 3 max output resolution to 720PPI. I worked with Eric and Kevin Tieskoetter of the LR engineering team to get the resolution changed and worked with Eric (who has refined the output sharpening in LR) to test the output sharpening at those higher resolutions...
-
Upsampling to 600/720 tends to eliminate them because at the highest image resolutions, there is far less risk of encountering moiré when the dither occurs and this follows along with the Nyquist theorem of oversampling reducing errors.
While true, I assume there is also an even simpler underlying reason. When the PPI results in a value above 300/360 for the output size chosen, at the default settings the printer driver will resample (yes, I do believe resampling takes place) down to 300/360 PPI. Sub-optimal interpolation/resampling is one thing, but improper (if any) AA-filtering before downsampling the pixel data can look terrible. I'm not suggesting that the Epson person was lying to you, but maybe he also didn't tell the whole story ...
Frankly, there is only so much one can expect from a printer driver. Image processing is not one of those things (although it would be nice), it's perhaps better left for dedicated output oriented software/RIPs/etc. that delivers the color-managed output data optimally resized to the printer driver. It's the kind of control spoken about earlier that one needs if components in the chain drop the ball.
Cheers,
Bart
-
so nobody really knows for certain if an Epson printer driver resamples/interpolates or whatever?
-
I think Jeff answered that question. But you know, once we run our tests to see what approach to resolution works best for each of us and we get our answers so we know what to do, it doesn't really matter, does it? I won't find a print looking better if I knew the data was resampled in Lightroom versus the Epson driver, quite frankly. Sure, it's educational to understand more about how these processes work, but there's so much beyond this one item we'll never know anyhow...........huh?
-
so nobody really knows for certain if an Epson printer driver resamples/interpolates or whatever?
To get that knowledge use hard edge B&W resolution targets with a 1440-720-360 LPI pattern (lines preferably) and print that with different printer quality settings that ask for 720-360-180 PPI input. It is easy to tell whether the driver downsamples first and then dithers or does the dithering right away. In the last case you can expect a more or less averaging to grey in total, in the first case it will stay B&W but lines disappear and/or complete black blocks appear, kind of moiré. I have done that some years back with 1200-600 LPI line targets to the HP Z3100 driver asking for 600 or 300 PPI input, to check whether the driver at least does anti-aliasing in its downsampling. It did not. That is also the flaw in this test, if the driver applies an anti-aliasing routine in downsampling then the print may not be that different from a right away dithered print, both more or less grey. For the Epson 9900 I expect that you will see B&W moiré but I do not have one here to test it.
If it is verified for downsampling to the native resolutions then I expect it to do the same in upsampling to the native resolutions. But I have no test for that.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
While true, I assume there is also an even simpler underlying reason. When the PPI results in a value above 300/360 for the output size chosen, at the default settings the printer driver will resample (yes, I do believe resampling takes place) down to 300/360 PPI. Sub-optimal interpolation/resampling is one thing, but improper (if any) AA-filtering before downsampling the pixel data can look terrible. I'm not suggesting that the Epson person was lying to you, but maybe he also didn't tell the whole story ...
Anyone here who knows more than is already in the public domain, can't say anything more than they've already said. It's a big leap to suggest that Jeff was mislead or not given the whole story in order to support your assumption. He's giving you information received directly from the source. Do with it as you like :-)
-
Do with it as you like :-)
I fully agree, and I do. However, when confronted with the following example, actual print enlargements from original Sigma SD14 1:1 crops, both with identical printer driver settings (first crop is from LR3, second crop is from Qimage):
(http://www.ddisoftware.com/qimage/lr-f.jpg) (http://www.ddisoftware.com/qimage/q-f.jpg)
I can only wonder why people question the benefit of better quality resampling versus letting the printer driver do it's thing.
Cheers,
Bart
P.S. The images were directly linked from a message from Mike Chaney. I hesitate to link to it directly because there seems to exist a somewhat allergic reaction with some forum dwellers when Qimage is mentioned, where I only want to underline the importance of better quality resampling, regardless of the method used.
-
I don't think anyone's suggesting that if you resample, you shouldn't do the best that you can. Without knowing the exact details of each workflow it's hard to compare or to say whether the best of each was being used. Furthermore, it doesn't show one way or the other whether a particular driver does or doesn't resample.
-
Furthermore, it doesn't show one way or the other whether a particular driver does or doesn't resample.
No, but I gave a route to test whether resampling in the driver is done or not when there is too much input resolution.
The resampling Bart shows looks nice to me, is it the new fusion algorithm or the older hybrid version of Qimage?
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
No, but I gave a route to test whether resampling in the driver is done or not when there is too much input resolution.
The resampling Bart shows looks nice to me, is it the new fusion algorithm or the older hybrid version of Qimage?
I'm not sure, Mike didn't mention, but it was an addition to his March 2010: Smart Photo Printing article on his Tech Support forum. I believe the new Fusion algorithm is of more recent date, so I'll have to give that a try some day as well. Hybrid SE is pretty good, because it produces no halo and thus is a good basis for either Smart sharpening or e.g. Deconvolution sharpening on a print file (if the size is still manageable). For large output QI seems to handle the memory for large files better than an average printdriver.
Having said that, I try to let this thread not turn into an application X versus Y pissing contest. Just some considerations about what is needed to get the best output, and what to avoid (or do) with printer driver settings, with examples to illustrate.
Cheers,
Bart
-
Having said that, I try to let this thread not turn into an application X versus Y pissing contest. Just some considerations about what is needed to get the best output, and what to avoid (or do) with printer driver settings, with examples to illustrate.
Cheers,
Bart
That is not my intention either but I am more familiar with Qimage and it would surprise me if a printer driver adapts itself to the application that calls it. So for samples., explanations and tests on what does the resampling I will refer to that application. Of course I could be wrong and what Jeff writes only applies to LR running on a Mac to drive an Epson.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I'm not sure, Mike didn't mention, but it was an addition to his March 2010: Smart Photo Printing article on his Tech Support forum.
Well, if you don't know exactly how each of the examples were made, the samples' usefulness drops considerably...so, was the LR 3 sample upsampled to 720 in LR3 and output sharpened? Without basic information, it really doesn't bring much to the discussion.
-
To get that knowledge use hard edge B&W resolution targets with a 1440-720-360 LPI pattern (lines preferably) and print that with different printer quality settings that ask for 720-360-180 PPI input. It is easy to tell whether the driver downsamples first and then dithers or does the dithering right away. In the last case you can expect a more or less averaging to grey in total, in the first case it will stay B&W but lines disappear and/or complete black blocks appear, kind of moiré. I have done that some years back with 1200-600 LPI line targets to the HP Z3100 driver asking for 600 or 300 PPI input, to check whether the driver at least does anti-aliasing in its downsampling. It did not. That is also the flaw in this test, if the driver applies an anti-aliasing routine in downsampling then the print may not be that different from a right away dithered print, both more or less grey. For the Epson 9900 I expect that you will see B&W moiré but I do not have one here to test it.
If it is verified for downsampling to the native resolutions then I expect it to do the same in upsampling to the native resolutions. But I have no test for that.
To make it easier, here is the Epson target to use so you can deliver evidence for Jeff's statement:
http://www.ddisoftware.com/qimage/quality/restest-epson.zip
more info on how to use them for normal testing
http://www.ddisoftware.com/qimage/quality/
In this case you should not just use the highest printer quality setting but also the lower quality settings as described above. For 1440 PPI input resolution, print the target image at 50% of its size.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
All of which is interesting, Ernst, but doesn't address the issues raised by Jeff, myself and others - without precise knowledge of the two workflows it is not useful to make a comparison.
Following your suggestion of testing is a much better idea than relying on the samples provided here, imho, because it allows a user to know what they are comparing and to use the optimal workflow in each case.
-
My experience shows that sending my Epson 3800 anything less than 360ppi risks degraded output. In contrast, the figure is 240ppi with my HP Z3100.
HTH.
-
All of which is interesting, Ernst, but doesn't address the issues raised by Jeff, myself and others - without precise knowledge of the two workflows it is not useful to make a comparison.
Following your suggestion of testing is a much better idea than relying on the samples provided here, imho, because it allows a user to know what they are comparing and to use the optimal workflow in each case.
The test with line targets with a resolution exactly twice the printer input resolution (as set by printer driver quality choices) reveals whether the driver does downsampling to that input resolution or starts dithering right away. To me that aims at the essence of this debate. It is a simple task for the owner of a recent model Epson. Use Photoshop or another application that will not interfere with resampling by itself.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
The test with line targets with a resolution exactly twice the printer input resolution (as set by printer driver quality choices) reveals whether the driver does downsampling to that input resolution or starts dithering right away. To me that aims at the essence of this debate. It is a simple task for the owner of a recent model Epson. Use Photoshop or another application that will not interfere with resampling by itself.
Indeed. If printing that file at any PPI other than the native printer resolution produces aliasing , then resampling is what caused it. Downsamping by the printer driver (native resolution set too low) will be most likely be worse than upsampling. Only the remote posibility of some strange software bug could keep the conclusion from being 100% certain. Stochastic dithering doesn't cause aliasing, and even simple dithering won't (it might cause a visble dither pattern, but no aliasing).
In fact, in some cases the resampling by the printer driver is so mediocre, that any resulting PPI other than the native one (for the driver settings) will cause artifacts on the critical test targets mentioned in this thread (like in the 359/360/361 test case). The artifacts do not necessarily show as prominent on normal images, but that doesn't mean there is no resampling.
Cheers,
Bart
-
My experience shows that sending my Epson 3800 anything less than 360ppi risks degraded output. In contrast, the figure is 240ppi with my HP Z3100.
HTH.
I do not think that the discussion is about which printer delivers best output etc but nevertheless:
With a variety of 4 and 6 picoliter droplets the Z3100 will not make an optical resolution as high as the Epson 3800 which can rely on 3.5 picoliter droplets for best quality. There is a rough rule to compare: 3.5 : 5 picoliter = 0.7 0.7 x 360 = 252 close enough to your 240. The HP will not ask for 360 PPI input but for 300 PPI which makes it all quite logical. Given good + enough data and maximum paper coating quality the higher input qualities of 600 or 720 PPI are available for resp the HP and Epson so the HP print quality is not limited to the 240 PPI you quote for 300 PPI input. A similar approach for a Canon iPF model, 4 picoliter throughout (in all dpi resolutions), this time not for input resolution but for for dot resolution: 3.5 : 4 = 0.875 0.875 x 1440 dpi = 1260, close enough to the 1200 dpi of the Canon. 0.875 x 2880 dpi = 2520, close enough to the 2400 dpi of the Canon. Whether the droplet size and/or dot placement are defined exactly by the numbers published is another question. A rough rule it is, the HP B9180 has to work with the same 4-6 droplet heads the Z3100 has but its dithering/weaving algorithms are better (and will take more processing time per square foot), so lays down a better print quality despite the same droplet sizes. With the exception of the B9180 all manufacturers use smaller droplet sizes, 2- 1 picoliter, for photo quality A3 and A4 printers as the expected viewing distance for print sizes like that will be less.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I do not think that the discussion is about which printer delivers best output etc but nevertheless:
With a variety of 4 and 6 picoliter droplets the Z3100 will not make an optical resolution as high as the Epson 3800 which can rely on 3.5 picoliter droplets for best quality. There is a rough rule to compare: 3.5 : 5 picoliter = 0.7 0.7 x 360 = 252 close enough to your 240. The HP will not ask for 360 PPI input but for 300 PPI which makes it all quite logical. Given good + enough data and maximum paper coating quality the higher input qualities of 600 or 720 PPI are available for resp the HP and Epson so the HP print quality is not limited to the 240 PPI you quote for 300 PPI input. A similar approach for a Canon iPF model, 4 picoliter throughout (in all dpi resolutions), this time not for input resolution but for for dot resolution: 3.5 : 4 = 0.875 0.875 x 1440 dpi = 1260, close enough to the 1200 dpi of the Canon. 0.875 x 2880 dpi = 2520, close enough to the 2400 dpi of the Canon. Whether the droplet size and/or dot placement are defined exactly by the numbers published is another question. A rough rule it is, the HP B9180 has to work with the same 4-6 droplet heads the Z3100 has but its dithering/weaving algorithms are better (and will take more processing time per square foot), so lays down a better print quality despite the same droplet sizes. With the exception of the B9180 all manufacturers use smaller droplet sizes, 2- 1 picoliter, for photo quality A3 and A4 printers as the expected viewing distance for print sizes like that will be less.
Hmmm... thanks Ernst for the information, you have given me something to think about (I've read it through three times and almost understood it).
I did not intend to comment on which printer produces best output, however I can see why you took my comment that way.
My intention was to comment on the marked difference in print quality produced when using the same ppi value on different printers. E.g. from the same source file and using my personal standard of print quality, I can produce a larger print using the HP, than I can with the Epson. I.e. when I am printing a BW image with fine diagonal lines, I have to up the source size and send it to the Epson at 360dpi to avoid jaggies which are not present in the HP's output at 240ppi. I am not commenting on the highest print quality that can actually be achieved by either printer.
Up until now I had thought that the HP (along with the driver etc.) merely had a better interpolation algorithm than the Epson.
Am I reading you right in saying that this is actually a result of the Epson having an overall higher resolution?
Thanks,
Richard.
-
Up until now I had thought that the HP (along with the driver etc.) merely had a better interpolation algorithm than the Epson.
Am I reading you right in saying that this is actually a result of the Epson having an overall higher resolution?
Thanks,
Richard.
Richard,
Both are possible. In theory the Epson should be able to achieve better print resolution given the same extrapolation algorithm before printing is used. But from which application did you print?
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Both are possible. In theory the Epson should be able to achieve better print resolution given the same extrapolation algorithm before printing is used. But from which application did you print?
I print from both QImage and Photoshop, under Windows Vista 64bit. I never let QImage resize the images. Both applications produce the same results.
Thanks,
Richard.
-
I've removed this post because I realised that the question I was asking was actually a non-question. Further thinking has caused light to be shed on the purpose of Ernst's test image, so I will have to go back to the beginning and run my trials again. A new (and hopefully more coherent) post will follow.
John
-
Hi Jeff & others,
I use the latest version of Photokit Sharpener.
I've followed this long thread and frankly am in an overload state. I'm now looking for some advice framed by this question: When using Photokit Sharpener, should I be upressing my file to 300 or 600 according to the Schewe Rule before I invoke PK Sharpener, or should I just let PK Sharpener work on the file at the "native" (ie, no up or down ressing) resolution that it has?
I print to only to an HP B9180, using only from Photoshop, and using a variety of papers with which I use the appropriate ICC profile and softproof.
TIA for any and all advice.
Ciao,
Doug
-
I'm now looking for some advice framed by this question: When using Photokit Sharpener, should I be upressing my file to 300 or 600 according to the Schewe Rule before I invoke PK Sharpener, or should I just let PK Sharpener work on the file at the "native" (ie, no up or down ressing) resolution that it has?
You should upsample (if you are going to) BEFORE you run the final output sharpening...PKS 2 can actually do both at once in the Output Sharpener for Inkjet.
-
Thanks for the reply, Jeff.
I think I'll leave PK Sharpener to do its thing--dare I say this, it is probably smarter than I am on the issue of sharpening.
Ciao,
Doug
-
I've removed this post because I realised that the question I was asking was actually a non-question. Further thinking has caused light to be shed on the purpose of Ernst's test image, so I will have to go back to the beginning and run my trials again. A new (and hopefully more coherent) post will follow.
John
Well I had my answer on your message ready and it had elements where I questioned the use of LR and the methods used. Some of what I wrote may have been visibIe in your results but other methods as described before by me should deliver more significant results. I would use Photoshop or the free Adobe Color Printer Utility or Qimage with all resampling/sharpening switched off. I simply do not know whether LR can pass image data without interfering, in whatever setting.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Well I had my answer on your message ready and it had elements where I questioned the use of LR and the methods used. Some of what I wrote may have been visibIe in your results but other methods as described before by me should deliver more significant results. I would use Photoshop or the free Adobe Color Printer Utility or Qimage with all resampling/sharpening switched off. I simply do not know whether LR can pass image data without interfering, in whatever setting.
Ernst, I am really sorry to have wasted your message. I used LR because that is my software which I use for everything, so of course I wanted to know what that does in combination with the printer I have. Where I was being clueless was in not fully understanding the purpose of your test. I think I now realise, after studying my results, that the test file is designed to show up any resampling by the software or printer, and will show banding artefacts in columns A and D unless absolutely no resampling has taken place. Is that correct?
John
-
Ernst, I am really sorry to have wasted your message. I used LR because that is my software which I use for everything, so of course I wanted to know what that does in combination with the printer I have. Where I was being clueless was in not fully understanding the purpose of your test. I think I now realise, after studying my results, that the test file is designed to show up any resampling by the software or printer, and will show banding artefacts in columns A and D unless absolutely no resampling has taken place. Is that correct?
John
John,
Normally this test is used to see the actual maximum optical print resolution possible with a paper/ink/printer/quality setting. That information is a good base for making print quality selections in your driver, say you should not select a 720 PPI input quality for an uncoated paper if it delivers the same quality with 360 PPI input. It uses those resolution numbers to avoid resampling done by the printer driver or application you print from. So it is based on the assumption that the driver likes to have those numbers for input. If that theory is not correct then the target still can be used to proof that that theory is not correct.
In this case you only want to know whether there are certain input resolutions the driver relies on before any computations on the data are done by it. If it does not need certain input resolutions but dithers on any input feed right away (Jeff's opinion) then aliasing, moiré, should not (or hardly) be visible in any case. But instead if it starts computing by resampling odd resolutions first to the resolutions it likes to start with then it has to resample one way or another. The quality of that resampling routine could reveal whether it uses certain input resolutions, most likely the downsampling routine is worse (does not have ant-aliasing) than the upsampling routine. If both are good then it becomes more difficult to see whether the driver relies on certain input resolutions. I do not expect that for driver downsampling 2:1 size, while your LR 2:1 size reduction did that excellent (must have anti-aliasing). So any interference of LR on the downsampling should be taken out otherwise we are not measuring what should be measured.
If you print a half size target 1440 PPI to a print quality setting that asks for 720 PPI do you see any kind of taxi block strips or worse emerge from the line patterns?
The same for a normal size target 720 PPI to a print quality setting that asks for 360 PPI ?
Any resampling done under the skin by the application = LR (or what I proposed) really taken out?
Blocking up that way should happen with an Epson 2400 in my expectation. It is not the latest model so we still will have that question for more recent models.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Thanks, Ernst.
All understood (I think). I will run further tests this evening. So far I can say with certainty -
* The R2400 does make use of 720ppi files when printing at its best quality settings. The difference in line thickness in column D is apparent to the naked eye as well as under a loupe. I already knew this, by comparing 10x8 prints made at 360 and 720ppi.
* When output at exactly 720ppi from LR no artefacts are visible in the print, so I assume LR does not do anything to the file if resampling and output sharpening are switched off - and neither does the printer, with a 720ppi input.
As you say, the R2400 is a six-year old design, so any further results will not tell us much about the latest x880 and R3000 printers. But surely everybody else on the LL Forum is at this moment running their own tests? ;)
John
-
John,
Normally this test is used to see the actual maximum optical print resolution possible with a paper/ink/printer/quality setting.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
Ernst,
When time permits I may try this test on my Epson 4900. It will be some weeks from now I'm afraid, because I'm swamped, but when I do it I'd be pleased to report back.
Now, one curiosity - several times you have mentioned the word "optical resolution" in the context of printers. Does this make sense? For scanners, yes, because there are optics involved, but printers? What do you mean by this term? Isit what others call "native" resolution?
-
The print resolution as seen by the naked eye of a myopic 20 year old or if it has to be scientific, measured with something like the best optical microscope around + Imatest and an appropriate target to get MTF results. Nobody does that yet for testing to publish results but it would be a good thing if it happened. I use that term to make the distinction between actual print resolution and the printer's droplet addressing in DPI or input resolution in PPI. The same distinction that has to be made when scanner manufacturers quote sampling resolution numbers with DPI added where it should be SPI and suggest they are true PPI numbers where Imatest will show the actual resolution is way less than that (Epson) or a 10% lower (Nikon). But in MTFs numbers. ColorFoto published scanner resolution numbers in PPI measured by Image Engineering in Germany. A company that will be qualified to do MTF measurements on inkjet papers if they do not already. It is done on halftone offset printing etc so could be done on inkjet printing.
I have seen "native resolution" used in many ways, for Apple/Adobe it seems to be the resolution of old Mac screens 72 PPI (at least I see that in some documents), others refer to it as the resolution of the camera sensor and the translation to different sizes without resampling, I thought for a long time that it meant the input resolution(s) of a printer discussed here so for Epson 360-720 PPI (add 180 and 1440 PPI for other printers/settings). I try to avoid the term now as it did become confusing and is to much device specific. So no, it is not what I see as optical print resolution. But the next guy may add a native resolution number to an inkjet paper :-)
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Ernst and everyone
Testing has proceeded this evening, and I do have some interesting results. My output was from Lightroom 3.4.1, running on Win 7 64-bit, and printing to my Epson R2400. The 2400 was set to its best output resolution, Photo RPM (5760 x 1440), High Speed off, and the paper was Epson Premium Glossy. The test file was sized in the print module to produce output at exactly 720 ppi initially, and then resized to produce other output resolutions for comparison. LR output sharpening was turned off.
I do understand that the original purpose of the restest file was establish what ppi and printer settings were appropriate for a given paper and print size combination. What we are trying to establish here (does the printer resample input data or not) is a bit different and not easy to pin down.
* If the test image is sized to exactly 720 ppi and sent to the printer direct so that there is no resampling whatsoever taking place, either in LR or at the printer (we assume), then we get a clean printed output with no artefacts, as you would expect. The difference between the 720 and 360 line widths in column ‘D’ is clearly apparent.
* I could not try Ernst’s 1440 ppi test, because LR will not report resolution settings above 720 – it just says 720+. So I had no way of accurately resizing the image to 1440.
* I could not try the 720 to 360 test, either, for reasons which will become clear.
* If the file is resized to any other value than 720 or 360 ppi, it results in strong vertical banding in columns A and D. The “wavelength” of the banding seems to vary according to the mathematical relationship between the resolution of the file and the output resolution at 720. The banding happens whether the file is downsampled to 720 or upsampled from a lower value. So any resampling results in artefacts, whether the resampling is done in LR (598 to 720, say) or we just send the file straight to the printer and let the Epson sort it out.
* If we take the image sized to say 686ppi and resample to 720ppi in LR, then send it to the 2400, or alternatively send the file to the printer at 686ppi with no processing in LR we get exactly the same banding, except that we can see even with the naked eye that the artefacts are more tightly defined and crisper in the LR processed version and the overall print quality is better. From this I infer that the printer does indeed resample a non-720 input, but that the LR sampling algorithms are superior.
* I also resampled the file in LR to 360 ppi and there were no artefacts, but the printed result was clearly inferior. The 720 and 360 lines in column ‘D’ become the same width, and the text becomes very smudgy.
* By now I had got rather curious, and tested the file with all the various quality settings on the 2400 for photo output on glossy paper. From the best downwards, they are called – Photo RPM, Best Photo, Photo, and Fine. Your Epson may have different names, but they probably do the same thing. I expected the output to shift from 720 to 360 somewhere down this list, but no – all of these are 720 ppi output. The quality does reduce a little, especially in the text, but the difference between the line thicknesses in column ‘D’ is still clearly apparent.
* So it seems that the Epson R2400, at any rate, when printing photos on glossy paper is always a 720 ppi device. Any other resolution you send to it gets resampled, and some quality is lost.
These tests back up my own observations when printing real photographs – a subject which we can expand on a bit tomorrow, perhaps. By the by, my “naked eye” may well be a bit different than yours, because I have been short-sighted since the age of 5 and I focus without spectacles at 6 inches. The results were also checked with an 8x loupe.
John
-
* If the file is resized to any other value than 720 or 360 ppi, it results in strong vertical banding in columns A and D. The “wavelength” of the banding seems to vary according to the mathematical relationship between the resolution of the file and the output resolution at 720. The banding happens whether the file is downsampled to 720 or upsampled from a lower value. So any resampling results in artefacts, whether the resampling is done in LR (598 to 720, say) or we just send the file straight to the printer and let the Epson sort it out.
Hi John,
Thanks for participating. What you have described here is completely consistent with sampling theory (Shannon/Nyquist theorem). Nyquist stated that, (in my own words) in order to unambiguously resolve a (spatial) frequency, it needs to be sampled at more than 2x that frequency. IOW, to resolve 360 PPI (or lower spatial frequency) image detail, it must be sampled at 720 or higher frequencies. Between 360 and 720 PPI there is still ambiguity (aliasing), and as always the aliases are larger than the sampling frequency. Ipso Facto, resampling takes place as witnessed by the resulting aliases.
* If we take the image sized to say 686ppi and resample to 720ppi in LR, then send it to the 2400, or alternatively send the file to the printer at 686ppi with no processing in LR we get exactly the same banding, except that we can see even with the naked eye that the artefacts are more tightly defined and crisper in the LR processed version and the overall print quality is better. From this I infer that the printer does indeed resample a non-720 input, but that the LR sampling algorithms are superior.
Indeed, as we can expect LR also uses some sharpening, but it cannot prevent (except for pre-blurring) the aliasing due to undersampling. The fact that you get exactly the same aliasing proves that the resampling takes place in both cases.
The final result on image quality is that when there is 360-720 PPI image detail in the available ink colors, it can be resolved as long as it is not a regular pattern. A regular pattern can aliase, other detail will be dithered to form intermediate colors. In addition, 720 PPI allows to sharpen at that output level, but it will also boost lower spatial frequencies.
Cheers,
Bart
-
Ernst,
When time permits I may try this test on my Epson 4900. It will be some weeks from now I'm afraid, because I'm swamped, but when I do it I'd be pleased to report back.
It would be nice if you do that with another program than Lightroom. I see that John's testing is hitting on limitations in that program, limitations that might be excellent for normal printing but obscure what we want to measure.
I think there is a chance to use a 3800 or 3880 of a friend next week so I can do a test too on an Epson.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
* I could not try Ernst’s 1440 ppi test, because LR will not report resolution settings above 720 – it just says 720+. So I had no way of accurately resizing the image to 1440.
* I could not try the 720 to 360 test, either, for reasons which will become clear.
* If the file is resized to any other value than 720 or 360 ppi, it results in strong vertical banding in columns A and D. The “wavelength” of the banding seems to vary according to the mathematical relationship between the resolution of the file and the output resolution at 720. The banding happens whether the file is downsampled to 720 or upsampled from a lower value. So any resampling results in artefacts, whether the resampling is done in LR (598 to 720, say) or we just send the file straight to the printer and let the Epson sort it out.
* If we take the image sized to say 686ppi and resample to 720ppi in LR, then send it to the 2400, or alternatively send the file to the printer at 686ppi with no processing in LR we get exactly the same banding, except that we can see even with the naked eye that the artefacts are more tightly defined and crisper in the LR processed version and the overall print quality is better. From this I infer that the printer does indeed resample a non-720 input, but that the LR sampling algorithms are superior.
* I also resampled the file in LR to 360 ppi and there were no artefacts, but the printed result was clearly inferior. The 720 and 360 lines in column ‘D’ become the same width, and the text becomes very smudgy.
* By now I had got rather curious, and tested the file with all the various quality settings on the 2400 for photo output on glossy paper. From the best downwards, they are called – Photo RPM, Best Photo, Photo, and Fine. Your Epson may have different names, but they probably do the same thing. I expected the output to shift from 720 to 360 somewhere down this list, but no – all of these are 720 ppi output. The quality does reduce a little, especially in the text, but the difference between the line thicknesses in column ‘D’ is still clearly apparent.
* So it seems that the Epson R2400, at any rate, when printing photos on glossy paper is always a 720 ppi device. Any other resolution you send to it gets resampled, and some quality is lost.
These tests back up my own observations when printing real photographs – a subject which we can expand on a bit tomorrow, perhaps. By the by, my “naked eye” may well be a bit different than yours, because I have been short-sighted since the age of 5 and I focus without spectacles at 6 inches. The results were also checked with an 8x loupe.
John
John,
Thank you for the trials. I have some questions. You hit on a limitation in LR for the higher than 720 PPI resolution, you can not half the size of the image based on your 720 PPI test an by that get 1440 PPI without bothering whether LR says what resolution it has? The other one, you hit on a driver limitation for the glossy paper media preset to get 360 PPI input resolution, there are no matte paper choices or even uncoated paper choices that brings it down to 360 PPI? I am still not happy with LR for tests like this. It could even be that if the driver asks for 360 PPI Lightroom still sends 720 PPI, even if you resampled the file to 360 PPi in Lightroom.
So far your testing says to a degree that the 2400 driver has one fixed input resolution at least: 720 PPI.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
Thank you for the trials. I have some questions. You hit on a limitation in LR for the higher than 720 PPI resolution, you can not half the size of the image based on your 720 PPI test an by that get 1440 PPI without bothering whether LR says what resolution it has? The other one, you hit on a driver limitation for the glossy paper media preset to get 360 PPI input resolution, there are no matte paper choices or even uncoated paper choices that brings it down to 360 PPI? I am still not happy with LR for tests like this. It could even be that if the driver asks for 360 PPI Lightroom still sends 720 PPI, even if you resampled the file to 360 PPi in Lightroom.
So far your testing says to a degree that the 2400 driver has one fixed input resolution at least: 720 PPI.
Ernst
As I said before, I use LR and I am interested in what LR does rather than running a printer test lab ;) Re-sizing the file to half size is not an option because the units of measurement available (.01 cm) are simply not accurate enough to get exactly 1440 ppi without any confirmation from LR itself. I had a hell of a job getting it to exactly 720 ppi as it is - this is one of the issues with LR, of course. You are correct, probably there are 360 ppi settings somewhere in the matt paper choices but I never print on matt so this is of no interest to me. And after a certain point I was losing the will to live . . .
John
-
The thing which annoys me about all of this is that we are having to establish, through tedious testing and experiment, the nuts and bolts of technicalities which should be freely available from the printer manufacturer. But still, another question -
How does the printer know what resolution the incoming print spool file actually is? Is this information in a file header somewhere, or does the printer analyse the data in some way? The reason I ask is because I think I can see a difference in output between -
* The file sized to 720 and sent direct to the 2400 and
* The file sized at 720 but with the resolution flag in LR also set to 720.
When there should be no difference at all. But if LR sets a flag in the file when it resamples, but not otherwise, this could account for it.
John
-
The thing which annoys me about all of this is that we are having to establish, through tedious testing and experiment, the nuts and bolts of technicalities which should be freely available from the printer manufacturer.
If they knew what to tell you, they would...fact is, they're not really sure themselves.
The bottom line is that through testing, I'm pretty sure that the suggestion to upsample in LR 3 to either 360/300 or 730/600 PPI (depending on your printer with the 360/720 for Epson and the 300/600 for HP & Canon) is legit and will result in optimal results...
A lot depends on the original image quality. Ya really can't make a silk purse out of a sow's ear. The better the original image, the better the final output.
-
If they knew what to tell you, they would...fact is, they're not really sure themselves.
Jeff, printer manufacturers always specify the dithered output - but never tell us what the optimal input should be. When it would be extremely useful to know that, for the range of paper types and output resolutions which are supported by the hardware. It's a bit like a film manufacturer giving you all the info on how to dev a film, but nothing on how to expose it.
John
-
How does the printer know what resolution the incoming print spool file actually is? Is this information in a file header somewhere, or does the printer analyse the data in some way? The reason I ask is because I think I can see a difference in output between -
* The file sized to 720 and sent direct to the 2400 and
* The file sized at 720 but with the resolution flag in LR also set to 720.
When there should be no difference at all. But if LR sets a flag in the file when it resamples, but not otherwise, this could account for it.
John
John,
As I understand it from discussions with Mike Chaney of Qimage the application may send a Windows image file format to the driver so not a Tiff etc and my best guess is that that happens in both cases without any extra flag. What you probably observe is a sharpening step (or another adaptation step to the media you selected) in LR that is not switched off, one of those black box things that makes this program not ideal to reveal what the driver itself does. It is like testing an Imacon scanner without knowing that even a zero on the sharpening setting still means sharpening is done, you have to set say a -60 or -120 number to get rid of the sharpening in its software.
The other way around, how the application could know what the driver likes to receive on input resolution, is by looking for an API call by the driver in Windows. that is where applications like Qimage and LR collect those numbers and use them for the best adaptations of the images to that request. Qimage does that more transparent and you can switch off all adaptations if needed, like for testing.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
What you probably observe is a sharpening step (or another adaptation step to the media you selected) in LR that is not switched off, one of those black box things that makes this program not ideal to reveal what the driver itself does.
Thanks, Ernst, that does make a lot of sense. Output sharpening in LR was certainly set to off, but as you say, something else may still be happening. The point is, that the output quality is better without the LR resolution flag set - you can see this particularly in the text. So you can see where this is going, can't you?
John
-
John,
So far it is telling me more about LR than any other thing. That text in images decreases in quality with programs aiming at good photo quality is not new to me either. There is a setting in printer drivers for fine detail that may enhance your text and lines even more (and could request 1440 PPI or 1200 input) but does not improve a photo image in most cases. Different horses for different courses.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
So far it is telling me more about LR than any other thing. That text in images decreases in quality with programs aiming at good photo quality is not new to me either. There is a setting in printer drivers for fine detail that may enhance your text and lines even more (and could request 1440 PPI or 1200 input) but does not improve a photo image in most cases. Different horses for different courses.
That's right. And what works well for a bunch of geometric lines is not necessarily best suited to the average photograph. But text - especially the serifed text used in the test file - does have a lot in common with the problems of digital landscape photography - smooth curves and fine lines have to be rendered using (not enough) square pixels. My photographs usually include a great deal of very fine, high-frequency detail - grass, trees against sky, complex vegetation. All in B/W with no colour information to trick the eye and flesh things out a bit. So this whole debate is very important to me.
John
-
It would be nice if you do that with another program than Lightroom. I see that John's testing is hitting on limitations in that program, limitations that might be excellent for normal printing but obscure what we want to measure.
I think there is a chance to use a 3800 or 3880 of a friend next week so I can do a test too on an Epson.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
I would use Photoshop for that.
-
John,
As I understand it from discussions with Mike Chaney of Qimage the application may send a Windows image file format to the driver so not a Tiff etc and my best guess is that that happens in both cases without any extra flag. What you probably observe is a sharpening step (or another adaptation step to the media you selected) in LR that is not switched off, one of those black box things that makes this program not ideal to reveal what the driver itself does. It is like testing an Imacon scanner without knowing that even a zero on the sharpening setting still means sharpening is done, you have to set say a -60 or -120 number to get rid of the sharpening in its software.
The other way around, how the application could know what the driver likes to receive on input resolution, is by looking for an API call by the driver in Windows. that is where applications like Qimage and LR collect those numbers and use them for the best adaptations of the images to that request. Qimage does that more transparent and you can switch off all adaptations if needed, like for testing.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
My experience working with Lightroom is that when you turn sharpening off by setting the Amount to Zero in both the Develop module and the Print module, it really is OFF. There is no sharpening taking place under the hood. But for safety it may still be better for testing purposes to do this in Photoshop where you know that unless you explicitly sharpen, there is none. But this is only an observation based on differences between settings the user can control. Aprt from the programmers, who knows, for any application, what really goes on under the hood before the image hits the display or printer that could be akin to *some* measure of default sharpening just to make the image minimally acceptable.
-
Apart from the programmers, who knows, for any application, what really goes on under the hood before the image hits the display or printer that could be akin to *some* measure of default sharpening just to make the image minimally acceptable.
Indeed. A good example of this was again LR itself, where in LR 2x some NR was applied even if you had all the NR settings at zero. Not that we realised this, of course, until LR 2.7 came out, and suddenly zero really meant zero. Which remains the case in LR 3x with PV 2010, of course. However, you have to work with the photo-editing and printing application of your choice, so the important thing is to optimise the results from that. I personally would prefer to work with a situation where all aspects were user controlled.
John
-
Indeed. A good example of this was again LR itself, where in LR 2x some NR was applied even if you had all the NR settings at zero. Not that we realised this, of course, until LR 2.7 came out, and suddenly zero really meant zero. Which remains the case in LR 3x with PV 2010, of course. However, you have to work with the photo-editing and printing application of your choice, so the important thing is to optimise the results from that. I personally would prefer to work with a situation where all aspects were user controlled.
John
John - beg to differ - I understand the logic of what you are saying - you want test results that show the merits of a workflow with what you are using - that makes sense. But from the point of view of "pure science" you want to use an application that is best suited for the purposes of the test. I think you are coming from the practical perspective of "I want to know how what I use works", whereas Ernst is more interested in the objective issue of what the Epson driver versus the application really does, and what settings are technically optimal. Both perspectives of course are valid but may call forth different approaches.
On the question of being in a situation "where all aspects were user controlled" - be careful what you wish for. I think we are in a situation where we don't even know quite what there is that needs to be controlled under the hood, and we may be overwhelmed if we had to manually set everything the software does to fashion the initial appearance of an image.
-
On the question of being in a situation "where all aspects were user controlled" - be careful what you wish for. I think we are in a situation where we don't even know quite what there is that needs to be controlled under the hood, and we may be overwhelmed if we had to manually set everything the software does to fashion the initial appearance of an image.
Mark, nice one. I was was thinking of that as I wrote it actually - "be careful what you wish for, you might just get it" ;)
I was rather hoping that by now a zillion other LL Forumites would have weighed in with their own tests using Ernst's file on their own, more recent Epson printers. Lazy lot, aren't they? I will probably be getting an R3000 later in the year, so I shall run the same tests again on that. Personally, contrary to other possibly more expert opinion, I anticipate the results to be the same - the printer expects 720 ppi input for all photos on gloss media, and anything else will be resampled. I know that there have been improvements on the new printers to the inksets and the dithering algorithms, but my guess is that all of this applies to the output side alone, not the input.
PS By-the-by, we still have not really had an answer to this question, I think - "how does the printer know what resolution the incoming print spool file actually is? Is this information in a file header somewhere, or does the printer analyse the data in some way?"
John
-
You should upsample (if you are going to) BEFORE you run the final output sharpening...PKS 2 can actually do both at once in the Output Sharpener for Inkjet.
Hi Jeff,
I'm somewhat confused by your response to the upsampling question. You mention that PKS 2 can do both at once in the Output Sharpener for Inkjet. I read that to mean that PKS 2 can both sharpen and upsample simultaneously. I've just checked in again to see if there was something I had missed, but can't seem to find an area within the PKS 2 dialog that will allow me to change the resolution while retaining the image output size. It's probably right there staring me in the face, but so far I haven't located that particular capability, nor is it mentioned in the manual. Perhaps I am misreading your response. If so, please advise.
Thank you,
Gary
-
Thank you for the trials. I have some questions. You hit on a limitation in LR for the higher than 720 PPI resolution, you can not half the size of the image based on your 720 PPI test an by that get 1440 PPI without bothering whether LR says what resolution it has? The other one, you hit on a driver limitation for the glossy paper media preset to get 360 PPI input resolution, there are no matte paper choices or even uncoated paper choices that brings it down to 360 PPI? I am still not happy with LR for tests like this. It could even be that if the driver asks for 360 PPI Lightroom still sends 720 PPI, even if you resampled the file to 360 PPi in Lightroom.
So far your testing says to a degree that the 2400 driver has one fixed input resolution at least: 720 PPI.
Ernst,
I was going to run some of the testing and use Qimage Pro, however, it keeps saying that the max for my Epson 4880 is 360 dpi so I am guessing that is a limitation of the older Qimage Pro, since the Epson driver runs at 720dpi.
Alan
-
You mention that PKS 2 can do both at once in the Output Sharpener for Inkjet. I read that to mean that PKS 2 can both sharpen and upsample simultaneously.
Yep, my bad...I was thinking of the ability to resize without resampling. So if you want to upsample you'll need to do it in Photoshop before the final output sharpening.
-
Yep, my bad...I was thinking of the ability to resize without resampling. So if you want to upsample you'll need to do it in Photoshop before the final output sharpening.
Thanks Jeff, much appreciated!
Gary
-
Ernst,
I was going to run some of the testing and use Qimage Pro, however, it keeps saying that the max for my Epson 4880 is 360 dpi so I am guessing that is a limitation of the older Qimage Pro, since the Epson driver runs at 720dpi.
Alan
Alan,
That is very unlikely. If you set the highest image qualities + 1440 or 2880 dpi etc for a gloss paper in the Epson 4880 driver it should jump to 720 PPI. You ought to see that 720x720 above Qimage's preview. The extrapolation options are below that preview window and you can select the resampling algorithms there + the Max = 720 PPI in this case, a lower setting = 360 PPI (so in this case the driver will do the rest = 2x ) and OFF (the driver does all) + the degree of smart print sharpening, default 5 and a range of 0 to 20. For testing check that it is all OFF or zero and also that no print filter is set and control that the job log shows the target image resolution as you would like to see in relation to the requested printer input resolution. I was almost at a point to install that driver but I rather avoid that, there are already 4 drivers installed and I can not afford a problem on that system right now. There must be 4000-4900 owners with Qimage that can verify what I expect. I recall it was possible more than 5 years ago with an older Epson wide format and an older Qimage.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Ernst,
I was going to run some of the testing and use Qimage Pro, however, it keeps saying that the max for my Epson 4880 is 360 dpi so I am guessing that is a limitation of the older Qimage Pro, since the Epson driver runs at 720dpi.
Hi Alan,
There is no limitation in QI, but you must set the Epson printer driver to "Finest Detail (http://people.csail.mit.edu/ericchan/dp/Epson3800/faq.html#native_res)" (assuming the driver settings are similar for your model) to enable it. QI uses whatever the printer driver reports as being available, when you select max quality.
Cheers,
Bart
-
For "Pro" model Epson printers if you wish to use 720ppi input resolution then as Bart says you need to select "Finest Detail", otherwise it will run at 360. This option/control is not available in non-Pro printers and is defined on the other media and resolution settings.
-
Thanks Phil and Bart, I was going to swear I had set to "finest details" but now that I have gone back and looked in the setup it appears to have not been set, I am guessing I forgot that I changed it doing some experiments. Doh! Doh! Thanks for dealing with the "dumb question" and senior moment. :-)
Alan
-
Thanks Phil and Bart, I was going to swear I had set to "finest details" but now that I have gone back and looked in the setup it appears to have not been set, I am guessing I forgot that I changed it doing some experiments. Doh! Doh! Thanks for dealing with the "dumb question" and senior moment. :-)
No problem, what matters is that it's solved. It's the result that counts, and that's what this thread happens to be about.
Cheers,
Bart
-
I thought the "Finest Details" setting was for printing vector graphics or text. Does it really improve quality in a photo?
-
That's the primary design purpose, and it does it partly by accepting 720 input res. Generally, I've not seen an improvement in output unless there are a lot of clean edges or straight lines. There's no downside, apart from the increased spool file size and perhaps slightly longer processing time.
Whether it makes a visible difference to your images at a normal viewing distance (or even a close viewing distance, excluding using a loupe) is really up to you to determine through testing.
-
I thought the "Finest Details" setting was for printing vector graphics or text.
Correct...it aliases edges in vector graphics such as type or EPS graphics...not intended for natural photographic images.
Upsample to 720 PPI is you want 720 PPI output.
-
I thought the "Finest Details" setting was for printing vector graphics or text. Does it really improve quality in a photo?
Hi Mike,
That as well. Do note that the setting is called finest detail, not "only for vector graphics or text". That option allows the printer to use (or upsample to) 720 PPI, which has the effect of rendering more detail (above 360 PPI) if it's present in the image data. Fine lines or text benefit from printing at 720PPI to avoid visible artifacts (e.g. jaggies).
Text and vector graphics are fed to the printer driver as rasterized image data, just like a continuous tone image. It makes no difference for the driver, it does not discriminate. A pixel is a pixel.
Obviously, processing 4x as much data (720 PPI versus 360 PPI in 2 directions) slows down the printing process, that's why it is an option, and not set by default. It helps to reduce complaints about slow printing.
Cheers,
Bart.
-
The 720 PPI requested input at Finest Detail etc setting is for the Wide Formats, I guess the desktops go for 1440 then. For the HP Z3100, the requested normal input resolutions are resp 300-600 PPI. While it becomes 1200 PPI with Maximum Detail selected (HP Prof Satin RC paper for example). In the Z3200 PCL3 or PS3 driver I can not get it that high, it stays on 600 PPI input but the droplet distribution jumps to 2400x1200 dpi like it does on the Z3100 (HP Prof Satin RC paper for example). The menu structure is different in the drivers but Qimage does not pick a 1200 PPI input either for the Z3200 so I guess HP brought it down for some reason, possibly an unrealistic long processing time needed for that setting.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Ernst
Going back to your "restest" file for a moment - if it is printed sized at 720 ppi, should you be able to see the individual lines in block 1, column A? (with a loupe, of course).
John
-
Bart - please do note that whilst it is called that, you should read the tool tip that comes up when you highlight that option - please see attached image.
One of the things that Finest Detail does is to provide for 720 input resolution - it's not the only thing it does and testing with lots of straight lines is not necessarily exactly the complete test between on and off that you might think if it were only changing input resolution.
It's not that the driver is receiving vector data, but it's expecting the rasterisation of that type of data and therefore can make certain assumptions when it does then process that data (not least of which is that the printer has higher resolution than PPI, because it has multiple dots per pixel.
Ernst - the non-Pro printers don't have a "Finest Detail" option. Input resolution is based on a variety of other driver choices.
Outside of the standard driver, various RIPs can achieve other options and settings.
-
Bart - please do note that whilst it is called that, you should read the tool tip that comes up when you highlight that option - please see attached image.
Well that's as cryptic as most 'info'. What does "make edges sharper" mean? Render them sharper than possible with upsampled 360 PPI, or does it add sharpening? I say the former, until someone can prove the opposite (which doesn't show from the printed test files sofar). In fact, increasing the sharpening on 720 PPI text would only increase the possibility of creating aliasing artifacts on text which is already optimally rasterized.
One of the things that Finest Detail does is to provide for 720 input resolution - it's not the only thing it does and testing with lots of straight lines is not necessarily exactly the complete test between on and off that you might think if it were only changing input resolution.
It's not the only thing it does? Sofar, nobody has come forward with any kind of proof for such a hypothesis.
Of course straight lines are not the only or best test. That's why I suggested the use of a zoneplate target (http://bvdwolf.home.xs4all.nl/main/foto/down_sample/down_sample_files/Rings1.gif) earlier in the thread (I also mentioned to convert the GIF from indexed to RGB mode to avoid artifacts from Photoshop's faulty resampling). Such a target has no bi-tonal black and white pattern, but rather an approximation of a sinusoidal pattern of gray tints with all possible spatial frequencies, and at all possible angles. It is also very sensitive to aliasing artifacts, so it becomes easy to differentiate between good and bad driver implementations. It's also likely to show if there is more going on than just enabling 720 PPI input (or resampling input to 720 PPI where needed).
It's not that the driver is receiving vector data, but it's expecting the rasterisation of that type of data and therefore can make certain assumptions when it does then process that data (not least of which is that the printer has higher resolution than PPI, because it has multiple dots per pixel.
So far, no proof of that has been presented, although I could think of a few things that could be done (but would take processing power that's available on a computer, but not in a printer). I'll repeat that a printer driver is not an image processing application, it's an interface to the basic functionality of a printer.
Outside of the standard driver, various RIPs can achieve other options and settings.
Indeed, and do a better job as well because there is much more processing power available than in an average printer, and data can be properly processed.
Cheers,
Bart
-
I ran several tests on my 4900 using Ernst's file. I've posted my horizontal dithering observations between my 4900 and 7880 on John's latest post and will try to keep this post to resolution and my 4900. Please note, all tests discussed herein were printed using PK ink printed at 2880 high speed (my print heads are aligned), finest detail off and printed on Epson Exhibition Fibre Paper with EFP chosen as the output paper within the 4900's printer driver (an option not available with the 7880 driver). Since, like John, I print from Lightroom, I'm interested in a file's output from LR and how changes within LR affect the output so all tests were printed from LR (no out sharpening or other alterations made to the file). When IR is different than print resolution/output resolution ("PR"), LR handled any file processing via its Print Resolution box. While I printed the full chart at an input resolution ("IR") of 180 across a couple of PRs, I decided to focus on Column A differences so other than IR 180, I'll just be discussing Column A output.
Not surprisingly, when IR is 720 and PR is 720, all appears fine. Unless I'm printing a pano, I rarely have a file with an IR 720 so I quickly moved on. With IR 360 and PR 360, there is no vertical banding across any sections. The 720 ppi section, however, appears very dark gray due to line spacing but faint vertical lines can be seen when viewing with a loupe. WIth IR 360 and PR 720, the 180 and 240 sections appear identical to the PR 360 result but the 360 and 720 sections appear lighter to the naked eye--with the 360 section appearing slightly lighter and the 720 section appearing much lighter. No loupe is required to see the vertical lines of the 360 section, but the 720 section appears to have narrower vertical spacing at PR 720 than PR 360. As a result, I'm not quite sure why PR 720's 720 ppi section appears lighter than PR 360's 720 ppi section--unless the driver is actually using a different black for that section.
IR 240. IR 240 and PR 240, all appears fine. While the 720 ppi section's vertical lines are not as clean looking as any of the others, vertical lines are visible with a loupe and evenly spaced. With PR 360, all sections appear lighter than their PR 240 counterparts (though 180, 240 and 360 barely so). Still no banding but the 720 section is not quite as "clean" as the other sections but is much cleaner than PR 240 (resulting in the lighter gray I am assuming). PR 720 produced the nicest results--no visible difference between the 180, 240 or 360 sections and a slightly lighter gray look to the 720 section probably due to "cleaner" vertical lines. This result seems to support Jeff's advice that if your file is below 360 PPI, print from LR with PR 360 for optimum results--at least for IR 240.
IR 180. IR 180 and PR 180, line thickness differences in Column D could be seen across all PPI sections. The 180 and 240 Column A sections looked fine, the 360 section had barely perceptible vertical banding and the 720 section had noticeable vertical banding which, when viewed with a loupe appeared to be due to varying, repeatable differences in vertical line thickness. WIth PR 360, all looks quite fine--no vertical banding of any sections and clearly defined vertical columns of the 720 section. I thought I also printed it at PR 720 but can't find it at the moment---I printed other IRs at PR 720 so I'm not overly concerned at the moment--perhaps I'll reprint.
More often than not, I find I'm printing files at an IR other than 180, 240, 360 or 720. As a result, I ran some additional tests with different IRs and had some surprising and interesting results.
IR 165. IR 165 and PR 165 is painful to view. The 180 ppi section looks fine, the 240 ppi section has vertical banding, the 360 section has vertical banding as well as vertical line convergence and the 720 section has an even greater frequency of vertical banding and vertical line convergence. At IR 165 and PR 360, the 180 section is fine, the 240 section has horizontal banding, the 360 section only has vertical line convergence, though at a much tighter frequency making it look better than PR 165 (and the banding is gone), and the 720 section is similar to the 360 section--no banding but a greater frequency of vertical line convergence, but it is not a painful print to view. Now the most surprising--IR 165 and PR 720-- everything looks perfect. No banding or vertical line convergence across any sections. The 720 ppi section's vertical lines could barely be made out by my naked eye and are clearly visible under the loupe. I was very surprised at this result.
IR 265. IR 265 and PR 265 looks horrible across all sections. The 180 section has vertical banding, the 240 section has vertical banding and convergence, the 360 section has some banding but a greater convergence and the 720 section has quite a lot of vertical convergence. The results are very bad. IR 265 and PR 360 showed vertical frequency banding on both the 180 and 240 sections and vertical line convergence on both the 360 and 720 sections. The 360 and 720 sections were ok to view, but the 180 and 240 sections were painful to view. IR 265 and PR 720--the 180 section was fine, the 240 section still had vertical banding (albeit narrower but more frequent), the 360 section had vertical line convergence, but looked much better than the 360 version. The 720 section had vertical line convergence (or thicker vertical lines at repeating distances), but at a greater frequency than the PR 360 version leading to a "cleaner" look. Overall, the IR 265 sections all looked better at PR 720 than PR 360.
For me, other than making the waters slightly murkier, if my IR is 240 or less, PR 360 should be fine. If IR is greater than 240, I'm going to change my PR to 720. I'd be interested to read other 4900 user results.
-
I ran several tests on my 4900 using Ernst's file. I've posted my horizontal dithering observations between my 4900 and 7880 on John's latest post and will try to keep this post to resolution and my 4900. Please note, all tests discussed herein were printed using PK ink printed at 2880 high speed (my print heads are aligned), finest detail off and printed on Epson Exhibition Fibre Paper with EFP chosen as the output paper within the 4900's printer driver (an option not available with the 7880 driver).
For me, other than making the waters slightly murkier, if my IR is 240 or less, PR 360 should be fine. If IR is greater than 240, I'm going to change my PR to 720. I'd be interested to read other 4900 user results.
Hi Shane,
Why do you turn Finest Details off? Lightroom can make use of it. When LR upsamples for an output resolution of 720 PPI, and you do not enable that resolution in the printer driver, then the printer driver will downsample it to 360 PPI (potentially leading to aliasing artifacts). Maybe LR is smart enough to restrict the resampling to 360 PPI even if you select 720 PPI, but then you are selling yourself short on resolution.
Cheers,
Bart
-
Bart,
I have finest detail turned off because it was my understanding, and Jeff posted in response to Mike's question that
Correct...it aliases edges in vector graphics such as type or EPS graphics...not intended for natural photographic images.
I don't have finest detail on for my images so I didn't turn it on for the test file.
-
I don't have finest detail on for my images so I didn't turn it on for the test file.
Shane, to perform this test (as I understand it) firstly you have to size the restest image to exactly 720 ppi. In LR this is tricky, but can be done. In the LR Print Module, choose a suitable paper size (A5 Portrait and place the file at the top) and size the file in the Layout Cell Size to precisely 14.11 x 9.326 cm. All you have to do is put in 14.11 as the long dimension, of course. You should see LR report this as 720 ppi. Then, you send it to the printer with no re-sampling in LR (in other words, the resolution box in the RH panel is not checked). If you can see a difference between the 720 and 360 line thickness in Column D, your printer is accepting 720 ppi input and outputting it correctly. If you can see no difference in line thickness (although the 720 lines might look a lighter gray) then the printer is resampling the incoming data to 360 ppi. In which case, check the "Finest Detail" box and try again. You should find that "Finest Detail" will switch the printer to look for 720 ppi input. This is more or less the present consensus of opinion ;)
John
-
I dont have anything to contribute, but I'd like to thank you for making this thread a good read.
-h
-
I dont have anything to contribute, but I'd like to thank you for making this thread a good read.
-h
h - we do our humble best ;)
Actually, I'm just trying to get up to 1,000 posts . . .
John
-
Well that's as cryptic as most 'info'. What does "make edges sharper" mean?
Well, the point being, you can't just make the assertion that you did earlier with any more certainty at this point. The vendor is telling us the purpopse of the design.
-
Well, the point being, you can't just make the assertion that you did earlier with any more certainty at this point. The vendor is telling us the purpopse of the design.
So does that mean that non-text and/or non-vector graphics, IOW continuous tone images, with fine edge/line detail (or even smooth gradients) will look worse when selecting the "Finest Detail" option?
Cheers,
Bart
-
So does that mean that non-text and/or non-vector graphics, IOW continuous tone images, with fine edge/line detail (or even smooth gradients) will look worse when selecting the "Finest Detail" option?
Cheers,
Bart
The best way to know for sure is to print real-world photographs with the characteristics you describe here at several settings, and examine them in normal viewing conditions, and if it's really important enough, under a magnifier. I think that's the only way to escape the risk of a prolonged discussion akin to angels dancing on the head of a pin.
-
So does that mean that non-text and/or non-vector graphics, IOW continuous tone images, with fine edge/line detail (or even smooth gradients) will look worse when selecting the "Finest Detail" option?
That is the implication I've gotten from Epson engineers...it either does nothing or does something slightly sub-optimal to non-vector graphics (raster)...
Course, you don't believe the rest of what I have reported regarding the Epson driver not resampling image data in the print pipeline...not much reason to expect you believe this :~)
-
That is the implication I've gotten from Epson engineers...it either does nothing or does something slightly sub-optimal to non-vector graphics (raster)...
Other than the engineers, there is also what the Manual says, and I'm looking at page 96:
"Choose a lower print quality setting for faster printing or SuperFine - 1440 dpi for best quality".
And for "Advanced Options, they say:
"Finest Detail for sharper edges on vecto-based data including text, graphics and line art. (This setting does not affect photographs and is not recommended for large files.)"
Adding the two statements together, this tells me we can select Finest Detail with SuperFine - 1440 and eat our cake and have it.
-
That is the implication I've gotten from Epson engineers...it either does nothing or does something slightly sub-optimal to non-vector graphics (raster)...
Course, you don't believe the rest of what I have reported regarding the Epson driver not resampling image data in the print pipeline...not much reason to expect you believe this :~)
Jeff,
You don't seem to be the kind of person to just believe what you are told, so what's your opinion after testing it yourself? You and Michael seem to have tested the "Finest Details" option, didn't you? So what's the outcome, does it hurt contone image quality? Does it allow to gain some resolution?
Cheers,
Bart
-
You and Michael seem to have tested the "Finest Details" option, didn't you?
No...I haven't tested with/without Finest Detail. I just now upsample in Lightroom to 720 PPI for printing. Since what the Epson guys have told me, that there's no benefit to photos, why would I bother?
-
No...I haven't tested with/without Finest Detail. I just now upsample in Lightroom to 720 PPI for printing. Since what the Epson guys have told me, that there's no benefit to photos, why would I bother?
Because without Finest Detail it will be resampled back down to 360ppi (or 180ppi if printing on roll media).
-
Because without Finest Detail it will be resampled back down to 360ppi (or 180ppi if printing on roll media).
Roll or sheet has nothing to do with it.
To answer Bart - as I said originally, there's no particular downside except for longer processing times, for real world images.
-
Roll or sheet has nothing to do with it.
http://www.luminous-landscape.com/forum/index.php?topic=48894.msg403784#msg403784
-
Well, if Shane (or anyone else) would just follow my instructions using the restest file, and send it to his 4900 with and without the "Finest Detail" box checked, we would have the answer to this endless debate (at least for the 4900 and its driver). I don't know what the result will be, because as I reported, on my 2400 all photo input for gloss paper is expected to be 720 ppi. Which really did surprise me.
John
-
To answer Bart - as I said originally, there's no particular downside except for longer processing times, for real world images.
Indeed, that seems to be the consensus amongst those who have tested it, although Jeff seems to have another opinion based on something that was told to him. That's why I asked him if his position was also based on personal observation, which it apparently isn't.
Which raises another question, since Jeff was involved in structuring the most recent LR output logic/workflow, why let Lightroom upsample to 720 PPI (assuming that is what it does when instructed to do so), and let the printer driver downsample to 360 PPI ? All the empirical evidence I've seen, and heard being described by others (including Eric Chan and Mike Chaney), indicates that resampling (up or down) to 360 PPI is happening unless the Finest Detail option is set, as witnessed by aliasing artifacts which would not be generated by dithering alone.
Cheers,
Bart
-
Gromit - yes, but that's not changing the input resolution.
-
John,
I did print the full test chart at 720. Living in the land of inches, I could never get an exact size within LR to print at 720 so I adjusted the canvas size (not image size) in PS to even numbers, went back to LR, set the cell size and printed (LR was reporting the image at 720). While I'm fairly certain there was a difference in line thickness between 360 and 720's Column D, I need to pull the print and look. I've been out of pocket since yesterday afternoon, but will look at it either this evening or in the morning and post the result.
-
Indeed, that seems to be the consensus amongst those who have tested it, although Jeff seems to have another opinion based on something that was told to him. That's why I asked him if his position was also based on personal observation, which it apparently isn't.
Always happy to learn something...I tried printing out on my 4900 with and without Finest Detail selected. With an image at 720 PPI, I can see a tiny improvement in high frequency image data when the Finest Detail option is selected. So, it looks like I'll be selecting that option when printing on printers that offer it–which is the Epson pro line.
The R3000 that I have does not offer that option. It does have two different 1440 dpi settings; SuperFine and Photo. The settings I've been using with that printer is SuperPhoto which states 5760 dpi.
-
Always happy to learn something...I tried printing out on my 4900 with and without Finest Detail selected. With an image at 720 PPI, I can see a tiny improvement in high frequency image data when the Finest Detail option is selected. So, it looks like I'll be selecting that option when printing on printers that offer it–which is the Epson pro line.
The R3000 that I have does not offer that option. It does have two different 1440 dpi settings; SuperFine and Photo. The settings I've been using with that printer is SuperPhoto which states 5760 dpi.
Great - glad you did that Jeff.
-
I did print the full test chart at 720. Living in the land of inches, I could never get an exact size within LR to print at 720 so I adjusted the canvas size (not image size) in PS to even numbers, went back to LR, set the cell size and printed (LR was reporting the image at 720).
Shane, you can't set the cell size in LR accurately enough if you have the measurements set in inches. A hundreth of an inch is bigger than a hundreth of a centimetre - 2.54 times bigger. You have to select the metric option in order to alter the cell size in small enough increments to get exactly 720 ppi for the test. Check your output in column D with a decent loupe.
John
-
Always happy to learn something...I tried printing out on my 4900 with and without Finest Detail selected. With an image at 720 PPI, I can see a tiny improvement in high frequency image data when the Finest Detail option is selected. So, it looks like I'll be selecting that option when printing on printers that offer it–which is the Epson pro line.
Hi Jeff,
That's good news. Who would say no to better quality at a price that cannot be beat? One more item that can be eliminated from the QED list. I assume you tested with Lighroom 3, which would mean that selecting the 720 PPI option there now really results in 720 PPI detail without downsampling by the printerdriver to 360 PPI, if available. Did you notice any adverse effects as alluded to by the Epson enigineer? Maybe that only happens in very rare and specific/theoretical scenarios?
Cheers,
Bart
-
I assume you tested with Lighroom 3, which would mean that selecting the 720 PPI option there now really results in 720 PPI detail without downsampling by the printerdriver to 360 PPI, if available.
Correct...
Did you notice any adverse effects as alluded to by the Epson enigineer? Maybe that only happens in very rare and specific/theoretical scenarios?
Well, my test was really only a few prints with and without Finest Detail checked so I would NOT call that exhaustive. I'll be seeing some of the Epson people next month and see if I can drill down on the Finest Detail setting.
Note, edited my response to add the word "NOT"
-
So... any chance we can have a summary of what's been discussed here?
It sounds like, for Epson printers, if the image resolution is below 360, then use a Print Resolution of 360 (in LR). Finest Detail won't have any benefit at this resolution. If IR is above 360, then use a PR of 720 and turn on Finest Detail. Yes? No? I know... it depends!
-
So... any chance we can have a summary of what's been discussed here?
How did you like your thread, Mike? It got quite good, didn't it ;) As for a summary, we're still in the thick of it here . . .
Seriously, as far as I can see yes it still all depends - on your output software, your printer model, and just as important the size of print you are making and whether that involves up or down-sampling to achieve that size on paper. The rule of thumb which Jeff set out at the beginning is still a good one in general, for LR and Epson printers. The debate has been about why it works, not whether it works. However, in some cases as we have seen with my R2400, you would probably be better off sending all your print data whatever size print at 720 ppi, otherwise it could be resampled twice (not a good thing). And the only way to be sure about this is to thoroughly test your own printer and software combination. As Mark said, you have to trust your own eyes.
John
-
Well, my test was really only a few prints with and without Finest Detail checked so I would call that exhaustive. I'll be seeing some of the Epson people next month and see if I can drill down on the Finest Detail setting.
Great, that's most helpful. Do not hesitate to also raise the issue of resampling to 360 or 720 PPI, versus dithering. All the effects we can see point to resampling, but it's all based on the resemblance of the artifacts with those from resampling. Although unlikely, it might well be something different and only they can know for sure.
Cheers,
Bart
-
How did you like your thread, Mike? It got quite good, didn't it ;) As for a summary, we're still in the thick of it here . . . John
Quite good doesn't quite cover it John. I've learned a lot - much of it is still over my head - but I think it's sinking in. That's why I asked for the summary. ;)
So, just to be sure I have the test right, I should try a Print Resolution (set in LR Print module) of 720 with Finest Detail On, and Off as well as the 360 setting to see what's best for my combination?
Now,lets say I get the best results at 720 with Finest Detail on... would there be a problem with using this setting all the time? I realize there may not be a visible difference, but if it doesn't lower the quality, then I shouldn't have to do a test for each image that I want to print.
-
Now,lets say I get the best results at 720 with Finest Detail on... would there be a problem with using this setting all the time? I realize there may not be a visible difference, but if it doesn't lower the quality, then I shouldn't have to do a test for each image that I want to print.
No currently known downside other than slightly slower spooler processing and printing.
-
Slightly slower doesn't bother me for this type of printing so that's good to know.
-
Do not hesitate to also raise the issue of resampling to 360 or 720 PPI, versus dithering.
At this point it's unclear if or when any upsampling occurs on Mac and Windows. There's some indication that with older print pipelines the device driver reported a "device resolution" to the system level OS print pipeline. I think that the older (meaning pre Win 7 and Carbon based on the Mac) pipeline "assumed" that the image data passed to the print driver needed to be a specific resolution based on what the devise was "asking for" (meaning, reporting to the OS). Whether or not the system level print pipeline resampled image data and the manner of the resampling is unclear.
I'm still kinda sticking to the statement that the print driver for Epson doesn't resample the image data line for now. I'm now thinking it's possible that somewhere in the print pipeline some sort of resampling gets done. What not clear is where the resampling may be done nor the manner of the resampling (flavor of the algorithm).
What is clear is that with regards to Lightroom and Epson pro printers, doing the upsampling to 360/720 (depending on the native resolution of the file) and then choosing the correct driver settings (Finest Detail off if upsampling to 360 from below 360 native rez and Finest Detail on if upsampling to 720) will produce the finest image detail in the print.
Eric Chan is somewhat involved in the discussion and so is the engineer for Photoshop. I'm also pushing this up the pole to people at Epson (I have zero contacts at HP & Canon) to get some sort of official response. I will say that my contacts at Epson are of course, very interested in advancing the art of printing to Epson printers. If it turns out that we can recommend a new "best practices" to them, they'll adopt it. Of course it'll take a while to trickle down to manuals and such. But there may be a "white Paper" that could be released that gives a definitive answer.
For now the essence is for Epson pro printers, if the native resolution of the image size is below 360, upsample to 360 and don't select the Finest Detail setting in the driver. If the native rez is above 360 but below 720, resample to 720 PPI and make sure to select the Finest Detail option. Presumably for Canon and HP the magic numbers are 300/600. But I have zero experience in configuring HP or Canon print drivers so you'll need to test this out for yourself.
-
[...]
What is clear is that with regards to Lightroom and Epson pro printers, doing the upsampling to 360/720 (depending on the native resolution of the file) and then choosing the correct driver settings (Finest Detail off if upsampling to 360 from below 360 native rez and Finest Detail on if upsampling to 720) will produce the finest image detail in the print.
Indeed, by sending resampled (to 360/720 PPI) data to the printer, we can make sure that the printer or printer driver don't do that. All risks are eliminated, and the sending application has full control over the quality of resampling and can do sharpening at that final resolution. All that people need to be aware of is that selecting borderless printing will again invoke a re/upsampling operation with a variable amount (and thus sensitive to the unknown quality of that resampling).
Eric Chan is somewhat involved in the discussion and so is the engineer for Photoshop. I'm also pushing this up the pole to people at Epson (I have zero contacts at HP & Canon) to get some sort of official response. I will say that my contacts at Epson are of course, very interested in advancing the art of printing to Epson printers. If it turns out that we can recommend a new "best practices" to them, they'll adopt it. Of course it'll take a while to trickle down to manuals and such. But there may be a "white Paper" that could be released that gives a definitive answer.
Great, that's helpful. Thanks for your cooperation.
For now the essence is for Epson pro printers, if the native resolution of the image size is below 360, upsample to 360 and don't select the Finest Detail setting in the driver. If the native rez is above 360 but below 720, resample to 720 PPI and make sure to select the Finest Detail option. Presumably for Canon and HP the magic numbers are 300/600. But I have zero experience in configuring HP or Canon print drivers so you'll need to test this out for yourself.
Canon/HP printers sofar usually default to 300/600 PPI with the appropriate driver settings for paper and quality. Programs like Qimage report back what the printer driver expects as input, and on-the-fly automatically adjusts its resampling and smart sharpening to that final output size. So if a specific printerdriver setting would require/allow anything different, it will be taken care of automatically. The user is given control over the amount of smart sharpening, and that's all that's needed because it will allow to adapt to different print media.
So, unless the printer manufacturers can point out any drawback for this safe practice of requested resampling and sharpening before sending to the printer(driver), it would seem a logical move for the software developers to optimize the workflow and quality at their end, based on what the printer driver reports to expect.
Cheers,
Bart
-
John,
I did look at the 720 input and output file and, as expected, there is a definite difference in line thickness between the 720 ppi and 360 ppi in column D. This is true whether the print driver is printing 8-bit, 16-bit, 1440 or 2880 or 2880 with fine detail on or off. One difference I did note in printing with finest detail off is the 720 ppi rectangle of Column A is a lighter gray to the naked eye than the other ppi rectangles when finest detail is not selected and I cannot make out any hints of vertical lines when viewing under a loupe (I just see darker spots splotched on top of a lighter, seemingly solid background). With finest detail on, the 240, 360 and 720 rectangles are all darker to the naked eye, and hints of vertical lines may be seen in the 720 rectangle, though not clearly defined.
On the image size front, yes, I could not get the image to the correct DPI within LR when working with US measurements. In PS, I merely changed the canvas size to 6x4 inches which did nothing to the image's ppi. In LR, I then set the cell width to 6 inches and all was fine.
-
I've used QImage for years on the Hp and it does indeed to an excellent job, especially when upsampling for very large prints from dslr files, as in sending them to the printer at 150 ppi.
I'm wondering if anyone knows exactly what is going on with the Canon IPF plug-in? It seems to be doing something similar for file optimization and also has a slider for output sharpening adjustments that can be changed for various media.
john
-
John,
I did look at the 720 input and output file and, as expected, there is a definite difference in line thickness between the 720 ppi and 360 ppi in column D. This is true whether the print driver is printing 8-bit, 16-bit, 1440 or 2880 or 2880 with fine detail on or off. One difference I did note in printing with finest detail off is the 720 ppi rectangle of Column A is a lighter gray to the naked eye than the other ppi rectangles when finest detail is not selected and I cannot make out any hints of vertical lines when viewing under a loupe (I just see darker spots splotched on top of a lighter, seemingly solid background). With finest detail on, the 240, 360 and 720 rectangles are all darker to the naked eye, and hints of vertical lines may be seen in the 720 rectangle, though not clearly defined.
Thank you for the update, Shane. Now what does all this mean? What your results seem to be telling us is that, with the 4900 -
* If you send your printer 720 ppi input, it will process it at 720 no matter what the output quality settings are in the print driver user interface. This mirrors my result with the 2400. (I have a nasty feeling that the opposite is also true, but I have another test to run)
* The “Finest Detail” box does something extra, but it does not select or deselect 720 ppi as has been suggested. It more likely refines the dither pattern in some fashion, hence your observations in block 1, column A. My 2400 when printing 720 ppi with no resampling also prints a block 1 where the lines are just visible with an 8x loupe.
This could change our view of the Epson world, yet again . . .
John
-
This could change our view of the Epson world, yet again . . .
That's why I have a bit of a problem with drawing conclusions based on Lightroom output. LR also does it's thing, and doesn't offer enough control to set an exact input PPI, or so it seems. That's fine for regular use (for which it is intended), but not optimal for testing. A test from e.g. Photoshop probably offers more control.
I'd wait a bit for (in)direct feedback from someone like Eric Chan, who also is familiar with Epson printers.
Cheers,
Bart
-
I've used QImage for years on the Hp and it does indeed to an excellent job, especially when upsampling for very large prints from dslr files, as in sending them to the printer at 150 ppi.
I'm wondering if anyone knows exactly what is going on with the Canon IPF plug-in? It seems to be doing something similar for file optimization and also has a slider for output sharpening adjustments that can be changed for various media.
Hi John,
Are you using any other settings than the ones on the Main tab of the print output plugin (as shown here (http://www.youtube.com/watch?v=Mhl9oowO-Bs))?
Cheers,
Bart
-
That's why I have a bit of a problem with drawing conclusions based on Lightroom output. LR also does it's thing, and doesn't offer enough control to set an exact input PPI, or so it seems. That's fine for regular use (for which it is intended), but not optimal for testing. A test from e.g. Photoshop probably offers more control.
Bart, the procedure that Shane described should have given him exactly 720 ppi for the input. And I can get the file to exactly 720 ppi also, using the cell size I posted earlier. I think the results he has are genuine, and are not influenced adversely by using LR in this particular instance. If that "Finest Detail" box was a 720/360 switch, he would have seen a difference in the 720 line thickness in column D of the restest printout with the box checked/not checked, there is just no way of avoiding that even if the file size had been 2 or 3 ppi off. This surprises me as much as it does you, but I have to say it looks as if Jeff and Mark are correct on this one, and that "Finest Detail" does something to the dither but not the ppi. On the 4900, at any rate.
John
-
So, unless the printer manufacturers can point out any drawback for this safe practice of requested resampling and sharpening before sending to the printer(driver), it would seem a logical move for the software developers to optimize the workflow and quality at their end, based on what the printer driver reports to expect.
There are a few potential drawbacks. I just talked to a friend at Epson and he indicated that if you are going to go to the trouble of upsampling to 720 and using the Finest Detail that the spool file will grow large (particularly if you are printing a file that is large to start with) and depending on the way the printer is connected; if over a network there could be problems or slow downs. If the network is fast and doesn't have a lot of other traffic and you don't hear the printer pausing then no problems, it's just taking more time.
The other thing he mentioned is that when printing at 720 with Finest Detail you should make sure the print head has been aligned on the exact media you'll be printing to and that you'll need to use a uni-directional setting not bi-directional. The spray alignment for bi-directional can cause problems when the second path of the head is also printing. This can be an issue because the printer will be 1/2 the speed. So you'll be sending more data and half the speed and the whole process will be less efficient. That would be an issue in a production environment.
As far as the question of whether or not the print driver is doing resampling or if it's the OS print pipeline, he reiterated that the print driver itself doesn't do resampling and the sieve analogy is indeed accurate. He's pretty sure it's the OS print pipeline that is doing the resampling and the nature of the resampling algorithm is unknown. But the odds of it being a really optimized resampling isn't high. The guess is either a bi-linear or nearest neighbor just from the standpoint of speed. I know a few people at Apple and MSFT but I don't know specifically whom to ask and I'm not sure Apple would even answer the question...
But The Epson guy agreed that sending the image data at the resolution the printer is expecting should take the print pipeline out of the equation.
-
There are a few potential drawbacks. I just talked to a friend at Epson and he indicated that if you are going to go to the trouble of upsampling to 720 and using the Finest Detail that the spool file will grow large (particularly if you are printing a file that is large to start with) and depending on the way the printer is connected; if over a network there could be problems or slow downs. If the network is fast and doesn't have a lot of other traffic and you don't hear the printer pausing then no problems, it's just taking more time.
Indeed, there is no free lunch, one either goes for optimal quality or better speed. In a production environment that latter is probably the more economical chioce.
The other thing he mentioned is that when printing at 720 with Finest Detail you should make sure the print head has been aligned on the exact media you'll be printing to and that you'll need to use a uni-directional setting not bi-directional. The spray alignment for bi-directional can cause problems when the second path of the head is also printing. This can be an issue because the printer will be 1/2 the speed. So you'll be sending more data and half the speed and the whole process will be less efficient. That would be an issue in a production environment.
Yes, Uni-directional and well aligned for the medium will help. Speed will suffer, unless one was already printing uni-directional.
As far as the question of whether or not the print driver is doing resampling or if it's the OS print pipeline, he reiterated that the print driver itself doesn't do resampling and the sieve analogy is indeed accurate. He's pretty sure it's the OS print pipeline that is doing the resampling and the nature of the resampling algorithm is unknown. But the odds of it being a really optimized resampling isn't high. The guess is either a bi-linear or nearest neighbor just from the standpoint of speed. I know a few people at Apple and MSFT but I don't know specifically whom to ask and I'm not sure Apple would even answer the question...
I'm not sure if your liason is a software or hardware guy, but there is also the posibility that the printer firmware (instead of the driver) does some of the processing. Firmware, or ASICs, are faster than software but less flexible. Maybe they moved some of the logic to the printer hardware for speed gain purposes. It seems unlikely to me that the OS meddles with the print data, but stranger things have happened. I'm assuming from his reaction that he also recognizes that aliasing is involved somewhere (because of the aliasing artifacts), so that source would still need to be identified for an explanation of the observations.
But The Epson guy agreed that sending the image data at the resolution the printer is expecting should take the print pipeline out of the equation.
That's a good observation to take away from the discussion sofar, assuming that the OS doesn't interfere. We can control the preparation part of the workflow to our heart's content and deliver it resampled and sharpened, ready for print at the requested PPI, thus bypassing as many variables as possible. One can alway deliberately choose for 360 PPI as a trade-off if speed is a concern.
Cheers,
Bart
-
It seems unlikely to me that the OS meddles with the print data, but stranger things have happened.
This analogy might seem far-out, but if you want to play a CD on your Windows PC, without Windows altering any bits, this could be quite cumbersome. Thing is, Microsoft had a mixer subsystem that did (low-quality) sample rate conversion in software in order to make it easier for multiple applications to access the system, and to be able to interface to various soundcards.
This just goes to show that consumer OS-er are made for the normal people that just want sound or printing to work, at a low prize. For nerds that will go to any length for even the most minimal improvement, they will often have to.... go to great lengths :-)
-h
-
It seems unlikely to me that the OS meddles with the print data, but stranger things have happened.
Well, if the OS isn't doing the resampling, why would the OS need the device to report it's resolution? Both the Epson guy and one of the Photoshop engineers are thinking it's the OS print pipeline that is doing the resampling to send the image data at the resolution asked for by the printer.
As far as the OS "meddling"' with the print data, well, we've all seen that before. Colorsync is noted for it's meddling ways. And while in general Windows tends to be far more hands off, print drivers are still at the mercy of the printing APIs.
I don't doubt that the printer's firmware is doing the heavy lifting of doing the dither using special purpose processing chips. But as I understand it (and re-confirmed during a phone call this AM) the Epson guys still maintains that the print driver and printer aren't doing the resampling just doing the dithering.
Unless and until we can confirm with Apple or MSFT, I'm sticking with the OS being the resampler and avoiding the issue by doing the resampling in Lightroom.
-
Jeff
The people who must know, of course, and who are not involved here, are the engineers who write RIPs.
John
-
The people who must know, of course, and who are not involved here, are the engineers who write RIPs.
Not necessarily...RIPS that work on Epson printers generally bypass the system APIs and have direct print head control. ImagePrint does that by completely bypassing the application, the OS print pipeline and sends it's own dither to the print head.
No, the ones that really know are the engineers at Apple and MSFT that are developing the print pipeline APIs. And I don't know any of them...
-
No, the ones that really know are the engineers at Apple and MSFT that are developing the print pipeline APIs. And I don't know any of them...
Dammit, Jeff, why not? ;)
We need you to broaden your social horizons . . .
John
-
Well one. On that Lexjet video they don't mention the Advanced printing menu where you can select Unidirectional. I always set it at Uni if I'm doing small prints or if time is not a factor for me for larger things. As for the output sharpening I haven't done tests with it yet and I'm curious to see if it matches QImage. I haven't set up QImage on the Canon. There are too many damn tests to do in this world when things are looking so good out of the box as they are.
john
Hi John,
Are you using any other settings than the ones on the Main tab of the print output plugin (as shown here (http://www.youtube.com/watch?v=Mhl9oowO-Bs))?
Cheers,
Bart
-
Well one. On that Lexjet video they don't mention the Advanced printing menu where you can select Unidirectional. I always set it at Uni if I'm doing small prints or if time is not a factor for me for larger things.
Okay, setting it to uni-directional makes sense to avoid the mechanical tolerances. One of those variables one would like to avoid when the magnitude is unknown. According to Canon Techsupport (as mentioned in the Canon IPF Wikispace) "... , the main difference is that there will be a different "angle" or "lean" to the droplets depending which was the head was moving, and that in some cases the reflectivity may vary enough from this to be perceived as banding". Do note that that was in relation to the IPF5000 model, newer ones may have solved that with the newer ink formulations for the x300 series. Again, it's better to be safe than sorry, unless time is at a premium.
As for the output sharpening I haven't done tests with it yet and I'm curious to see if it matches QImage. I haven't set up QImage on the Canon. There are too many damn tests to do in this world when things are looking so good out of the box as they are.
I agree, but for the best quality one needs to suffer a bit, apparently. Anyway, Qimage remembers the printer driver settings per type of paper or job used, so the learning curve is short and the reward is sweet ...
Cheers,
Bart
-
This day I had some time to test further. I could use a friend's R2880 and R3880 driven by a Mac and printing from Photoshop CS5 but due to some limitations, he uses the 2880 for gloss and the 3880 for matte, so I did not want to switch between extreme print qualities related to those choices and the results were not convincing. Add to that my clumsiness on the Mac. It could even be that PS CS5 or OS-X interferes more. I see some moiré when the target is twice the resolution of the requested input resolution 1440 to 720 PPI, 720 to 360 PPI, but not the shift to complete black or white or huge blocking to B&W. I had Adobe Color Printer Utility for Mac there too but ran into a problem with its image sizing, a flaw already reported on its normal use for printing profiling targets. Too little time to cope with that on the Mac. All in all not a convincing test method and test result.
At home I tried the things I have done with the HP Z3100 and Qimage some years ago but this time with an office printer HP K5400 as it works faster with individual sheets. Printing from ACPU and Photoshop CS4 on a Windows Vista 64 system. The printer knows roughly 3 print quality grades: 300 PPI input, 600 PPI input, 1200 PPI input (one photo paper only). With Photoshop feeding line targets of 600 PPI, 1200 PPI, 2400 PPI respectively (last two sized to 50%, 25%) I loose the complete top field of column A, the rectangle becomes white. Feeding a line target 1200 PPI to the printer asking 1200 PPI input gives a dark grey rectangle, lines bleed completely to one another but no resampling happening if I check the top field in column D, so a genuine 1200 PPI input request.
With ACPU I get blocks in B&W in the A column top field and more. Problem with ACPU is that it resizes the original size so I have to measure the print and resize (without resampling) in Photoshop to get the exact print size of the original target tiff. Could be that a more precise printed target will also have a white or black rectangle there instead of B&W blocks. Sizing factor to be used in PS is somewhere between 1.037 to 1.042. But aliasing happens heavily with any print close enough. I would not expect any sophisticated downsampling/anti-aliasing in ACPU as it is intended for simple profile target printing at 1:1 or let us say sort of 1:1 :-)
In short resampling to 300, 600 and 1200 PPI happens somewhere en route and no anti-aliasing is done to overcome heavy aliasing/moiré in both PS CS4 and ACPU on Windows Vista 64.
Will test the Z3200 and Z3100 (again) to see whether something has changed between the two models/firmware/drivers but the Z3100 behaved very similar to the K5400 when Qimage was used with all its resampling off, running on Windows XP 32.
Edit: tested them too from Qimage running on Vista 64 and all extrapolation + sharpening OFF on Qimage, the Z3100 behaved like the K5400: 300, 600, 1200 PPI rendering resolutions. The Z3200 has 300 and 600 PPI as rendering resolutions. The Z3200 printed a better 600 PPI target at 600 PPI than the Z3100 at both 600 and 1200 PPI rendering resolutions. The resolution on both printers is better at the head travel direction than at the media trans[port direction, more or less predictable. Second Edit: The most recent Z3100 firmware and/or driver has just 300 and 600 PPI rendering resolution. 1200 PPI is no longer there.
The targets are from ddisoftware and placed there and/or created by Mike Chaney. Do not give them my name, I only gave the link to them.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Not necessarily...RIPS that work on Epson printers generally bypass the system APIs and have direct print head control. ImagePrint does that by completely bypassing the application, the OS print pipeline and sends it's own dither to the print head.
No, the ones that really know are the engineers at Apple and MSFT that are developing the print pipeline APIs. And I don't know any of them...
And do not underestimate the knowledge of the developers that made GimpPrint > GutenPrint for more OS systems. When in doubt Robert Krawitz often had an answer.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Thanks for that update, Ernst. I have some more results coming shortly using LR with real photographs rather than a test target, this time.
John
-
What is clear is that with regards to Lightroom and Epson pro printers, doing the upsampling to 360/720 (depending on the native resolution of the file) and then choosing the correct driver settings (Finest Detail off if upsampling to 360 from below 360 native rez and Finest Detail on if upsampling to 720) will produce the finest image detail in the print.
Jeff,
I think there are cases where data with less that 360 PPI at print size and resampled to 360 PPI will deliver a lower "general" image quality if the same paper/printer can deliver better image quality with enough data available. Detail is one thing, smoothness of gradients is another thing. If the paper/printer combination can not deliver a better quality with data above 360 PPI at print size then two conditions are met to set the limit at 360 PPI input.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
It's not the only thing it does? Sofar, nobody has come forward with any kind of proof for such a hypothesis.
Of course straight lines are not the only or best test. That's why I suggested the use of a zoneplate target (http://bvdwolf.home.xs4all.nl/main/foto/down_sample/down_sample_files/Rings1.gif) earlier in the thread (I also mentioned to convert the GIF from indexed to RGB mode to avoid artifacts from Photoshop's faulty resampling). Such a target has no bi-tonal black and white pattern, but rather an approximation of a sinusoidal pattern of gray tints with all possible spatial frequencies, and at all possible angles. It is also very sensitive to aliasing artifacts, so it becomes easy to differentiate between good and bad driver implementations. It's also likely to show if there is more going on than just enabling 720 PPI input (or resampling input to 720 PPI where needed).d.
Cheers,
Bart
Bart,
For testing image quality to the eye and a loupe with different workflow/paper/printer/print quality settings your zoneplate target is way better. To get a clear answer on the resampling question I preferred the simple line targets.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I'm not sure if your liason is a software or hardware guy, but there is also the posibility that the printer firmware (instead of the driver) does some of the processing. Firmware, or ASICs, are faster than software but less flexible. Maybe they moved some of the logic to the printer hardware for speed gain purposes. It seems unlikely to me that the OS meddles with the print data, but stranger things have happened. I'm assuming from his reaction that he also recognizes that aliasing is involved somewhere (because of the aliasing artifacts), so that source would still need to be identified for an explanation of the observations.
Cheers,
Bart
Printers taking part in the preparation of print data is more likely to happen on wide to medium format pro models than on (consumer) A3+ desktop models. The last seem to depend on 720 PPI input as a normal condition while the pro models most of the time use 360 PPI input if this thread delivered the right information. Computing a 1/16 or 1/8 square meter of printer data is of course demanding less of system power than one square meter of data. Laying down 720 PPI information on a 112 cm wide paper roll asks more of the hardware components addressing and of climate control. Translating 720 PPI information to 2 picoliter droplets (that are harder to place than 5 picoliter droplets) is not a optimal choice for production wide formats and would increase processing and printing time while hardware and climate control set a limit anyway on real optical resolution. So the pro models may do more in the printer itself, the desktop models may need the same work done but delegate it to the computer system. It would be interesting to see the information density that goes to a desktop inkjet head and a pro inkjet head.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I expect that most of the Forum readership is now beyond bored with this epic thread, but for the sake of completeness here are the latest results from Smith Labs. If you print from Lightroom to an Epson printer, they could be of interest. I have not solved the mystery of whether the OS or the Epson print driver resamples incoming data or not, but I have found out a bit more about what LR is doing. As before, my output was from Lightroom 3.4.1, running on Win 7 64-bit, and printing to my Epson R2400. The 2400 was set to its best output resolution, Photo RPM (5760 x 1440), High Speed off, and the paper was Epson Premium Glossy. LR Output Sharpening was off. My test files were 39MP Hasselblad 3FR raw files, and the specific file I used for this test is attached at the bottom of the post just for reference – this VW Campers pic is an old warhorse, but I use it because it is of good quality, well-focused and has no camera shake. I used just a section of the picture from the left of the frame, the cups on the table.
I decided to print my test only at exact divisions of the maximum 720 ppi input for the Epson, so I used 180, 360 and 720 ppi, in order to eliminate any re-sampling artefacts which might be introduced by the OS or the Epson Driver (if they do in fact resample). I started out by cropping a section of the image and sized it in the LR print module to exactly 360 ppi, fitting nicely on one third of an A5 sheet (so that I could print three tests side-by-side for comparison). This represents a section from a 20x15 print of the full-frame image. I printed it first to the 2400 directly at 360, and then tried resampling it to 720 ppi. The second result was obviously better even to the naked eye, with crisper definition and more detail on the pattern on the cups, crisper text on the tablecloth and an overall better internal contrast, particularly in the half-shadow areas. But why? No extra detail has actually been added, of course – the detail in the file is exactly the same, but you would swear it had been enhanced. At first I thought that perhaps the uprezzing routine was invoking an enhanced dither pattern in the print driver, but the dot density seemed the same under the 8x loupe. So I decided to go BIG, in order to get a closer look at what was going on.
This time I cropped a section of just the cups to 180 ppi and ran the test again – firstly a straight output at 180, then upsampled to 360 and also 720 ppi. This time I could see that LR is in fact not just resampling but running a very sophisticated anti-aliasing and smoothing routine on the image data into the bargain. It’s good at 360, and actually slightly better again at 720. And that’s why the “Schewe Rule” works. Have a look at the two files attached below – these are scans from the test prints, a section of the saucer rim on the right-hand cup. The first is the straight print at 180 ppi, the second has been resampled by LR to 720 ppi. Both are the same size in print, of course. You should be able to see how LR has smoothed the saucer rim.
Conclusions? Or perhaps a summary for Mike . . .
• We are talking about striving for the very finest prints here, squeezing the last little drop of quality out. Many people would never notice the difference – the current printers are so good, that almost any settings produce very acceptable output.
• For the best quality, set your print driver to either 5760x1440 (for the desktops) or 2880x1440 (for the Stylus Pro range). On the SP printers checking the Finest Detail box to on may give you a touch more definition in grass, hair, and fine lines. Make sure High-Speed or Bi-Directional printing is off.
• If your image sizes to less than 180 ppi in the LR Print Module for your selected paper size, you are trying to print it too large – but you may still get a reasonable result. A purpose-built uprezzing package might do better.
• From 180 ppi upwards to (presumably) 719 ppi, you will get the very best results by resampling the file in LR to 720 ppi. Between 180 and 359, you can resample to 360 to save time and spool file size, and get almost as good a result, but 720 is the business. What happens is that of course no extra detail can be added, but LR not only resamples but uses a very clever anti-aliasing and smoothing routine on the image, which enhances the apparent detail and resolution.
• This does not happen when LR is downsampling (as from 741 to 720 ppi). The image is not smoothed and no anti-aliasing is applied, so you should avoid this situation. The print quality suffers as a result (although it takes a loupe to see it). Unfortunately, with 60MP and now 80MP backs (and probably 100MP+ before too long) this is going to be of increasing concern for small prints.
• Strangely, the absolute worst thing you can do is to resample the image in LR to the same value as reported for the cell size – for example, take a 360 ppi image resolution and then set the LR output resolution to 360 as well. Theoretically, it should make no difference – but the quality of the print is degraded compared with simply sending it to the printer with no LR resampling. I can’t explain why this happens, but it is observable and repeatable. I can also see the same result with the Restest file – comparing output direct at 720 with the same file output with the LR Resolution box also set to 720 ppi (the sample text is degraded).
• Setting the LR output sharpening to “Standard” with the correct paper type will give the print a little bit of “snap” without introducing any artefacts or halos. It’s not a huge difference with my 3FR files, just nice and subtle.
That’s my conclusions for tonight, at any rate. You are now invited to weigh in and prove me wrong ;)
John
-
That’s my conclusions for tonight, at any rate. You are now invited to weigh in and prove me wrong ;)
Thanks for the work and reporting. I think unless told otherwise, your conclusions are nicely laid out and easy to comprehend.
-
Yes, thanks John for the 'summary'. I like those!
I just printed an image that natively came down to about 121 ppi after some cropping and wanting to print at about 36" wide. I set the Printer Resolution to 360 and it came out beautifully at a 'normal' viewing distance - at least to my eyes. I'm happy.
-
I just printed an image that natively came down to about 121 ppi after some cropping and wanting to print at about 36" wide. I set the Printer Resolution to 360 and it came out beautifully at a 'normal' viewing distance - at least to my eyes. I'm happy.
Mike, doing these tests has made me realise that I could print my 39MP files to 40x30 ins at 180 ppi using LR and they would look absolutely fantastic - no grain, pefectly smooth, beautiful tonality. Well, I could if I had a printer which would print that big. Bound to be snags . . .
John
-
Well done, John. You can't really beat actual printing to test these things. Very nicely written up summary, too.
-
It may be long but this has definitely been an illuminating thread. Mr. Smith, appreciate that last summary but need one clarification. Am I correct in assuming that the same upsampling rules would apply to printing from Photoshop? That may have been made clear earlier in the thread but I'm scared to go back looking for it ;-)
I should add I'm just about to upgrade my 4800 to a 9890 or 9900 and the regular cast of contributors here have provided lots of useful information across the board. I recently did my own print comparisons (http://www.velodramatic.com/archives/6995) between the 4800 and a 7900 and clearly I didn't understand the fine detail selection until I came here.
thanks all.
-
But The Epson guy agreed that sending the image data at the resolution the printer is expecting should take the print pipeline out of the equation.
Not to throw a monkey wrench into the equation but we may not actually be done yet regarding output resolution...
It seems the x900 series printers actually ship with the ability to report 1440 as their native resolution. But to get there is kinda tricky...
Here's the way to do it but note this ONLY works if you select one of the "Proofing Paper" media settings...(also note I've not yet tested this but I'm gonna)
Here's the info:
"On the PC side it is Level 3 Fine 1440x1440
Check Finest Detail in this mode and Qimage reports 1440 ppi MAX resolution (as a confirmation)
White Semimatte is usually the default LUT of choice if you are going to make a profile in this mode for a different media. It has the highest ink duty limit.
On the MAC side select be Fine - 1440 and Finest Detail".
So, I'll be testing this to see if one can eek out any more image detail this way...it'll be a couple of weeks before I can get into this though. But for the geeks this may be another rabbit hole to go down :~)
Note, it requires the x900 series so a 4900, 7900 or 9900 is required. There is no similar ability on any of Epson other printers...
-
So, I'll be testing this to see if one can eek out any more image detail this way...it'll be a couple of weeks before I can get into this though. But for the geeks this may be another rabbit hole to go down :~)
Note, it requires the x900 series so a 4900, 7900 or 9900 is required. There is no similar ability on any of Epson other printers...
In my testing, the proofing paper media presets do have better ink-loading for modern (read non-Epson) rag papers but conversely significantly reduced gamut. Not sure exactly why the latter would be. Since ABW isn't supported for these, the presets are of limited use ... unless I guess you're using them for proofing.
-
That's an interesting thought, Jeff. This mode was added really to support various RIP vendors for the proofing market (as would be obvious from the driver media mode you choose to access it).
How well this mode performs for general photographic work will be interesting to see because, as Jeff says, it's designed for highest ink density so in the driver it's using a larger dot size (which also makes it quite fast). This may counter any benefits from a higher input resolution and probably also accounts for the gamut decrease that Gromit notes.
-
It may be long but this has definitely been an illuminating thread. Mr. Smith, appreciate that last summary but need one clarification. Am I correct in assuming that the same upsampling rules would apply to printing from Photoshop?
I did no testing using Photoshop, and I just don't have the energy to do so (I never use PS, so it would be a steep learning curve for me). You really would have to run your own tests, but I would have thought that if you needed to downsample, PS would be a better choice.
John
-
In my testing, the proofing paper media presets do have better ink-loading for modern (read non-Epson) rag papers but conversely significantly reduced gamut. Not sure exactly why the latter would be. Since ABW isn't supported for these, the presets are of limited use ... unless I guess you're using them for proofing.
Are the proofing papers not for 90% using PK black? Do you mean the Fiber and Baryta versions of rag papers? Matte fine art rag papers are not really compatible with the proofing media presets.
Epson Proofing White Semimatte should deliver a very wide gamut, Epson advertises the gamut of its recent models based on prints with that paper. For example the 4900 set at 2880x1440 dpi should cover 98% of Pantone colors. It has a high white reflectance and little or no FBA content (in my measurements only two RC papers show that low FBA content). Lab: 97.2 0.3 1.3. It holds its white in time too if my several months old window test tells something. The only problem I have with it is that there is a lot of bronzing with the HP Z ink sets, it will be better with Epson and Canon ink sets.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I did no testing using Photoshop, and I just don't have the energy to do so (I never use PS, so it would be a steep learning curve for me). You really would have to run your own tests, but I would have thought that if you needed to downsample, PS would be a better choice.
John
John,
With your critical eye it would be nice if you spend some energy on comparing Qimage's on the fly extrapolation and sharpening with LR's qualities. I do not think that Photoshop will improve your images but Qimage might.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
• Strangely, the absolute worst thing you can do is to resample the image in LR to the same value as reported for the cell size – for example, take a 360 ppi image resolution and then set the LR output resolution to 360 as well. Theoretically, it should make no difference – but the quality of the print is degraded compared with simply sending it to the printer with no LR resampling. I can’t explain why this happens, but it is observable and repeatable. I can also see the same result with the Restest file – comparing output direct at 720 with the same file output with the LR Resolution box also set to 720 ppi (the sample text is degraded).
John
John, a nice summary and all about a normal image.
The quote above describes the same observation you had earlier with the test target. I then thought it was related to algorithms not suitable for text, vectors. That is not the case it seems . It makes me wonder whether intelligent sharpening, anti-aliasing etc is not activated in LR if it does not resample on the fly to the requested printer input, possibly related to a simultaneous resampling/sharpening/anti-aliasing process when it is done on the fly. If I understand the difference between the routes of your two examples correctly.
On anti-aliasing in downsampling, it is available in Qimage and comes with a slider to adjust the strength. And I have written before that downsampling quality will become more important with increasing MP in cameras. It has been good too when scanned film images have (aliased) grain aboard.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Ernst
I have downloaded the trial version of Qimage, but I am not going to have any time to test it out for a while, as I am going away next week. I just had a brief look at it, and so far I have to say I found the user interface a bit off-putting.
Doing tests like this certainly forces you to re-think a lot of assumptions, which is a very good thing. My first reply to Mike on this thread was basically correct on procedure, but backed up by the wrong science. Essentially, the reason that upsampling a file to 360 or 720 in LR produces a better print is probably absolutely nothing to do with some hypothetical matching of image resolution to printer resolution (as I implied), and everything to do with the fact that LR upsamples the image using some pretty neat anti-aliasing and smoothing routines. The results I have come up with now are strictly empirical (they work, but I’m not entirely sure why) and the hell with the theory.
Which leads me to another thought, and an incorrect assumption. I notice when running these tests that at 720 ppi size and output, even with an 8x loupe on the print I cannot see anything like the detail which is in the original file at 100% on screen. I still don’t see all this detail at 360 ppi (equivalent to a 20x15 ins print), and in fact it is only at 180 ppi (equivalent to a 40x30 ins print) that I can see everything on the print that is in the original file. Upsampled in LR to 720 ppi this prints as smooth as you like and is in fact the closest representation of the original image. Which means that, up to a certain point (which is probably about 180 ppi or so) the bigger you go with a digital image and an inkjet print, the better it gets. This is of course entirely contrary to darkroom printing, where the best possible quality is from a contact print from the negative, and anything larger than that progressively degrades the IQ.
Now I, in my innocence, had assumed that the best print quality would come from printing the file at exactly the same ppi as the highest output from the printer – so in my case, it would be a 720 ppi output file and 720 ppi from the printer, giving a print size of 10x7.5 ins. After all, one pixel in the file would print as one pixel on the page, so all of the image detail and information should be there. This would be my digital “contact print”. But no, it doesn’t work – in fact I don’t see anything like all of the image data until I get up to 180 ppi or so. I’m really not sure why this is (dither pattern obscuring detail? Inability of printer to lay down a continuous tone in highlight areas? Droplet size too large?). It seems that to do what I would like to do (print very high-quality small prints) I would need a printer with a much greater nozzle density on the print-heads. We could speculate – the current Epsons have 180 nozzles per head, so perhaps I would need 720 packed into the same space with probably a sub 1 picoliter droplet size to do the job.
I don’t think that we will be seeing such a printer anytime soon – think of the clogging issues for a start ;)
John
-
I have downloaded the trial version of Qimage, but I am not going to have any time to test it out for a while, as I am going away next week. I just had a brief look at it, and so far I have to say I found the user interface a bit off-putting.
Hi John,
That's a common first reaction. Not that I like every aspect of it, but it's actually very efficient for the plethora of options it offers. The best way to get acquainted is by following some of the learn by example scenarios or the videos. By mastering common and frequently recurring tasks, it becomes easier to learn some of the more infrequently used ones.
[...]
Which leads me to another thought, and an incorrect assumption. I notice when running these tests that at 720 ppi size and output, even with an 8x loupe on the print I cannot see anything like the detail which is in the original file at 100% on screen. I still don’t see all this detail at 360 ppi (equivalent to a 20x15 ins print), and in fact it is only at 180 ppi (equivalent to a 40x30 ins print) that I can see everything on the print that is in the original file. Upsampled in LR to 720 ppi this prints as smooth as you like and is in fact the closest representation of the original image. Which means that, up to a certain point (which is probably about 180 ppi or so) the bigger you go with a digital image and an inkjet print, the better it gets. This is of course entirely contrary to darkroom printing, where the best possible quality is from a contact print from the negative, and anything larger than that progressively degrades the IQ.
There are (at least?) two possible explanations.
The first is the limitations of human visual acuity. Normal resolution of the human eye is limited to something in the 300 PPI ballpark. This is also a function of contrast, lower contrast loses resolution sooner. You may ask yourself, why then do we use higher resolution? Well, there is also a thing called Vernier resolution, and our brain/eyes are very good at discriminating much higher levels of detail when slight offsets are involved.
The second reason has indeed to do with dithering. Dithering is used to blend colors in between the readily available ink colors. So for certain colors the resolution is sacrificed for accurate colors. However, for the detail that's close to one of the available ink colors resolution can be boosted at 720 PPI. The benefit of being able to do that at 720 PPI is that small sharpening artifacts are virtually invisible (below visual acuity) but lower spatial frequencies (say at 360 PPI and lower) will also be boosted (potentially without artifacts). It is therefore also possible to increase the amount of sharpenng at 720 PPI, if the printing application allows to tweak that. Hence my earlier remark about limited control in LR, whereas e.g. Qimage allows to adjust the amount of smart sharpening that's applied after resampling. It also allows to tweak for the specific output material used.
Of course, when you upsample enough (e.g. your 180 PPI input resolution) then every little bit of resolution available in the original file will be much easier to see, but also more risky to sharpen without creating visible artifacts. Again, having control over the amount is helpful. Also the type of sharpening will make a difference. Deconvolution sharpening is more likely to avoid halos such as we know them from edge contrast enhancing methods.
Cheers,
Bart
-
Thanks John,
I did in fact mean, uprez when I used the term upsample. Sorry if that confused the issue/question. Though my workflow is Lightroom centric I've stuck to using Photoshop for printing out of habit. In conjunction with the new printer I'll invest some time in understanding the Lightroom print module.
::Michael
-
Which means that, up to a certain point (which is probably about 180 ppi or so) the bigger you go with a digital image and an inkjet print, the better it gets. This is of course entirely contrary to darkroom printing, where the best possible quality is from a contact print from the negative, and anything larger than that progressively degrades the IQ.
That's something I would agree with, and I've heard too 200dpi as the maximum of MTF for inkjets - sounds plausible to me but I didn't see any hard proof btw.
Dithering would made an ideal suspect for that indeed, thanks Bart!
And thanks to all for this informative thread.
-
Ernst
I have downloaded the trial version of Qimage, but I am not going to have any time to test it out for a while, as I am going away next week. I just had a brief look at it, and so far I have to say I found the user interface a bit off-putting.
I’m really not sure why this is (dither pattern obscuring detail? Inability of printer to lay down a continuous tone in highlight areas? Droplet size too large?). It seems that to do what I would like to do (print very high-quality small prints) I would need a printer with a much greater nozzle density on the print-heads. We could speculate – the current Epsons have 180 nozzles per head, so perhaps I would need 720 packed into the same space with probably a sub 1 picoliter droplet size to do the job.
I don’t think that we will be seeing such a printer anytime soon – think of the clogging issues for a start ;)
John
John,
Qimage's user interface will never score a prize in a world that likes iGadgets and iApps. It will beat them on print functionality though.
I think you need more inkjet print area to get all the detail of an old contact print. And you must have very good eyes if you can see all the detail in a contact print.
Beyond the Epson R2400 B&W quality there are inkjet printers with 1.5 to 2 picoliter droplet sizes, 16 bit drivers, 4 to 7 monochrome B&W ink sets. Black Only inksets squirting 1.5 picoliter droplets only from 4 channel heads to improve Black Only printing. Check the DigitalBlackandWhiteThePrint forum at Yahoo to see what has been developed in time, Jon Cone's products and methods of course, Paul Roark's many experiments, Roy Harrington's QuadTone RIP, the Bowhaus boys, all of them and the interaction with forum members created a wealth of information. That forum has almost 10K of members, must be one of the largest lists on inkjet printing.
Clogging does not have to be an issue, Paul mixed HP Vivera PK ink with ink medium to quad sets for several Epsons and they behave better on the printer than the OEM inkset does. There are turn key solutions available with a short learning curve and they are as easy and convenient to use as the OEM solutions.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Are the proofing papers not for 90% using PK black? Do you mean the Fiber and Baryta versions of rag papers? Matte fine art rag papers are not really compatible with the proofing media presets.
Yes, you're right. It was a while ago that I did the tests so the comparison in gamut would have been against baryta papers.
-
With your critical eye it would be nice if you spend some energy on comparing Qimage's on the fly extrapolation and sharpening with LR's qualities. I do not think that Photoshop will improve your images but Qimage might.
I did tests comparing Lightroom 3, Qimage and Blow Up 2 (all at 720ppi) and got the best results to my eye with Blow Up 2. I encourage others to do similar tests and form their own conclusions.
-
So for certain colors the resolution is sacrificed for accurate colors. However, for the detail that's close to one of the available ink colors resolution can be boosted at 720 PPI. The benefit of being able to do that at 720 PPI is that small sharpening artifacts are virtually invisible (below visual acuity) but lower spatial frequencies (say at 360 PPI and lower) will also be boosted (potentially without artifacts). It is therefore also possible to increase the amount of sharpening at 720 PPI, if the printing application allows to tweak that.
Bart, that started me thinking. At first it seemed counter-intuitive to me that you could use more sharpening on a smaller print, but when you test it out it makes sense. Sharpening artefacts do of course become more visible the further that you enlarge the pixels in the file, and less so when the detail is in fact very small. So I ran another routine on the cups on the table from my test print of the VW Campers, sized to exactly 720 ppi which results in a print size of 10x7.5 ins (a small print in everybody’s view around here). I did this with the LR output sharpening set to off, set to Standard, and set to High. There is visibly enhanced detail in the pattern on the cups with Standard, and High gives the best result. Most importantly, I can see no obvious sharpening artefacts as a result – and I am fanatical about sharpening halos in particular, if I see them I will just tear the print up.
I then ran another test to see if altering my capture sharpening could improve things further. As I think you suggested, deconvolution sharpening could result in fewer artefacts, so I went back to the Develop Module and altered my sharpening to Radius 0.6, Detail 100, and Amount 38 (my original settings were Radius 0.9, Detail 35, Amount 55). The next print gained a little more acutance as a result with output sharpening still set to High, with some fine lines on the cup patterns now becoming visible under the loupe. Just for fun, I am going to attach 1200 ppi scans of the prints so you can judge for yourselves, bearing in mind that this is a very tiny section of the finished print.
And I think that has now about hit the limit for my “small” prints at 720 ppi via Lightroom to my 2400. The real gain with inkjet printing seems to come from the freedom we get to print very large, and actually gaining quality as we do so with big MF files. I have learned an interesting lesson, though, which is that you can really ramp up the sharpening on a small print from a large file.
John
-
I then ran another test to see if altering my capture sharpening could improve things further. As I think you suggested, deconvolution sharpening could result in fewer artefacts, so I went back to the Develop Module and altered my sharpening to Radius 0.6, Detail 100, and Amount 38. The next print gained a little more acutance as a result with output sharpening still set to High, with some fine lines on the cup patterns now becoming visible under the loupe.
The output sharpening in LR 3 is designed to work optimally when the capture sharpening in Develop's Detail panels is set optimally. If the Detail settings are off (or not set optimally) then the output sharpener in Print will be off.
The key really is to nail the capture sharpening so the Amount, Radius, Detail and Masking as well as Luminance noise reduction (I consider luminance noise reduction to be the 5th sharpening slider) are all set so the image looks good at 1:1. I'm often in the mid 40's with Amount and occasionally higher and similar Radius and Detail (plus some Edge Masking) with medium format digital captures from by P65+. YMMV...
-
If the Detail settings are off (or not set optimally) then the output sharpener in Print will be off.
That's interesting, Jeff, I didn't know that. I didn't mean to give the impression that I had no capture sharpening set at the outset - I have now modified my post to include those settings.
The key really is to nail the capture sharpening so the Amount, Radius, Detail and Masking as well as Luminance noise reduction (I consider luminance noise reduction to be the 5th sharpening slider) are all set so the image looks good at 1:1. I'm often in the mid 40's with Amount and occasionally higher and similar Radius and Detail (plus some Edge Masking) with medium format digital captures from by P65+. YMMV...
I'd certainly coincide with all of that advice - and I do use use some edge masking to avoid sharpening clear skies and similar areas of even tone.
John
-
That's interesting, Jeff, I didn't know that.
When I said the output sharpening would be off I meant incorrect (not optimal), not actually turned off.
-
When I said the output sharpening would be off I meant incorrect (not optimal), not actually turned off.
I see, yes, off as in f*%^-up, rather than switched off.
J
-
Speaking of output sharpening, is there a definition of "Low, Standard" and "High" sharpening in LR? Is this something that can be measured or calculated in such a way as to know what setting to use without actually making a test print first?
-
Speaking of output sharpening, is there a definition of "Low, Standard" and "High" sharpening in LR? Is this something that can be measured or calculated in such a way as to know what setting to use without actually making a test print first?
Well, JP Caponigro pretty much thought the PK output sharpener over-sharpened while Mac Holbert of Nash editions thought it under-sharpened...so you might say that the the the Low setting was for JP and the High setting was for Mac...in truth the intent was to only provide 1 single setting but Adobe decided to offer 3.
So, if you DON'T optimize the capture settings in Develop, try High (since that tries to overcome under sharpening) and if you tend to over-sharpen in Develop try Low...if you know what you are doing (optimizing the capture sharpening when at 1:) use Standard...And, if you think I'm kidding, I'm really not...that's the way the discussion took place...
-
Well Jeff.. you have a way of putting things that makes sense. I usually use Low since I hate oversharpening, but I'll try a few at Standard. I've got a pretty good handle on capture sharpening so the Standard may provide better results overall. Thanks for clearing that up.
-
I then ran another test to see if altering my capture sharpening could improve things further. As I think you suggested, deconvolution sharpening could result in fewer artefacts, so I went back to the Develop Module and altered my sharpening to Radius 0.6, Detail 100, and Amount 38 (my original settings were Radius 0.9, Detail 35, Amount 55). The next print gained a little more acutance as a result with output sharpening still set to High, with some fine lines on the cup patterns now becoming visible under the loupe. Just for fun, I am going to attach 1200 ppi scans of the prints so you can judge for yourselves, bearing in mind that this is a very tiny section of the finished print.
John
John,
This is not the forum to discuss scanning and sharpening but I am intrigued by some aspects/contradictions of deconvolution sharpening, flatbed scanning and film grain. I have seen a thread on another LL forum (you were there too) that discussed deconvolution sharpening but little information on flatbed scanners and film grain. I have a suspicion that on flatbeds diffraction plays an important role in loss of sharpness (engineers deliberately using a small stop for several reasons) while at the same time the oversampling as used on most Umax and Epson models keeps (aliased) grain in the scan low and delivers an acceptable dynamic range. In another thread Bart mentioned the use of a slanted edge target on a flatbed to deliver a suitable base for the sharpening. I would be interested in an optimal deconvolation sharpening route for an Epson V700 while still keeping grain/noise at bay. Noise too as I use that scanner also for reflective scans.
The old thread was
http://www.luminous-landscape.com/forum/index.php?topic=45038.0
I could add this reply there too if the thread is still open.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Ernst
I have never done any serious film scanning (I’ve always sent mine out to a lab), so I have no experience of those problems (perhaps it’s just as well).
There is a (possibly insignificant) update from Smith Labs, however, as the latest round of testing reaches a conclusion. No scans today though, I just don’t have time.
* Testing shows that even at a small print size where LR reports a print resolution of 600 ppi, there is a definite gain to be had by resampling in LR to 720 ppi, and there is a visible and worthwhile improvement in fine detail in the print.
* This may be a contentious result, but never mind. My tests seem to show that it is best to avoid print size resolutions with odd numbers like 351, 577 ppi etc. For example, 600 ppi resampled to 720 looks better under a loupe than 601 ppi resampled to 720. Subtle tonal transitions in the original file seem to be rendered more smoothly. However, it may be that my eyesight is just giving up with so much testing and too many late nights ;)
PS In other words, if you have set up (like I did) for a printed image size of 12x9 ins and the resolution reported by LR is 601 or 603 or 597 or something, resize the cell very slightly to get 600 ppi instead, upsample to 720 and you will get a better print for a fractional difference in printed image size.
If you make use of the User Templates within the Lightroom Print Module, you can get all your favourite print and paper sizes set up with ideal resolutions (in and out) plus your profiles and everything else. It takes a while to sort it all out, but once it’s done you never have to worry about it again.
John
-
This thread has been a really interesting read.
Are there any limitations in file size for large prints? Would it be possible to send a 64x80" 720 ppi print file to an 11880?
I'm thinking large format scans or stitches where it's possible to have more than 360 ppi detail at this size.
-Dominique
-
This thread has been a really interesting read.
Are there any limitations in file size for large prints? Would it be possible to send a 64x80" 720 ppi print file to an 11880?
I'm thinking large format scans or stitches where it's possible to have more than 360 ppi detail at this size.
-Dominique
Usually that hits on some memory problems. Right now Qimage may handle up to 2 GB image data (64 bits Windows and ample memory) and with the best upsampling + sharpening routines to say 720 PPI at the size you intend the print data becomes huge and that is causing the problems in the end. The easy compromise will be 360 PPI input on the printer. Other routes with RIPs etc may be able to offload more printer data to a 60" printer but their upsampling routines are no improvement. With Qimage there is the Poster Tiles method that splits up the image in more parts, stitched again on the printer so to speak and that can help in this case. It is used for printing longer lengths than the driver allows but has been used to overcome this limitation too. There was another trick for getting image data above the 2 GB limit but I can not recall that. I guess viewing distance is not something that will stop you in adding detail so I did not start with that comment :-)
Edit: to overcome the 2 GB limitation an image could be chopped up in Photoshop and stacked/stitched again by nesting on Qimage's print page and then printed with the Poster Tiles method. That was in a thread some years ago and it never became clear whether that first part ever worked, the Poster Tiles method works for Epsons and Canons though, I have used it.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
This thread has been a really interesting read.
Are there any limitations in file size for large prints? Would it be possible to send a 64x80" 720 ppi print file to an 11880?
I'm thinking large format scans or stitches where it's possible to have more than 360 ppi detail at this size.
Hi Dominique,
Technically there should (!) be no problem sending a 2.7 GP file to the printer driver and spoolfile, assuming they manage memory as they should. However, in 3-channel RGB output that would be almost 8 GB of data, certainly more that e.g. a TIFF file format can store. On a 32-bit OS it will cause problems due to memory management, but on a 64-bit OS it should be 'fine', but there is a chance that drivers or spoolers choke on the amount of data, or the intermediate file size. There are some settings and precautions that can reduce the risk of OS memory problems, as outlined here (http://ddisoftware.com/tech/articles/february-2006-wide-load-tips-for-printing-large/).
If all the software components play together nicely, it will take a while (=understatement) to print but the resulting image should look very good, and you'll be able to use all the detail that's in the original file when the "Finest Detail" is selected, even up close. Of course you do want to make sure that the heads are well aligned for the medium you are going to use.
Another issue might be, how are you going to mount and transport the beast?
Cheers,
Bart
-
But the highest possible resolution per inch is most relevant for really small prints, where there will be sufficient detail from the camera, and viewers are likely to view it up close, isnt it?
-h
-
Viewing distance in relation to print size is the key to this whole topic.
Roy Lictenstein, and long before him Georges Surat had the answer to pixel peepers.
Stand back and view- the pixels (dots), blend and dissappear.
http://www.bing.com/images/search?q=georges+seurat&id=D6660BEE75DFE08B586F9F4C6DFBE25AA0D2188E&FORM=IGRE2#x0y0
http://www.bing.com/images/search?q=roy+lichtenstein+images&qs=IM&sk=IM1&pq=roy+lichtenstein+&sp=2&sc=8-17&form=QBIR#x0y4246
-
But the highest possible resolution per inch is most relevant for really small prints, where there will be sufficient detail from the camera, and viewers are likely to view it up close, isnt it?
Well, yes when talking about normal people. the intended viewing distance is considered about 2x the diagonal of the image...but that's for " normal" people...as Bruce Fraser has said, the normal viewing distance for a photographer is only limited by the length of his nose...
Any way you cut it, more is usually better...and if you "can" make something better, why wouldn't you? (and yes, there are times that time prevents you).
-
Are there any limitations in file size for large prints? Would it be possible to send a 64x80" 720 ppi print file to an 11880?
I'm thinking large format scans or stitches where it's possible to have more than 360 ppi detail at this size.
[/color]
I've encountered some limitations with Lightroom 3, where a file that was 44" x 102" @ 720 ppi simply didn't print (Lightroom produced a blank print from my Epson 9900) - in this case, I downsampled until I reached a resolution where it would print (in my case I eventually downrezed to 180 dpi). Your mileage may vary. For what it's worth, I wasn't memory or disk space limited. This is just one experience though - I haven't taken the time to be scientific about it, so there may have been another factor at work.
-
Well, yes when talking about normal people. the intended viewing distance is considered about 2x the diagonal of the image...but that's for " normal" people...as Bruce Fraser has said, the normal viewing distance for a photographer is only limited by the length of his nose...
Any way you cut it, more is usually better...and if you "can" make something better, why wouldn't you? (and yes, there are times that time prevents you).
It is like that in the graphic arts too. Detail in a mezzotint, aquatint, washed litho, even the texture of paper and ink in a crude woodblock print, it all counts to give a total impression of a print. It will not be different when landscapes, cityscapes, are displayed on digital picture frames of 10 by 5 feet, you will step closer when detail like a window shows something interesting and when you step back again the picture has changed in your head. Pictures made with wide angles or teles, we never stand at the right nodal point or whatever it is called to get the perspective of the original scene, if the lens perspective ever matches our eyes perspective. Sometimes that is done on purpose, think Bill Brandt. In the end a print is an abstraction of the original scene. Viewing distance is just one factor. Bruce Fraser was right.
met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Thanks for the input to both of you, Ernst and Bart. I was asking out of sheer curiosity.
So far my largest print has been 56x70" at 304.8 ppi on a Lightjet from a scanned 8x10" color neg. The master file is about 28k x 35k pixels. The downsized print file retained pretty much all the detail. But it's really tight and starts to look digital when some details are reduced to single pixel lines where the the full res file has some room to ramp up the contrast on such lines. There is a natural beauty in oversampling. On the other hand the Lightjet couldn't make use of all the detail of the print file. Some single pixel lines clearly visible on screen, in this case a slight contrast of lighter green lines against a darker green background, didn't make it into the print. It seems the Lightjet needs some threshold contrast at its native pixel resolution. I take it from John's comment about having to reduce the resolution to 180 ppi in order to get all the detail of the file on paper that an inkjet wouldn't be any better when it has to put down high res detail of low to moderate contrast but within similar hue.
Another issue might be, how are you going to mount and transport the beast?
I'm outsourcing my prints. The mounting is done by the lab. There are shippers who can manage the transport of this size.
-Dominique
-
I have read this thread with great intrest! It has been a great learning curve and confirmed a few 'rumours' regarding upsampling etc. Thanks to all who have contributed to this thread!
I do have a question, and if this has been answered, my apologies for asking again.
All this upsampling etc, seems to based on the final print size then upsampled from there in LR3. That suggests that the end user has control over the orginal image in some form. What if for example, I am given a file that is 8x10 at 360ppi ( from within camera say) but the customer wants it printed to a 16x20? So both the image dimensions and therefore possibly ppi change. Do I change the file dimension in Photoshop leaving the ppi at 360 or resize it and allow the ppi to drop leaving the pixels the same or, from LR set the page to 16x20 and leave the print output at 360ppi? My gut feeling is the later.
Your thoughts would appreciated.
-
Well, yes when talking about normal people. the intended viewing distance is considered about 2x the diagonal of the image...but that's for " normal" people...as Bruce Fraser has said, the normal viewing distance for a photographer is only limited by the length of his nose...
I've noticed something interesting about viewing distance, which is that when I watch people looking at the prints on my walls, they are surprisingly consistent in how close they come to each of the prints.
That "viewing distance" formula is supposed to include the angular view represented in the print, and in theory at least, is the distance at which the apparent perspective (based on depth cues rather than stereoscopic vision) matches what you would have seen at the point at which the photograph was taken. The theory is that our brains like it when the visual clues we use to gauge distance mesh up when they decode a 2D projection of a 3D scene - which they should do at that calculated viewing distance. Other perspectives are of course used for dramatic effect, but I'm wondering if people naturally tend to stand more or less where the perspective is right.
I've really noticed this on a particularly large print of mine, which is a stitched image where the angular view corresponds to what a very wide angle lens would have recorded. People tend to get quite close to this image. If I look at the image, and imagine being at the scene in question, I find that something clicks when I'm surprisingly close. This matches up with the viewing distance formula (the calculated viewing distance is shorter for wide angles of view, and longer for narrow angular angles of view).
Does this match up with other people's experience? Has anyone else tried observing where people tend to stand when they look at particular prints? Come to think of it, this might be a good time for me to see what's been published on the subject...
-
I've noticed something interesting about viewing distance, which is that when I watch people looking at the prints on my walls, they are surprisingly consistent in how close they come to each of the prints.
That "viewing distance" formula is supposed to include the angular view represented in the print, and in theory at least, is the distance at which the apparent perspective (based on depth cues rather than stereoscopic vision) matches what you would have seen at the point at which the photograph was taken. The theory is that our brains like it when the visual clues we use to gauge distance mesh up when they decode a 2D projection of a 3D scene - which they should do at that calculated viewing distance. Other perspectives are of course used for dramatic effect, but I'm wondering if people naturally tend to stand more or less where the perspective is right.
Hi John,
The anamorphic projection distortion does have an effect on how we perceive the image, e.g. wide angle effect when viewed from too far or a compressed telephoto effect when we're watching from too close. It also explains why people feel there is something wrong with the 17mm T/S shots of high buildings on a 35mm camera, while they are perfectly undistorted by the lens.
Whether that initial effect is enough to make us adjust the viewing distance, I'm not so sure. In general we want to be able and view a scene as composed by the photographer in 'one view', just scanning the image with our eyes (not by also turning our head). That usually results in a viewing distance that's roughly equivalent to the diagonal dimension of the image.
I think that the main driver for close inspection is curiosity. When we see that there is enough detail to inspect, then we tend to get closer, and see if there is more detail when viewed close up.
If anything, it warrants getting the most out of our prints ...
Cheers,
Bart
-
I really wonder whether we fit our viewing position to the original perspective all the time and whether it itches when we can not. I would expect it for larger prints and image content that shows enough depth information and the viewer able to find the right place but with smaller prints it becomes quite impractical, lighting etc blocked by the viewer's presence. In the past photos were generally smaller on whatever image carrier. Larger prints made of telephoto lens images would make it difficult too. I also wonder how image composition and that natural perspective should relate then. It could become a complex research and considering more abstraction in a B&W print versus color there should be testing done for both categories. Magritte's -Ceci n'est pas une pipe- goes beyond painting.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
Hi All
I would still like some input on the below:
All this upsampling etc, seems to based on the final print size then upsampled from there in LR3. That suggests that the end user has control over the orginal image in some form. What if for example, I am given a file that is 8x10 at 360ppi ( from within camera say) but the customer wants it printed to a 16x20? So both the image dimensions and therefore possibly ppi change. Do I change the file dimension in Photoshop leaving the ppi at 360 or resize it and allow the ppi to drop leaving the pixels the same or, from LR set the page to 16x20 and leave the print output at 360ppi? My gut feeling is the later.
Thanks
-
Hi All
I would still like some input on the below:
All this upsampling etc, seems to based on the final print size then upsampled from there in LR3. That suggests that the end user has control over the orginal image in some form. What if for example, I am given a file that is 8x10 at 360ppi ( from within camera say) but the customer wants it printed to a 16x20? So both the image dimensions and therefore possibly ppi change. Do I change the file dimension in Photoshop leaving the ppi at 360 or resize it and allow the ppi to drop leaving the pixels the same or, from LR set the page to 16x20 and leave the print output at 360ppi? My gut feeling is the later.
Thanks
My gut feeling is the same and I would let Qimage resample the 180 PPI on the fly to what the printer requests, in my case 300 PPI for an HP printer. I think that is not in conflict with the conclusions in this thread.
met vriendelijke groeten, Ernst
Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html
-
Thanks for the reply. Much appreciated!
-
I'm wondering if I'm doing something wrong here...just did two test prints of leaves and grasses from a Leica M9 shot. Printed on my 7900 with the correct profile settings in the epson driver. am printing through Lightroom. For the size print I printed (letter sized paper) the dpi resolution was 500 dpi something (can't remember exactly)..anyway, printed one print at 360 dpi through LR and another print at 720 dpi through LR. Looking at both prints with my 7x loupe, the 360 dpi print is clearly sharper and has better micro contrast. The 720 dpi print detail in leaves and grasses looks lower contrast and is actually a bit mushy. Wondering what gives here? What could I be doing wrong..if anything?? thanks!, Eleanor
No, not really...here's a simpler guideline for Epson printers; if your image's native size is less than 360PPI, set the resolution in LR to 360 for the output resolution, if the image's native size is above 360 but less than 720PPI, set the resolution to 720PPI in LR.
For Canon and HP the numbers are 300/600PPI...
-
I'm wondering if I'm doing something wrong here...just did two test prints of leaves and grasses from a Leica M9 shot. Printed on my 7900 with the correct profile settings in the epson driver. am printing through Lightroom. For the size print I printed (letter sized paper) the dpi resolution was 500 dpi something (can't remember exactly)..anyway, printed one print at 360 dpi through LR and another print at 720 dpi through LR. Looking at both prints with my 7x loupe, the 360 dpi print is clearly sharper and has better micro contrast. The 720 dpi print detail in leaves and grasses looks lower contrast and is actually a bit mushy. Wondering what gives here? What could I be doing wrong..if anything?? thanks!, Eleanor
Eleanor
Your experience is completely contrary to my own test results and observations. In one case you are downsampling from ca 500 ppi to 360, in the other upsampling to 720. If the downsampled print to 360 looks sharper, then I am at a loss to suggest why. Except that -
Are you sure that what you are seeing really is sharpness and better detail, rather than edge artefacting which can make the print look a little crunchy and crisp but is not really detail at all? Often a correctly upsampled file with edge smoothing and very moderate output sharpening applied will look softer under a loupe, but actually is a more faithful representation of the original file. With my own work I aim for a “look” which approaches that of my darkroom prints – all the detail is there, but without any obvious digital edge artefacting (or haloes, of course).
Here is another Smith Labs idea. If the ppi reported in LR at your output size is 500 ish, adjust the cell size until you have an exact multiple of 60 ppi – in this case 480 or 540 ppi. Then upsample that to 720 ppi and compare.
John
-
I'm wondering if I'm doing something wrong here...just did two test prints of leaves and grasses from a Leica M9 shot. Printed on my 7900 with the correct profile settings in the epson driver. am printing through Lightroom. For the size print I printed (letter sized paper) the dpi resolution was 500 dpi something (can't remember exactly)..anyway, printed one print at 360 dpi through LR and another print at 720 dpi through LR. Looking at both prints with my 7x loupe, the 360 dpi print is clearly sharper and has better micro contrast. The 720 dpi print detail in leaves and grasses looks lower contrast and is actually a bit mushy. Wondering what gives here? What could I be doing wrong..if anything?? thanks!, Eleanor
Eleanor...
Did you change the print driver to "Finest Detail"? If not, the driver will take the LR 720 dpi output and then "downsize it" to 360 dpi. Effectively resulting in manipulations which will damage quality.
John
-
Many thanks to both Johns! I will do more testing today and report back... I may not have had finest detail checked. Eleanor
-
Eleanor...
Did you change the print driver to "Finest Detail"? If not, the driver will take the LR 720 dpi output and then "downsize it" to 360 dpi. Effectively resulting in manipulations which will damage quality.
John
That was on my mind too but I am less familiar with LR. If LR first upsamples (with good routines) to 720 PPI and the driver still asks for 360 PPI then aliasing etc could happen in the driver's downsampling. While the LR downsampling when everything is kept at 360 PPI should have the correct algorithms. In Qimage the (maximum) upsampling is always done to the driver's requested rendering resolution and not more. Something I would expect of LR too.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
Yes yes!! with 720 up res. one needs to check finest detail!!! Makes all the difference in the world. (I hadn't checked it before in my test print.... Thanks!!! eleanor
-
That was on my mind too but I am less familiar with LR. If LR first upsamples (with good routines) to 720 PPI and the driver still asks for 360 PPI then aliasing etc could happen in the driver's downsampling. While the LR downsampling when everything is kept at 360 PPI should have the correct algorithms. In Qimage the (maximum) upsampling is always done to the driver's requested rendering resolution and not more. Something I would expect of LR too.
met vriendelijke groeten, Ernst
New: Spectral plots of +250 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
Ernst...
LR does not query the driver. You can set it to any dpi you want...let the user beware :-)
I guess that is why we print with Qimage. Protects us from ourselves....while delivery the best quality prints...+ other great features.
John