Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: steveblennis on June 07, 2014, 12:49:56 am
-
I suppose its very dependent on quality and a lot of other things.
But....Im curious about letting the printer print above 100%, vs. (how Ive always done it) Software uprez, I use Alien Skin Blow up.
Im usually uprezzing some, if not a lot. Im just starting into some sample prints. But curious what the positives, negatives and any info
about uprezzing at the printer. i.e if it tends to degrade, blur, or otherwise be worse or better than software uprezzing.
In an ideal scenario, uprezzing a little to clean up and sharpen, then letting the printer do the major uprez work, sounds great.
Greg
my current project is below, in this case its more of an illustrative work, which Blow up is doing a good job on.
starting image
2000x1994 px, 11x11 inches, at 180 ppi
uprezz to:
7600x7577 px, 30x30 inches, at 240 ppi
-
Hi,
I got the impression that the optimal workflow is to upsize to the image size needed for printing before applying output sharpening.
Why?
- The output sharpening needs to be done on the actual pixels sent to the printer (or more correctly to the printer driver)
- The printer driver is said to do coarse uprezzing, probably using nearest neighbour
Best regards
Erik
I suppose its very dependent on quality and a lot of other things.
But....Im curious about letting the printer print above 100%, vs. (how Ive always done it) Software uprez, I use Alien Skin Blow up.
Im usually uprezzing some, if not a lot. Im just starting into some sample prints. But curious what the positives, negatives and any info
about uprezzing at the printer. i.e if it tends to degrade, blur, or otherwise be worse or better than software uprezzing.
In an ideal scenario, uprezzing a little to clean up and sharpen, then letting the printer do the major uprez work, sounds great.
Greg
-
You may want to read this thread
http://www.luminous-landscape.com/forum/index.php?topic=90225.0
In an ideal scenario, uprezzing a little to clean up and sharpen, then letting the printer do the major uprez work, sounds great.
-
You may want to read this thread
http://www.luminous-landscape.com/forum/index.php?topic=90225.0
And: this thread (http://www.luminous-landscape.com/forum/index.php?topic=54798.msg446922#msg446922), and this thread (http://www.luminous-landscape.com/forum/index.php?topic=77949.msg624358#msg624358) may also add some more insight.
Cheers,
Bart
-
I suppose its very dependent on quality and a lot of other things.
But....Im curious about letting the printer print above 100%, vs. (how Ive always done it) Software uprez, I use Alien Skin Blow up.
Hi Greg,
Blow-up will do a much better job on straight edge detail than the printer can.
Im usually uprezzing some, if not a lot. Im just starting into some sample prints. But curious what the positives, negatives and any info about uprezzing at the printer. i.e if it tends to degrade, blur, or otherwise be worse or better than software uprezzing.
For small upsampling percentages the differences may be subtle, unless patterns are present in the image that are sensitive to aliasing.
Upsampling by the printer driver, prevents the possibility to output sharpen at the final resolution that the printer uses/creates.
In an ideal scenario, uprezzing a little to clean up and sharpen, then letting the printer do the major uprez work, sounds great.
Unfortunately it produces lower quality, do to the mediocre resampling algorithms the printer driver uses, and the lack of sharpening after resampling.
my current project is below, in this case its more of an illustrative work, which Blow up is doing a good job on.
starting image
2000x1994 px, 11x11 inches, at 180 ppi
uprezz to:
7600x7577 px, 30x30 inches, at 240 ppi
That also doesn't use the native resolution of the printer, which is 360 or 720 PPI, so the printer driver will add another resampling that 240 PPI to 360 PPI. Blow-up should be able to produce much better image quality if you let it do the resampling from 180 to 360 PPI directly, and when you add some detail enhancement at that 360 PPI level, the print should start to really rock.
Cheers,
Bart
-
At the Academy of Learning the question arose: how many teeth does a horse have? A committee of august professors was formed, learned books were consulted, and there was much lively discussion. At some plint the ju nior member suggested finding a horse and counting its teeth. He was immediately thrown off the committee.
Bottom line? Make some prints and see for yourself.
-
Bottom line? Make some prints and see for yourself.
Indeed. There are those who prefer to find their way through the mine-field themselves.
Others appreciate some guidance to get to the other side faster, and in one piece ...
For instance, make some prints, with which printer driver settings? Perhaps 90% of the people here at LuLa were unaware of the benefits of checking the "Finest Detail" option in their Epson LF printer driver until they were told here at LuLa.
Cheers,
Bart
-
That also doesn't use the native resolution of the printer, which is 360 or 720 PPI, so the printer driver will add another resampling that 240 PPI to 360 PPI. Blow-up should be able to produce much better image quality if you let it do the resampling from 180 to 360 PPI directly, and when you add some detail enhancement at that 360 PPI level, the print should start to really rock.
Cheers, Bart
Thanks. I arrived at 240 after a ton of searches and articles a few years ago. I had heard of that native resolution, but if I remember, if it was divisible by say 12, it would work. Well Im back and ready to relearn. Maybe I should revisit my luminous-landscapes videos.
Im on Canvas, so (not thinking) I needed 360, Ive been using 240.
Back to the books ! ;) Thanks
-
Thanks for the links to other threads. I know there are the 'RTFM' and 'plow through it yourself' mindsets, but
Im all about learning from those who either know or at least have a much better solution than I do.
Thanks everybody.
The Native Resolution thing is starting to sink in.
-
The Native Resolution thing is starting to sink in.
Hi Greg,
That's the essential starting point. The printer driver usually only supports one or two native resolutions. For Epsons it's 360 PPi and 720 PPI, and for Canons an HPs, Lexmark, etc. it's 300 PPI and 600 PPI. There may be other settings available in the printer driver, but effectively those PPIs are what it's at. Other output modalities, like C-print equipment (using photochemical paper), or dedicated (high resolution or solvent) inkjet print stations may require different settings (ranging from 254 PPI to some 720 PPI).
Any other PPI sent to the printer will be resampled to those values, and it's done with relatively low quality resampling algorithms, so one better does it oneself (which additionally allows to output sharpen at the final output size).
Cheers,
Bart
-
Indeed. There are those who prefer to find their way through the mine-field themselves.
Others appreciate some guidance to get to the other side faster, and in one piece ...
For instance, make some prints, with which printer driver settings? Perhaps 90% of the people here at LuLa were unaware of the benefits of checking the "Finest Detail" option in their Epson LF printer driver until they were told here at LuLa.
Cheers,
Bart
I have nothing against asking advice from more experienced photographers. The point I was trying to make, albeit in a round-about way, was that following the advice of experts may get you good prints, but you don't learn much. Figuring it out yourself, at least partly, is the best way to learn. As an analogy, slavishly following recipes versus experimenting. We learn a lot from mistakes!
-
... Im on Canvas, so (not thinking) I needed 360, Ive been using 240...
Given that you are on canvas, you probably won't see much difference anyway.
-
Any other PPI sent to the printer will be resampled to those values, and it's done with relatively low quality resampling algorithms, so one better does it oneself (which additionally allows to output sharpen at the final output size).
Reading through all those linked threads addressing the subject of "low quality resampling algorthms" I was frustrated as usual to find not one poster went through the trouble of taking a side by side comparison shots of two prints using each upsampling method and posted it or even provided a link to a site that did.
Andrew's post stating he could only see the differences viewing through a loupe pretty much told me the time I took reading through all of it was the equivalent of mining the Library of Congress for info on how to screw in a light bulb.
-
... Andrew's post stating he could only see the differences viewing through a loupe...
If I am not mistaken, Andrew was looking at a letter-size print. As usual, the size does matter ;) Otherwise, I agree with you, it would be nice to see the difference.
-
Reading through all those linked threads addressing the subject of "low quality resampling algorthms" I was frustrated as usual to find not one poster went through the trouble of taking a side by side comparison shots of two prints using each upsampling method and posted it or even provided a link to a site that did.
Read this and get back to me with any questions...The Right Resolution (http://www.digitalphotopro.com/technique/workflow/the-right-resolution.html#.U5O6gCjW-bA).
-
Reading through all those linked threads addressing the subject of "low quality resampling algorthms" I was frustrated as usual to find not one poster went through the trouble of taking a side by side comparison shots of two prints using each upsampling method and posted it or even provided a link to a site that did.
Here (http://blog.kasson.com/?s=resampling)you go...
-
Read this and get back to me with any questions...The Right Resolution (http://www.digitalphotopro.com/technique/workflow/the-right-resolution.html#.U5O6gCjW-bA).
Thanks I'll check it out
-
Thanks folks, Im on an Epson 9900 by the way. ok so Im getting some clarity.
Good point from Bart and Slobodan Blagojevic
Any other PPI sent to the printer will be resampled to those values, and it's done with relatively low quality resampling algorithms, so one better does it oneself (which additionally allows to output sharpen at the final output size).
and
Given that you are on canvas, you probably won't see much difference anyway.
I probably wont see a wildly better print, but I wont make it much worse "resampling with relatively low quality resampling algorithms".
Thanks
I have two projects for an Interior Designer in Miami, I'll put the new information to use.
-
Photographs: http://www.kasson.com/
Jim, really nice UW shots! I hadn't known you did that :~)
-
I probably wont see a wildly better print, but I wont make it much worse "resampling with relatively low quality resampling algorithms".
Actually, resampling to the printer's reported resolution CAN help canvas if you have image detail that has strong high contras diagonals (or circles or other texture that has a specific pattern) because even with canvas and matte watercolor paper, you can still see the printer resampling aliasing errors in cases like this.
-
Photographs: http://www.kasson.com/
Jim, really nice UW shots! I hadn't known you did that :~)
Thanks, Jeff. The person who took the underwater stuff was a much younger man. He even used film for all but one series...
Jim
-
Here (http://blog.kasson.com/?s=resampling)you go...
wow all of info thanks. i see some are 'finest detail' related. I was wondering about that checkbox.
-
Actually, resampling to the printer's reported resolution CAN help canvas if you have image detail that has strong high contras diagonals (or circles or other texture that has a specific pattern) because even with canvas and matte watercolor paper, you can still see the printer resampling aliasing errors in cases like this.
Thanks nice tip.
-
Actually, resampling to the printer's reported resolution CAN help canvas if you have image detail that has strong high contras diagonals (or circles or other texture that has a specific pattern) because even with canvas and matte watercolor paper, you can still see the printer resampling aliasing errors in cases like this.
I agree with Jeff. I've produced canvas prints at 720 (!) PPI (which allowed to also output sharpen to a higher degree!) that looked very sharp. A lot also depends on the particular type of canvas, and for the surface areas that have the same distance to the jet nozzles as with glossy paper, the sharpness can be identical.
I see no reason to lower the quality if it takes no more effort than resampling to a fixed PPI. I still need to output sharpen after that, and I won't need to depend on whatever flavor of resampling the printer driver utilizes, which might even differ between OS versions (OS X versus Windows) according to some.
Cheers,
Bart
-
First test done. I'd say pretty good success, looks good considering it wasnt a huge file in the first place.
I did click the 'finest detail' with a little bit of a test in mind. After the fact I see it seems to resample to 720,
and only be good for more vector type images. but it still came out good.
Im going to do another copy without the 'finest detail'.
Im on to a 30x30 next.
Thanks for all the info, with time, canvas and ink = money. Im glad to have prepared first.
What do the woodworkers say, "measure twice, cut once".
-
I did click the 'finest detail' with a little bit of a test in mind. After the fact I see it seems to resample to 720,
and only be good for more vector type images. but it still came out good.
To be clear, Epson has pretty much said Finest Detail was only good for vector art, but, they've been too conservative in stating it's usefulness. We've proven that Finest Detail CAN help photographic images where there is high contrast diagonal (and circles) and other high frequency image detail. In these cases, the images get the same benefit of improving aliased edges as the vector artwork does. In essence, it's solving the problem of refining the aliased edges of both photos and vector art. So, what's good for the goose is good for the gander, so to speak.
I've talked at length with the EPson people and they are aware of this new found use for printing photos. It's just they've never gotten around to doing the testing and work they would need to do to make these claims of improvement.
In point of fact, there's actually a technique to have the printer report it's native resolution to be 1440 which is 2x the Finest Detail. The only trick (and limitation) is it'll ONLY work if you select Finest Detail and ALSO select certain media types-specifically, certain proofing papers. The reason for this is in proofing for halftone screening, printing out at 1440 can better help resolve an actual halftone dot.
I tested this using proofing papers to compare 720PPI with Finest Detail and 1440 with Finest Detail and with the same photos I could see no visible difference–nowhere near what I could see printing at 360 without Finest Detail and with 720 with Finest Detail.
In order to use this 1440 trick, you would have to make custom profiles using the Proofing media settings and using your own paper (non-proofing paper) to get optimal results for tone & color.
Since I couldn't see any difference (and I know what to look for) I decided that was a rabbit hole I didn't need to go down. So, the bottom line is if the native rez is less than 360, I upsample to 360 (and then sharpen) if the native rez is above 360 but below 720, I upsample and sharpen. If the native rez is about 720, I downsample to 720 and sharpen. The reasoning for downsampling is again to take the resampling away from the print pipeline.
-
Thanks great info. it really puts it all in perspective.
Ive been selling prints now for about 2 years, I started off pretty green, but with decent results.
In the last year getting into groove, and now really trying to dig into some of these nuance things to bring my prints to as much of a pro level as I can.
Thanks great stuff.
-
High Speed setting question.
I pretty much always have this checked. As a Side note: I did see that if you have 'Finest Detail' checked, you should probably uncheck 'High Speed'.
But the question is.......
Does unchecking 'High Speed' give you any better quality difference. or is 'High Speed' the way to go generally?
I would take quality over speed.
-
Does unchecking 'High Speed' give you any better quality difference. or is 'High Speed' the way to go generally?
If memory serves, checking High-Speed enables bidirectional printing. There is a possibility of increased dynamic registration errors when the print head has to hit the same spot moving from either side. I always leave it unchecked, because I'm not in a hurry. But I've never done any comparative testing. I would think the results would depend on your particular (serial #, not model #) printer.
Jim
-
Does unchecking 'High Speed' give you any better quality difference. or is 'High Speed' the way to go generally?
Depends, when was the last time you did a head alignment? If you have the head well aligned (I use auto-align) then high speed on/off is gonna be real close. However, if you are working with high rez images and printing out 720ppi images with the printer set to Finest Detail and 2880 print resolution, you might ("might") see some slight benefit due to the more accurate per swath registration...I do head alignments whenever I change papers and generally print with high-speed on. But, when I'm printing out at 720ppi with Finest Detail, I turn it off. Note, when printing at 720ppi, I'm generally making smaller prints (less than 16x20) that will be viewed much closer than larger prints. For that reason I think it's marginally optimal to print with high-speed off.
-
High Speed setting question.
I pretty much always have this checked. As a Side note: I did see that if you have 'Finest Detail' checked, you should probably uncheck 'High Speed'.
But the question is.......
Does unchecking 'High Speed' give you any better quality difference. or is 'High Speed' the way to go generally?
I would take quality over speed.
High speed (bidirectional) printing can also cause banding in large black/near black areas on certain media types. Some can handle it, some can't.
-
High speed (bidirectional) printing can also cause banding in large black/near black areas on certain media types. Some can handle it, some can't.
ah, yeah i did an almost all black canvas print. that is tough enough with white spotting etc.
im on canvas, but i bet thats a real pain on a nice glossy print.
-
wow ok, knocked out my 2nd print, up to 30x30ish.
came out great
starting:
2000 x 1994 180ppi 1.43MB
to:
11999 x 12031 360ppi 2.39GB
super happy ;D
-
Jim would liketo see the results i a sort of split view, as the PR maybe specatacular on the graphical part but is not without artefacts on the tree branches and the circular grey gradient is not as smooth as f.i.the LR result.
-
My 2nd print. Its a Frantisek Kupka watercolor, 1920-ish, so pretty smooth flat, like vector graphic.
Probably lends itself to uprezzing, but looks great. 31x31 printed area.
Update on 'Finest Details' - Made 2 prints, canvas, 360 ppi etc etc
On a very painterly artist print, 'Finest Details' wins out. Ever so subtle at 6-12 inches. About a 10%-ish pop in sharpness (contrast) -as best as I can describe.
Oddly enough it noticeable in the finest details. :) The very small canvas pattern in areas is mostly noticeable.
-
Steve, same effect with my prints. This "pop" is quite visible. On my epson 4900 my standard setup is 720 ppi and finedetails on, and highspeed on. Even for fine art papers, as the effect is at least less bleed of ink at high contrast edges on the paper. (720 ppi does put ( a little) more ink on the paper) Also as Jeff's article points out, a smoother result on tonal and saturation changes.
-
On a very painterly artist print, 'Finest Details' wins out. Ever so subtle at 6-12 inches. About a 10%-ish pop in sharpness (contrast) -as best as I can describe.
Hi Steve,
There is more ... Once we have upsampled to the 600/720 PPI native printer driver resolution, we can add not only output sharpening to counteract ink diffusion, but we can also do more accurate detail enhancement, e.g. with a plugin such as Topaz Labs Detail (http://www.topazlabs.com/detail). It allows to not just sharpen the micro-detail, but it also allows to give larger structural detail in the image more (or less) definition (typically used a bit earlier in the creative sharpening part of processing). There will be no halos added, just detail enhancement which will also allow the image to 'pop' at more than a nose length away. And of course one can also add noise which becomes almost subliminal at this detail level.
Oddly enough it noticeable in the finest details. :) The very small canvas pattern in areas is mostly noticeable.
Yes, that's what I see on canvas prints as well. Details remain very sharp, especially at the highest native printer driver resolution, and obviously the canvas structure dominates in regions that otherwise have little subject feature content.
Cheers,
Bart
-
Even for fine art papers, as the effect is at least less bleed of ink at high contrast edges on the paper. (720 ppi does put ( a little) more ink on the paper)
Hi Jan,
The dithering pattern at 720 PPI is supposed to be made up of smaller droplets, which allows to have sharper edge detail that otherwise would be rendered less accurately with larger droplets. In other regions of the image that may take more droplets to build up density and cover the paper substrate when the ink doesn't diffuse much. I assume that the dithering algorithms could also be changed to use larger droplets in those not detail regions, so things may change (although selling more ink (~liquid gold) is a goal for the manufacturers).
Cheers,
Bart
-
Jim would liketo see the results i a sort of split view, as the PR maybe specatacular on the graphical part but is not without artefacts on the tree branches and the circular grey gradient is not as smooth as f.i.the LR result.
I understand. The split-screen approach has a problem in that you're not looking at the same part of each image, unless you have a slider like they do with the before and after tsunami pictures, and I don't know how to do that. What I've done with some other images that were hard to compare on the screen was supply Photoshop composite images with each sub image in a layer. I like what Lloyd Chambers does for his comparisons, but I don't know how to do that either.
To your second point. Yes, unfortunately, there's no one algorithm that delivers stunning performance for every kind of image.
That said, I'm not anxious to repeat the resampling algorithm testing. One, The proprietary algorithms can change over time, so there's a finite shelf like to testing them. Two, since Lightroom 4, I've been happy enough with the job that Lr does that I only use less-convenient techniques when I get a print order or I have an exhibition. Neither one of those things is a weekly occurrence -- I wish it were -- so I mostly just use Lr. My motivations for my blog are so that others can profit from what I'm doing for myself, so if I'm no longer spending time testing resampling algorithms for my own work, I'm no longer doing it at all. Sorry.
The good news is that, if you have a scanner, you can do everything but the photomicrographs yourself. Or, skip the scanner and get up close and personal with the print. You now know that the resampling algorithm can make a difference, and that, among good algorithms, the difference is usually subtle.
I'll be doing another post on the Lr3-->Lr4 print sharpening change. Foreshadowing: it wasn't a change to the sharpening algorithm per se.
Jim
-
I'll be doing another post on the Lr3-->Lr4 print sharpening change. Foreshadowing: it wasn't a change to the sharpening algorithm per se.
I was on a panel on raw processing about a year ago, and so was Eric Chan. At the break, I congratulated him on the big improvements in resizing in Lightroom 4. He said that the algorithms were similar; the big change was that, rather than doing the calcs in the standard Lr working space, which has a gamma of one, they were doing the resizing in a gamma encoded space which is more perceptually uniform. We didn't discuss the gamma, but it's probably either the ProPhotoRGB gamma of 1.8, or 2.2.
That's what I was going to report to you all. But, me being me, I had to do some testing first. I wrote a little Matlab script to upres a test image by 80% at gammas from 1.0 to 3.8 using bicubic interpolation:
(http://www.kasson.com/ll/scalegammatest.PNG)
When I ran it, I did see some differences. But they were very subtle; nothing like the differences between Lr 3 resampling and Lr 4 resampling.
I scratched my head. Then I thought that there's usually sharpening associated with up-resing. So I modified my script to do unsharp masking at the same gamma set:
(http://www.kasson.com/ll/sharpengammacode.PNG)
Now there was a big difference, with the images that were sharpened at higher gammas being crisper. This could be a good thing or a bad thing, depending on what effects you're trying to achieve.
Lessons:
1) although it's a weak effect, people comparing resizing algorithms should probably take pains to make sure they do all their resizings at the same gamma, and should report that gamma when they report the results.
2) people comparing sharpening algorithms should definitely take pains to make sure they do all their sharpening at the same gamma, and should report that gamma when they report the results.
I'll have images up on my blog in a few days -- I've got a backlog.
Jim
-
I was on a panel on raw processing about a year ago, and so was Eric Chan. At the break, I congratulated him on the big improvements in resizing in Lightroom 4. He said that the algorithms were similar; the big change was that, rather than doing the calcs in the standard Lr working space, which has a gamma of one, they were doing the resizing in a gamma encoded space which is more perceptually uniform. We didn't discuss the gamma, but it's probably either the ProPhotoRGB gamma of 1.8, or 2.2.
Well, you've discovered Eric's secret :~)
In an attempt to improve certain aspects of the final resampling and sharpening, it was determined that sharpening in linear space was suboptimal. I'm not sure when Eric made the change, but there was a certain point where some breakthroughs in Eric's testing lead to the improvements in LR's resampling and sharpening for output.
Not for nothing, if anybody can advocate additional improvements, I'm sure Eric would like to hear about it :~)
-
In an attempt to improve certain aspects of the final resampling and sharpening, it was determined that sharpening in linear space was suboptimal.
Hi Jeff,
I'm afraid that's not universally true, but I assume Eric knows that but also has to deal with complexity trade-offs of the current sharpening dialog (it could use a serious overhaul). Deconvolution makes more sense in a linear gamma space, edge enhancement makes more sense in a gamma pre-compensated space (to reduce differences in acutance on the light side and the dark side of the edges/lines), and it would benefit from a possibility to target lighter, medium, and darker tones because their contrast is different in non-linear gamma space.
I'm not sure when Eric made the change, but there was a certain point where some breakthroughs in Eric's testing lead to the improvements in LR's resampling and sharpening for output.
He probably, amongst others, paid attention to the discussions on the ImageMagick board (http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=22005&start=15#p91283) and website where it was concluded quite some time ago that downsampling (http://www.imagemagick.org/Usage/filter/nicolas/#downsample) would benefit from linear gamma space filtering, and upsampling (http://www.imagemagick.org/Usage/filter/nicolas/#upsampling) would perceptually benefit from doing that in non-linear gamma space.
Professor Nicolas Robidoux also did extensive trials (http://www.imagemagick.org/Usage/filter/nicolas/) of various methods used by ImageMagick, including temporarily (sigmoidal) contrast adjusted upsampling, especially on the more advanced Elliptical Weighted Averaging (EWA) filtered approaches but also on the tensor based variants of Lanczos windowing.
Cheers,
Bart
-
Hi Bart,
On a side note, I wonder why Adobe had not sent you an invitation and hired you like Eric many years ago already! With your Kodak background and overall expertise, it certainly would be a win-win. Then again, who knows, perhaps you would have declined the invite. ;D
On this occasion, and seriously, many thanks for years of ruthless expertise sharing.
Hope all is well with you.
Best
Georg
-
Hi Bart,
On a side note, I wonder why Adobe had not sent you an invitation and hired you like Eric many years ago already! With your Kodak background and overall expertise, it certainly would be a win-win. Then again, who knows, perhaps you would have declined the invite. ;D
On this occasion, and seriously, many thanks for years of ruthless expertise sharing.
Hope all is well with you.
Hi Georg,
Thanks for the kind words. I'm fine, hope you are fine as well.
I suppose Adobe (and some others), prefer to work with local (USA) folks.
I'm always looking for new developments in all sorts of fields (lots of interesting research going on), and tend to combine the best practices and high potentials from those various fields with common sense and pragmatism, sometimes with surprising results when applied to a specific (e.g. photographic) issue. I know when good enough is good enough, but that doesn't mean one should stop to try raising the bar and differentiate from competition.
Aim low, and you'll shoot something low, aim higher and you'll shoot something higher ... (metaphorically speaking, I do not endorse the use of guns ;) )
Cheers,
Bart
-
I know when good enough is good enough….
That in itself is an art that "develops" ;) with experience only.
I'll send you a PM with my updated contact details. Would love to have your thoughts on the recent Sony CMOS development and inevitable? (Software solution possible perhaps, I am not sure.) channel crosstalk.
-
...
He probably, amongst others, paid attention to the discussions on the ImageMagick board (http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=22005&start=15#p91283) and website where it was concluded quite some time ago that downsampling (http://www.imagemagick.org/Usage/filter/nicolas/#downsample) would benefit from linear gamma space filtering, and upsampling (http://www.imagemagick.org/Usage/filter/nicolas/#upsampling) would perceptually benefit from doing that in non-linear gamma space...
I am not so sure anymore that enlargement (upsampling) generally benefits from being performed, say, through Lab (=L*a*b*) or other perceptual color space (with gamma different than 1), or through a sigmoidal color space (which is a bit too blunt a tool, at least in the published versions, to be used blindly even though it sometimes gives very good results).
If you are using an upsampling method that does not create strong halos, enlarging through linear light appears to be generally better at preventing color artifacts.
With an upsampling method that adds overshoots and undershoots (like Lanczos filtering), it appears that using a perceptual color space (or sigmoidal one) may often be better, like it may be when sharpening.
The advice given in http://www.imagemagick.org/Usage/filter/nicolas (http://www.imagemagick.org/Usage/filter/nicolas) is not "wrong" but it is slowly going stale. Let's call it a starting point.
For example, there are two filters that I now quite like that are not on the list: EWA RobidouxSoft http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=19823&p=109820#p109820 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=19823&p=109820#p109820) or actually any Keys cubic with alpha (C in ImageMagick) larger than the RobidouxSoft value but strictly less than the Robidoux value, and EWA LanczosSharpest 4 http://www.imagemagick.org/discourse-server/viewtopic.php?f=2&t=25268#p109493 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=2&t=25268#p109493).
(Windows users: http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=23850&p=101409#p101409 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=23850&p=101409#p101409).)
I wish I had the time for "round 2". But I don't: It requires a lot experimentation, comparison and discussion to put such things together. At this point I can only ask questions.
P.S. In other words, I wish my "extensive trials" had been more extensive.
-
Sometimes, however, just asking the question is a step forward: http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=21804 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=21804)
-
...
1) although it's a weak effect, people comparing resizing algorithms should probably take pains to make sure they do all their resizings at the same gamma, and should report that gamma when they report the results.
...
Sections 2.6 and 6.6 of my former student Adam Turcotte's Masters thesis
THE EXQUIRES (EXTENSIBLE QUANTITATIVE IMAGE RESAMPLING) TEST SUITE: IMPACT OF THE DOWNSAMPLER, DIFFERENCE METRIC, TEST IMAGE, RESAMPLING RATIO AND COLOUR SPACE ON UPSAMPLER RANK
http://web.cs.laurentian.ca/nrobidoux/misc/AdamTurcotteMastersThesis.pdf (http://web.cs.laurentian.ca/nrobidoux/misc/AdamTurcotteMastersThesis.pdf)
-
Hi Jeff,
I'm afraid that's not universally true, but I assume Eric knows that but also has to deal with complexity trade-offs of the current sharpening dialog (it could use a serious overhaul). Deconvolution makes more sense in a linear gamma space, edge enhancement makes more sense in a gamma pre-compensated space (to reduce differences in acutance on the light side and the dark side of the edges/lines), and it would benefit from a possibility to target lighter, medium, and darker tones because their contrast is different in non-linear gamma space.
He probably, amongst others, paid attention to the discussions on the ImageMagick board (http://www.imagemagick.org/discourse-server/viewtopic.php?f=22&t=22005&start=15#p91283) and website where it was concluded quite some time ago that downsampling (http://www.imagemagick.org/Usage/filter/nicolas/#downsample) would benefit from linear gamma space filtering, and upsampling (http://www.imagemagick.org/Usage/filter/nicolas/#upsampling) would perceptually benefit from doing that in non-linear gamma space.
Professor Nicolas Robidoux also did extensive trials (http://www.imagemagick.org/Usage/filter/nicolas/) of various methods used by ImageMagick, including temporarily (sigmoidal) contrast adjusted upsampling, especially on the more advanced Elliptical Weighted Averaging (EWA) filtered approaches but also on the tensor based variants of Lanczos windowing.
Cheers,
Bart
I wish I could take credit for inspiring Lightroom improvements but as far as purposely using something else than gamma 1 when upsampling I started openly pushing the idea in the Spring of 2012 (quite emphatically in http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=20992 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=20992)) which is after the release of Lightroom 4.
-
Aim low, and you'll shoot something low, aim higher and you'll shoot something higher ... (metaphorically speaking, I do not endorse the use of guns ;) )
I assumed you were referring to cameras ;)
Jeremy
-
...
Deconvolution makes more sense in a linear gamma space, edge enhancement makes more sense in a gamma pre-compensated space (to reduce differences in acutance on the light side and the dark side of the edges/lines), and it would benefit from a possibility to target lighter, medium, and darker tones because their contrast is different in non-linear gamma space.
...
:)
P.S. http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=25736 (http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=25736)