Luminous Landscape Forum
Raw & Post Processing, Printing => Printing: Printers, Papers and Inks => Topic started by: Michael H. Cothran on March 01, 2010, 09:03:27 am
-
In the LL video tutorial Camera to Print, Jeff Schewe (corroborated by Michael Reichmann and previously by Bruce Fraser) advocates no manual interpolation when the native resolution falls between 180-480 ppi. Since viewing the video, I began practicing this method.
However (and there's always a "however" or a "but..."), I was unclear as to what would be the better treatment of files when they fall BELOW or ABOVE the 180-480ppi parameters. Obviously, up-/down-rezing is then necessary, but I see more than one way of doing this, and would appreciate your opinion as to which way would be preferred, if it makes any difference at all:
Method 1. If the native resolution if above or below 180-480 ppi, rez the file to 360 ppi, and send that to the (Epson) printer.
Method 2. If the native resolution is below 180 ppi, uprez to 180 ppi, then send that to the (Epson) printer, allowing it to do the final interpolation to 360 ppi. Conversely, if the native resolution is above 480 ppi, downrez to 480 ppi, and again, send that to the printer, allowing it to perform the final interpolation to 360 ppi.
Method 3. As an alternative to method 2, I could down rez directly to 360 ppi, while uprezing only to 180 ppi. This would help speed up the spooling process.
Any thoughts would be appreciated. Jeff or Michael - if you happen to be reading this inquiry, I'd certainly appreciate your opinion.
Thanks,
Michael H. Cothran
Nashville, TN
-
Micheal, I would also appreciate an answer to your question. Like you, I came away from the LL printing tutorial not understanding this issue very well.
-
Any thoughts would be appreciated. Jeff or Michael - if you happen to be reading this inquiry, I'd certainly appreciate your opinion.
Thanks,
Michael H. Cothran
Nashville, TN
I too await Micheal's and Jeff's reply. Bruce Fraser's approach (updated by Jeff) is succinctly outlined on pp. 82 and 83 of Image Sharpening, 2nd Ed.
There is some controversy in this area and you can read a paper by Qimage (http://www.ddisoftware.com/qimage/quality/) to get one viewpoint and you can do your own testing with the images posted in that paper. I downloaded the images in that article, and as Bruce predicted, I could see some advantage at 720 ppi with the line chart but little difference with my printer (an old Epson 2200) between 360 and 720 ppi with a normal photographic image. At odd resolutions, one can resample to the native resolution of the printer or let the printer driver do the job. The choice would depend on the relative merits of the downsizing algorithms in one's imaging software and printer driver.
As camera resolutions increase, downsizing is often needed, and ideally the image should probably be prefiltered with a low pass filter to avoid aliasing of the high frequencies as discussed in a recent thread (http://luminous-landscape.com/forum/index.php?s=&showtopic=41589&view=findpost&p=347184). Photoshop's bicubic sharper algorithm was criticized in that thread.
BTW, I have fond memories of Nashville. My daughter just graduated from Vanderbilt, and I really enjoyed my visits there.
Regards,
Bill
-
Epson heads
F8 print head = 180PPI (38x0, 78x0, 98x0, 4xx0 etc)
TFP print head = 360PPI (7900, WT7900, 9900, 11880)
If file is <180 up to 180PPI.
If file is <240 up to 240PPI.
If file is < 360 up to 360PPI.
The aim is in hitting the size you want while also hitting the correct res for your printer.
-
I too have posted a question about this 2 weeks ago, as the "From camera to Print" is indeed in need of additional clarification.
see here:
http://luminous-landscape.com/forum/index....showtopic=41797 (http://luminous-landscape.com/forum/index.php?showtopic=41797)
-
Epson heads
F8 print head = 180PPI (38x0, 78x0, 98x0, 4xx0 etc)
TFP print head = 360PPI (7900, WT7900, 9900, 11880)
Can you tell me where this information comes from? Eric Chan states that the native resolution of the 3800 is 360 with "finest detail" unchecked, and 720 with "finest detail" checked.
In any event, to respond to the OP, the way to get the answer to this question -- as it specifically relates to your situation -- is to take a little time to do some empirical testing.
In my case, I print 12MP files to 17x22 paper, so my base resolution with minimal or no cropping is a bit under 180 ppi (usually around 176, as I recall). So I did some tests printing crops of a typical file to the scale the full image would be on a 17x22 print after doing my usual post-processing workflow in ACR with futher refinement as needed in PS, trying three methods: uprezzing to 360 ppi with bicubic smoother and printing from PS applying output sharpening I thought appropriate; uprezzing to 360 ppi with Genuine Fractals, applying output sharpening I thought appropriate in PS and printing from PS; and printing from the corrected but original size file from Lightroom and letting LR/printer driver do everything. In other words, I was doing the workflow exactly like I would if I were going to a final real print. I found I liked the look of the print uprezzed with GF the best (though it really took an examination from a foot away to really see a difference compared to the Bicubic Smoother file, and I liked both of them better than printing from LR and letting the printer driver do the uprezzing.
So my conclusion is that, for files from MY camera/lens, shooting MY typical subject matter, usign MY typical post-processing adjustments, printing on MY printer at MY preferred size on MY preferred paper, the method that gives me the look I like best is to uprez to 360 ppi using GF. I don't know why that is, or what I am doing or not doing in the three methods to get that result. That's what works best for ME working the way I work. What works best for YOU may or may not be something completely different. But the only way to find out what does work best for you is to take a little time to test it out.
-
Hopefully Jeff chimes in here, but I believe some of the confusion lies in the fact that Camera to Print predates Lightroom 2 and especially LR 3. In the video Jeff talks about leaving the file in it's native size unless it gets below 180dpi or above 480 dpi. I believe he also mentions that you are better off uprezzing in an even amount (200%) which utilizes Photoshop's bicubic algorithms bettter rather than uprezzing to a specific DPI to fit your printers output. The premise I believe is that the printer driver will do a better job resizing the iamge as part of the screening process. But this is all based on printing from Photoshop. From my perspective this workflow is pretty nice because for most images you only need one master file for all your print sizes, and I believe the quality is quite acceptable if you print this way for most images.
However I believe more recently Jeff advocates printing from Lightroom's print module which has been highly optimized to produce the best results using technology from Pixel Genius and as such talks about a little different philosophy. I'm pretty sure he's mentioned this in at least one thread on this forum so a little effort with searching should turn it up.
One note as far as using the setting finest detail, this setting is not designed for use with photographic work, but specifically designed for vector graphics. You may find some issues in edges in your photographs if you use this setting. For the highest quality photographic output, I would use 2880 printing for an Epson printer, but not finest detail.
-
However I believe more recently Jeff advocates printing from Lightroom's print module which has been highly optimized to produce the best results using technology from Pixel Genius and as such talks about a little different philosophy. I'm pretty sure he's mentioned this in at least one thread on this forum so a little effort with searching should turn it up.
So, as it relates to Lightroom 2.3 and above, Eric Chan "fixed" the resampling to the point where if your image is on the low end of the 180-480ppi scale, in Lightroom you COULD see a benefit with "certain images" if they had strong contrast diagonal or circles and/or the images contained a lot of high frequency detail.
In those cases, with Lightroom's adaptive upsample you CAN get better results printing from Lightroom by upsampling about a 50% amount-so if at 180ppi, add 90 to get to say, 270ppi.
As far as automatically upsampling to 360ppi, not sure that is warranted...also not sure going to 480ppi (the current max in Lightroom 2.x) is warranted...my suggestion is to test it yourself with your own printer and paper...none of this applies directly to Photoshop however since Photoshop does not have an adaptive upsample nor the PG output sharpening...
As far as the x800 series being in effect 180 nozzles per inch and the x900 series 360 nozzles per inch, that's correct...but just understand that relates to how many droplets per inch can be printed...that is not at all the same as dots per inch (DPI).
Ink jet printer resolution is a lot more complicated and Epson's newest printers are using and all new exotic math to create droplets that is different on the x900 printers...
As far as "future developments" in either software (or videos) I really can't comment other that to say, yes, in the near future, all this stuff will change...
-
In the LL video tutorial Camera to Print, Jeff Schewe (corroborated by Michael Reichmann and previously by Bruce Fraser) advocates no manual interpolation when the native resolution falls between 180-480 ppi. Since viewing the video, I began practicing this method.
However (and there's always a "however" or a "but..."), I was unclear as to what would be the better treatment of files when they fall BELOW or ABOVE the 180-480ppi parameters. Obviously, up-/down-rezing is then necessary, but I see more than one way of doing this, and would appreciate your opinion as to which way would be preferred, if it makes any difference at all:
Method 1. If the native resolution if above or below 180-480 ppi, rez the file to 360 ppi, and send that to the (Epson) printer.
Method 2. If the native resolution is below 180 ppi, uprez to 180 ppi, then send that to the (Epson) printer, allowing it to do the final interpolation to 360 ppi. Conversely, if the native resolution is above 480 ppi, downrez to 480 ppi, and again, send that to the printer, allowing it to perform the final interpolation to 360 ppi.
Method 3. As an alternative to method 2, I could down rez directly to 360 ppi, while uprezing only to 180 ppi. This would help speed up the spooling process.
Any thoughts would be appreciated. Jeff or Michael - if you happen to be reading this inquiry, I'd certainly appreciate your opinion.
Thanks,
Michael H. Cothran
Nashville, TN
Or just use Qimage today and select the right extrapolation algorithm for the image content, the 5 degrees of upsampling, the anti-aliasing slider on downsampling and the smart print sharpening slider. All done on the fly at print processing time. Enough ways in Qimage to make a small proof print when in doubt.
met vriendelijke groeten, Ernst Dinkla
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/ (http://groups.yahoo.com/group/Wide_Inkjet_Printers/)
-
Can you tell me where this information comes from? Eric Chan states that the native resolution of the 3800 is 360 with "finest detail" unchecked, and 720 with "finest detail" checked.
In any event, to respond to the OP, the way to get the answer to this question -- as it specifically relates to your situation -- is to take a little time to do some empirical testing.
In my case, I print 12MP files to 17x22 paper, so my base resolution with minimal or no cropping is a bit under 180 ppi (usually around 176, as I recall). So I did some tests printing crops of a typical file to the scale the full image would be on a 17x22 print after doing my usual post-processing workflow in ACR with futher refinement as needed in PS, trying three methods: uprezzing to 360 ppi with bicubic smoother and printing from PS applying output sharpening I thought appropriate; uprezzing to 360 ppi with Genuine Fractals, applying output sharpening I thought appropriate in PS and printing from PS; and printing from the corrected but original size file from Lightroom and letting LR/printer driver do everything. In other words, I was doing the workflow exactly like I would if I were going to a final real print. I found I liked the look of the print uprezzed with GF the best (though it really took an examination from a foot away to really see a difference compared to the Bicubic Smoother file, and I liked both of them better than printing from LR and letting the printer driver do the uprezzing.
So my conclusion is that, for files from MY camera/lens, shooting MY typical subject matter, usign MY typical post-processing adjustments, printing on MY printer at MY preferred size on MY preferred paper, the method that gives me the look I like best is to uprez to 360 ppi using GF. I don't know why that is, or what I am doing or not doing in the three methods to get that result. That's what works best for ME working the way I work. What works best for YOU may or may not be something completely different. But the only way to find out what does work best for you is to take a little time to test it out.
Hi,
Eric's talking about the driver screen, i'm talking about the mechanical rez of the heads which the screen is built on.
-
Epson heads
F8 print head = 180PPI (38x0, 78x0, 98x0, 4xx0 etc)
TFP print head = 360PPI (7900, WT7900, 9900, 11880)
You seem to suggest that the resolution is limited to 360PPI. If so, then how would you explain that a resolution difference in output (upto 720PPI) can be demonstrated (http://www.ddisoftware.com/qimage/quality/)?
The printer driver also feeds back info to the calling program that the native resolution is 720PPI with glossy paper. This allows to apply sharpening at the interpolated data at 720PPI, which allows to apply more sharpening without visible artifacts.
Cheers,
Bart
-
Thanks for the responses so far, but either above my head, or not applicable.
Since watching the C to P video, I now know how to treat native resolutions in the 180-480 range.
I just wanted to know what PPI to set in PS CS3 if my native resolution is, say, 150 ppi or 800 ppi for example?
There was the suggestion to use Qimage, which I understand is a Windows-only application. I'm Mac-based. Also, I don't use LR, but ACR 4.6, and print from PS CS3. Since I'm still using software that was current about the time Jeff & Michael made the Camera to Print video, I'd like to continue following their advice at the time.
Was hoping for a quick & easy answer.
Michael H. Cothran
Nashville, Tennessee
-
Was hoping for a quick & easy answer.
If you are under 180ppi, either upsample or make the image size smaller...if you are over 480ppi, I would simply print with the higher rez. I would also use PhotoKit Sharpener's output sharpening after you set the final PPI of the file.
I would also suggest looking at Lightroom if for no other reason that it makes printing a lot, lot easier (and the fact you don't have to spawn off image versions for each and every print size you do).
-
Was hoping for a quick & easy answer.
Michael H. Cothran
Nashville, Tennessee
I can tell you what I used to do most of the time when I was printing from PS.
If over 480 dpi, as Jeff suggests I didn't do anything about it.
If under around 180dpi, I would uprez 200% using bicubic smoother , and then that becomes my new "native" file size that was sent to the printer, the printer driver then handling final sizing. Some say you can get better quality uprezzing to a specific DPI of 240 or 360, and while that may be true, this workflow keeps things simple and output is very good. The critical piece to all of this is the sharpening, which LR simplifies and pretty much automates for you.
LR has changed the workflow dramatically, from conversion to output. You may want to download the 30 day trial just to see what it might do for you.
-
The printer driver also feeds back info to the calling program that the native resolution is 720PPI with glossy paper. This allows to apply sharpening at the interpolated data at 720PPI, which allows to apply more sharpening without visible artifacts.
Only with Epson consumer level photo printers or with large format printers with the driver set to Finest Detail. Normally, large format Epsons will use and report 360 for all paper types. The 720 option (Finest Detail) is not designed for photo work, but for vectors (such as text).
-
I can tell you what I used to do most of the time when I was printing from PS.
If over 480 dpi, as Jeff suggests I didn't do anything about it.
If under around 180dpi, I would uprez 200% using bicubic smoother , and then that becomes my new "native" file size that was sent to the printer, the printer driver then handling final sizing. Some say you can get better quality uprezzing to a specific DPI of 240 or 360, and while that may be true, this workflow keeps things simple and output is very good. The critical piece to all of this is the sharpening, which LR simplifies and pretty much automates for you.
LR has changed the workflow dramatically, from conversion to output. You may want to download the 30 day trial just to see what it might do for you.
Just a note... is there a method of soft proofing your final output in Lightroom. I am unable to find a method so for this reason I still use Photoshop for my output as result tend to be unpredictable on different paper stocks.
-
You seem to suggest that the resolution is limited to 360PPI. If so, then how would you explain that a resolution difference in output (upto 720PPI) can be demonstrated (http://www.ddisoftware.com/qimage/quality/)?
The printer driver also feeds back info to the calling program that the native resolution is 720PPI with glossy paper. This allows to apply sharpening at the interpolated data at 720PPI, which allows to apply more sharpening without visible artifacts.
Cheers,
Bart
Printing the Qimage test chart on my lowly Epson 2200 at maximum resolution does show a difference between 720 and 360 dpi as shown here. Sorry for the typo on the right.
[attachment=20609:Qimage2.png]
-
Just a note... is there a method of soft proofing your final output in Lightroom.
Sadly no...but you CAN of course use Photoshop to open the image FROM Lightroom and soft proof in Photoshop while adding the adjustments needed for a specific printer/paper combination and then save and print from Lightroom-which is what I do.
-
Printing the Qimage test chart on my lowly Epson 2200 at maximum resolution does show a difference between 720 and 360 dpi as shown here. Sorry for the typo on the right.
[attachment=20609:Qimage2.png]
Absolutely. If you feed the right sort of data, you can of course see a difference.
-
If you are under 180ppi, either upsample or make the image size smaller...if you are over 480ppi, I would simply print with the higher rez. I would also use PhotoKit Sharpener's output sharpening after you set the final PPI of the file.
I would also suggest looking at Lightroom if for no other reason that it makes printing a lot, lot easier (and the fact you don't have to spawn off image versions for each and every print size you do).
I will buy Photokit Sharpener. Never having used LR, though, I don't understand the need for it. ACR in conjunction with PS makes logical sense in my mind. But where does LR fit in? What you state above about printing from LR does whet my interest, as I do print in a variety of sizes.
One more thing referencing the 180-480ppi - I get it if I'm over 480 - just leave the native rez alone. BUT...I still would like a more specific response to files less than 180. You state "upsample." I know that much. But how much do I upsample?? Do I go up to 180, then let the Epson take over, do I go to 360, or where? The answer to this question was the base intent of my original inquiry here. Thanks.
(No funny remarks on where else I can go please!)
Michael H. Cothran
Nashville, Tennessee
-
I will buy Photokit Sharpener. Never having used LR, though, I don't understand the need for it. ACR in conjunction with PS makes logical sense in my mind. But where does LR fit in? What you state above about printing from LR does whet my interest, as I do print in a variety of sizes.
One more thing referencing the 180-480ppi - I get it if I'm over 480 - just leave the native rez alone. BUT...I still would like a more specific response to files less than 180. You state "upsample." I know that much. But how much do I upsample?? Do I go up to 180, then let the Epson take over, do I go to 360, or where? The answer to this question was the base intent of my original inquiry here. Thanks.
(No funny remarks on where else I can go please!)
Michael H. Cothran
Nashville, Tennessee
"Ignorance is bliss" I rarely use Photoshop anymore because of Lightroom and V3.0 (Beta) is even better. Lightroom just makes everything easier. I'm not a huge fan of overly processed images and LR is simple, intuitive and does a great job with the basics (cloning not being one of them). Printing from LR is much easier than from Photoshop, just make the presets and you are off to the races.
-
BUT...I still would like a more specific response to files less than 180. You state "upsample." I know that much. But how much do I upsample?? Do I go up to 180, then let the Epson take over, do I go to 360, or where? The answer to this question was the base intent of my original inquiry here.
As Wayne said, I would simply upsample 200% using Bicubic Smoother and add additional sharpening and perhaps even a touch of grain. See this article: The Art Of The Up-Res (http://www.digitalphotopro.com/technique/software-technique/the-art-of-the-up-res.html)
-
Wayne and Jeff -
Thanks for your replies. Jeff, I went back and watched the episode on resolution in the CtoP video again, and everything you and Wayne stated corroborate. But I'm really having a senior moment here, maybe a senior "day," as my mind is drawing a blank on how to calculate 200%.
If you have a 150ppi file, and increase it 200%, how big will the resulting file be? I know that sounds dumb, but if I multipy 150ppi times 200%, I get 300ppi, and if I add it to the original 150ppi, I get 450ppi, so I'm not sure what the correct answer is. Hmmm...
Forgive my stupidity here, but if you, or anyone, can just tell me how big a file you would have if you increase a 150ppi by 200%, then I can backtrack the math.
I promise this will be my last question on this matter.
Thanks.
Michael H. Cothran
-
Wayne and Jeff -
Thanks for your replies. Jeff, I went back and watched the episode on resolution in the CtoP video again, and everything you and Wayne stated corroborate. But I'm really having a senior moment here, maybe a senior "day," as my mind is drawing a blank on how to calculate 200%.
If you have a 150ppi file, and increase it 200%, how big will the resulting file be? I know that sounds dumb, but if I multipy 150ppi times 200%, I get 300ppi, and if I add it to the original 150ppi, I get 450ppi, so I'm not sure what the correct answer is. Hmmm...
Forgive my stupidity here, but if you, or anyone, can just tell me how big a file you would have if you increase a 150ppi by 200%, then I can backtrack the math.
I promise this will be my last question on this matter.
Thanks.
Michael H. Cothran
Just use the percentage option in the image size dialog box, and plug in 200. This will actually calculate the pixel dimensions at the top to accomplish that. For the DPI box, put in any number you want, it doesn't affect the upsample. This is now your new "native" file size, when printing in photoshop you should be well over the 180dpi lower limit. And yes, this will result in a file that is 4 times larger than your original file in size. The reason you use 200% is it will use the optimum resample algorithms which only happen when you do even sampling, such as 12.5%, 25%,50%, 200%, 400%.
One last thing ... 180dpi is not a hard fast rule. Depending on file quality and subject matter, using a number lower than that may yield a perfectly acceptable image. Something with a lot of edge detail and sharp contrasting elements probably not, however a soft focus portrait of a smooth skinned child would probably look just fine at a lower dpi setting.
Here's a screenshot
[attachment=20633:DIALOG.jpg]
-
Printing the Qimage test chart on my lowly Epson 2200 at maximum resolution does show a difference between 720 and 360 dpi as shown here. Sorry for the typo on the right.
[attachment=20609:Qimage2.png]
Hi Bill,
The more advanced printers also vary droplet size <4 pico litre, and will show even more difference (paper quality also makes a difference). Also don't forget that the human eye (or rather the brain) can resolve much more detail than the sample would suggest. The phenomenon is called "Vernier acuity". Besides, sharpening the output at native printer resolution gives cleaner results than sharpening with larger radii.
Cheers,
Bart
-
Hi Bill,
The more advanced printers also vary droplet size <4 pico litre, and will show even more difference (paper quality also makes a difference). Also don't forget that the human eye (or rather the brain) can resolve much more detail than the sample would suggest. The phenomenon is called "Vernier acuity". Besides, sharpening the output at native printer resolution gives cleaner results than sharpening with larger radii.
Cheers,
Bart
Bart,
I wasn't familiar with vernier acuity even though I have used it many times when taking readings from instruments. A quick google search shows a couple of interesting articles: here (http://www.springerlink.com/content/v0717737418221k3/)and here (http://www.pc.ibm.com/ww/healthycomputing/vdt13eyee.html). The essential points are that vernier acuity can be as high as 4-5 arcseconds rather than the commonly reported value of one arcminute, but that this type of acuity is contrast dependent. I'm not certain how this type of acuity would apply to most photographs which do not have high contrast line patterns.
Human vision is astounding in that it uses a relatively crude sensor (the eye) and then processes the information in the neural pathways and brain to achieve very good perceived results. If only we could to this with our cameras!
Although the eye can resolve down to one arcminute (giving a resolution of 30 cycles per degree), the contrast function peaks at about 6 cycles per degree and this is the basis for SQF (subjective quality factor) as explained by Bob Atkins (http://www.bobatkins.com/photography/technical/mtf/mtf4.html) in an essay for the intelligent lay person. MTF in this range is most important for the perceived quality of a print and this value is considerably below the limits of visual resolution.
Although most photographers want their images to be as sharp as possible, most observers would not view an 16 by 24 inch print at nose tip distances and good results can be obtained with resolution that is lower than the limits of visual acuity, since the eye is most sensitive to 6 cycles per degree.
Theoretical considerations aside, I think that the suggestions of Bruce Fraser and Jeff Schewe as outlined in the Real World Sharpening are reasonable for practical photography of most scenes and that is what I use in my own work. Most of the time we do not have as much resolution as we would like, but as the megapixel count of cameras increase, downsizing is necessary either in one's editing software or in the printer driver. As per recent discussions on this site, we probably should give more attention to aliasing and down sampling our sharpening halos out of existence in the process. Also, more attention should be given to image restoration algorithms that go beyond the early 20th century unsharp mask.
I would value your opinion in these matters.
Regards,
Bill
-
Just use the percentage option in the image size dialog box, and plug in 200. This will actually calculate the pixel dimensions at the top to accomplish that. For the DPI box, put in any number you want, it doesn't affect the upsample. This is now your new "native" file size, when printing in photoshop you should be well over the 180dpi lower limit. And yes, this will result in a file that is 4 times larger than your original file in size. The reason you use 200% is it will use the optimum resample algorithms which only happen when you do even sampling, such as 12.5%, 25%,50%, 200%, 400%.
One last thing ... 180dpi is not a hard fast rule. Depending on file quality and subject matter, using a number lower than that may yield a perfectly acceptable image. Something with a lot of edge detail and sharp contrasting elements probably not, however a soft focus portrait of a smooth skinned child would probably look just fine at a lower dpi setting.
Here's a screenshot
[attachment=20633:DIALOG.jpg]
Thanks Wayne. And especially for the screenshot. Things like that really help sort out confusion. In my further ignorance, I did not realize one could change from inches to percentage in the image size dialog box. What a revelation!
I just purchased Photokit Sharpener last night, so tonight I'm in for a treat trying it, along with the resolution practices you and others have shown me here.
I do have pretty good files - from a Hasselblad V camera with Zeiss lenses, from scanned film (producing roughly 8900x8900 pixels) and a CFV 16 mpix digital back (4080x4080 pixels). My digital back will produce a 24" print @ 170 ppi. I feel pretty confident in leaving it alone, but I plan to sample a file tonight with 200% uprezing, using the Photokit Sharpener's Creative Sharpening as described by Jeff in the CtoP video, then the Output sharpening, and see what happens.
I also understand the concept of "even sampling." I currently use the same concept in monitor viewing, utilizing only even sizes in which to view and work. Due to the size of my files, I normally use 12.5% and 6.25% for full image view, and the larger even sizes for local work. You mention smaller even upsampling amounts like 100%, 50%, 25%, etc. Is there ever a time when one might want to use these smaller amounts to uprez, or is it best to stick with 200%? Jeff made mention of 400% uprez, but only on the very best quality of files.
Thanks again,
Michael H. Cothran
Nashville, Tennessee
-
You seem to suggest that the resolution is limited to 360PPI. If so, then how would you explain that a resolution difference in output (upto 720PPI) can be demonstrated (http://www.ddisoftware.com/qimage/quality/)?
The printer driver also feeds back info to the calling program that the native resolution is 720PPI with glossy paper. This allows to apply sharpening at the interpolated data at 720PPI, which allows to apply more sharpening without visible artifacts.
Cheers,
Bart
i'm talking about the mechanical rez of the heads, as in how many nozzles per row.
-
The more advanced printers also vary droplet size <4 pico litre, and will show even more difference (paper quality also makes a difference).
Just to be clear, Epson pro printers do not use variable droplet size when the output is set to 2880-it ONLY outputs the smallest droplet. Depending on the paper they may also not use variable droplets at 1440. They will use variable droplet sizes when set to 720.
-
Bart,
I wasn't familiar with vernier acuity even though I have used it many times when taking readings from instruments. A quick google search shows a couple of interesting articles: here (http://www.springerlink.com/content/v0717737418221k3/)and here (http://www.pc.ibm.com/ww/healthycomputing/vdt13eyee.html). The essential points are that vernier acuity can be as high as 4-5 arcseconds rather than the commonly reported value of one arcminute, but that this type of acuity is contrast dependent. I'm not certain how this type of acuity would apply to most photographs which do not have high contrast line patterns.
Human vision (eye+brain) is mostly sensitive to edges. Edges may be of high spatial frequency and, if at adequate luminance level, they will be resolved at higher than perhaps anticipated levels. Edges are also easier to discern than gratings. Also, human vision doesn't mind at which angle the edges are presented (due to the random placement of the (mostly) rods and (less important for resolution) cones. Digital images however, in a common rectangular X/Y grid configuration, have a (sqrt(2)) higher resolution in the diagonal direction.
Although the eye can resolve down to one arcminute (giving a resolution of 30 cycles per degree), the contrast function peaks at about 6 cycles per degree and this is the basis for SQF (subjective quality factor) as explained by Bob Atkins (http://www.bobatkins.com/photography/technical/mtf/mtf4.html) in an essay for the intelligent lay person. MTF in this range is most important for the perceived quality of a print and this value is considerably below the limits of visual resolution.
Yes, but it is important to understand that boosting the image resolution at a given (e.g. limiting) resolution, will also boost the resolution of slightly lower spatial frequencies (there is no sharp cut-off, it's gradual).
Although most photographers want their images to be as sharp as possible, most observers would not view an 16 by 24 inch print at nose tip distances and good results can be obtained with resolution that is lower than the limits of visual acuity, since the eye is most sensitive to 6 cycles per degree.
At common resolution, yes. That doesn't mean that at optimal luminance levels, and for slightly displaced features, we cannot see a difference between optimal an sub-optimal resolution. How important that distinction is, remains open for debate. However, when the higher quality level can be achieved with only an investment in technique, I don't see why we shouldn't use it (unless 'cost' is an issue). 'Cost' can be expressed as a monetary effort, but also as a computational (or intellectual ...) one. It also has something to do with risk management, the willingness to revisit 'some' percentage of the results (if we're given a chance), to promote them from 'good enough' to excellent. From a commercial standpoint 'good enough' is not competition resistant, especially when the 'cost' is relatively minor.
I would value your opinion in these matters.
It depends on one's goals, (hopefully) good enough versus (as good as it reasonably gets) excellent.
Cheers,
Bart
-
i'm talking about the mechanical rez of the heads, as in how many nozzles per row.
guess I'm confused as well ... how is that relevant to output resolution? Lower mechanical resolution just means more passes to get the same dots on the paper.
-
Exactly. Works great with the Z3100, esp with dslr files.
Or just use Qimage today and select the right extrapolation algorithm for the image content, the 5 degrees of upsampling, the anti-aliasing slider on downsampling and the smart print sharpening slider. All done on the fly at print processing time. Enough ways in Qimage to make a small proof print when in doubt.
met vriendelijke groeten, Ernst Dinkla
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/ (http://groups.yahoo.com/group/Wide_Inkjet_Printers/)