Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Printing from QimagU: Interpolation?  (Read 2025 times)
walter.sk
Sr. Member
****
Offline Offline

Posts: 1332


« on: January 19, 2013, 09:02:01 PM »
ReplyReply

Just out of curiosity, what method of interpolation do you use when printing from Qimage?  I had been using "Fusion" or "Hybrid" but the Help file says that Hybrid SE bests Hybrid by being smoother and having fewer artifacts.  I am printing on an Epson 4900 at up to 16 x 24" on Ilford Smooth Pearl paper from files from a 5DIII and a 1DII.
Logged
David Good
Full Member
***
Offline Offline

Posts: 145


« Reply #1 on: January 20, 2013, 08:20:14 AM »
ReplyReply

Walter, with the Ultimate version I have mine set to "Fusion".

http://www.ddisoftware.com/qimage-u/tech-fus.htm

Dave
Logged

kdphotography
Sr. Member
****
Offline Offline

Posts: 729


WWW
« Reply #2 on: January 20, 2013, 08:38:26 AM »
ReplyReply

I'm using Qimage Ultimate.  My default is set on Fusion as well.

ken
Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3749


« Reply #3 on: January 20, 2013, 09:52:36 AM »
ReplyReply

Just out of curiosity, what method of interpolation do you use when printing from Qimage?  I had been using "Fusion" or "Hybrid" but the Help file says that Hybrid SE bests Hybrid by being smoother and having fewer artifacts.  I am printing on an Epson 4900 at up to 16 x 24" on Ilford Smooth Pearl paper from files from a 5DIII and a 1DII.

Hi Walter,

It depends on how much magnification is required, and if you are going to work on the output file (e.g. adding text at highest resolution, or enhance sharpening) or not.

I use Hybrid SE (because it doesn't introduce any halo) if I subsequently want to apply deconvolution sharpening to the upsampled result. On the other hand, if the file is the end product (e.g. for off-site printing) already, and the magnification is no more than 2x or 3x the number of pixels in the original, I'd use Fusion for a good compromise.

For the very best results I use Photozoom Pro to do the upsampling, and then take the resulting file through Qimage for it's consistent handling of the settings for specific media, and its layout features.

Cheers,
Bart
Logged
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #4 on: January 20, 2013, 11:43:22 AM »
ReplyReply

Hi Walter,

It depends on how much magnification is required, and if you are going to work on the output file (e.g. adding text at highest resolution, or enhance sharpening) or not.

I use Hybrid SE (because it doesn't introduce any halo) if I subsequently want to apply deconvolution sharpening to the upsampled result. On the other hand, if the file is the end product (e.g. for off-site printing) already, and the magnification is no more than 2x or 3x the number of pixels in the original, I'd use Fusion for a good compromise.

For the very best results I use Photozoom Pro to do the upsampling, and then take the resulting file through Qimage for it's consistent handling of the settings for specific media, and its layout features.

Cheers,
Bart

Bart - what software gave you halos by doing a resolution up-sample?  I've never seen that with the different flavors of bi-cubic in Photoshop.  I do all my image editing and sharpening once the image is at my printer's native resolution.

Sal
Logged
davidh202
Sr. Member
****
Offline Offline

Posts: 551


« Reply #5 on: January 20, 2013, 02:49:41 PM »
ReplyReply

Th combination of the new DFS (deep focus  sharpening) for relatively artifact free sharpening,  http://www.youtube.com/watch_popup?v=AiVoXcB1uzk&vq=hd1080  ,  and Fusion interpolation, cannot be beat!
I think the help files were written before the debut of these features last year.
David
Logged
walter.sk
Sr. Member
****
Offline Offline

Posts: 1332


« Reply #6 on: January 20, 2013, 06:56:03 PM »
ReplyReply

Thanks, all, for your input.  So far, I don't need to do further work on a file after the Qimage on-the-fly upsampling, so I guess Fusion would work just fine (I had been using it until now).  But I do use the Deep Focus Sharpening in QimageU, and since David points out that the help files were probably written before DFS and HybridSE were introduced, the HybridSE is probably the best uprezzing paradigm.  I'll use it for a while on some prints and see if I can even see a difference.  If I get a stretch of time I might "print to file" some large prints and then print some 8.5x11 crops done, one with HybridSE and one with Fusion.  I'll look for a file that has some high frequency, low frequency and solid areas near each other.
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3749


« Reply #7 on: January 21, 2013, 05:00:36 AM »
ReplyReply

Bart - what software gave you halos by doing a resolution up-sample?  I've never seen that with the different flavors of bi-cubic in Photoshop.  I do all my image editing and sharpening once the image is at my printer's native resolution.

Hi Sal,

They all produce halos to a certain extent, with the exception of PhotoZoom Pro. That will of course only become visible after substantial magnification and close-up inspection, especially on glossy media.

Here is a screenshot of the results after resampling a slanted edge to 300% of the original size in pixels, without sharpening, only resampling. This will just start becoming visible in print, when printed at 600/720 PPI on glossy media, on sharp high contrast edges (branches against sky, architecture, powerlines, etc.) in otherwise smooth areas:

Click on the image for full size.


Sharpening will make the artifacts become more visible, so the one with the least visible artifacts will stand more sharpening, but also needs more sharpening. Going larger, will also mean longer viewing distances, but on close inspection the artifacts will of course become even easier to see.

The only applications that avoid halos are the ones that vectorize edges and smoothly interpolates other image features, such as PhotoZoom Pro:

Click on the image for full size.

Here the sharp edges remain sharp, and thin, and are only anti-aliased. The actual edge resolution has increased. It does require some moderation (which the program allows) when going to very large format output, because the edges will remain relatively too sharp compared to the lower contrast image features.

Cheers,
Bart

P.S. Here is another test image with edges in all directions, feel free to resample by whatever factor you most often require:
« Last Edit: January 21, 2013, 06:12:38 PM by BartvanderWolf » Logged
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #8 on: January 21, 2013, 09:22:05 AM »
ReplyReply

Thanks for doing the test Bart.  It did make me go back and test the different BiCubic options.  BiCubic smooth does do the best at enlarging an image, as Adobe recommends. 

These tests are good to show extreme limits of different technology but are of little to no value, at least to me, in real world printmaking.  I'm sure that viewing maximum contrast images in geometrically perfect edges will show many problems when viewed at 400% after a 300% enlargement.  I guess QI gets bragging rights in that situation, but barely.  I only up-sample from 50% to 100-ish% (20x30 prints), and views of 100% are already pretty extreme as related to a final print.  I tried a quick duplication of the test at 100% up-sampling, and viewed it at a size that I care about (50%) and there are no visible halos.  That's probably why I've never had issues with BiCubic I suppose. 

I went to the QI website to see if their sharpening algorithm was indeed revolutionary.  I found that their comparison test shows PS unsharp mask at a radius of 10 (!) and strength of 100.  Of course it looked like garbage.  They mention that the QI example had the same settings, but who can say that the settings of their new technology are anywhere close to the settings for USM.  They then show a "more modest" comparison but don't mention any settings.

QI is a great product and has loyal fans, but IMO it doesn't offer anything new to experienced users who know how to use existing tools (like PS) properly.  It seems it's USP is convenience, simplicity and price.  Not that there's anything wrong with that!  Smiley

Sal
Logged
davidh202
Sr. Member
****
Offline Offline

Posts: 551


« Reply #9 on: January 21, 2013, 09:45:11 AM »
ReplyReply

   It seems it's USP is convenience, simplicity and price.  Not that there's anything wrong with that!  Smiley Sal

Precisely ;-)

David
Logged
mchaney
Newbie
*
Offline Offline

Posts: 15


« Reply #10 on: January 21, 2013, 10:02:48 AM »
ReplyReply

Thanks for doing the test Bart.  It did make me go back and test the different BiCubic options.  BiCubic smooth does do the best at enlarging an image, as Adobe recommends.  

These tests are good to show extreme limits of different technology but are of little to no value, at least to me, in real world printmaking.  I'm sure that viewing maximum contrast images in geometrically perfect edges will show many problems when viewed at 400% after a 300% enlargement.  I guess QI gets bragging rights in that situation, but barely.  I only up-sample from 50% to 100-ish% (20x30 prints), and views of 100% are already pretty extreme as related to a final print.  I tried a quick duplication of the test at 100% up-sampling, and viewed it at a size that I care about (50%) and there are no visible halos.  That's probably why I've never had issues with BiCubic I suppose.  

I went to the QI website to see if their sharpening algorithm was indeed revolutionary.  I found that their comparison test shows PS unsharp mask at a radius of 10 (!) and strength of 100.  Of course it looked like garbage.  They mention that the QI example had the same settings, but who can say that the settings of their new technology are anywhere close to the settings for USM.  They then show a "more modest" comparison but don't mention any settings.

QI is a great product and has loyal fans, but IMO it doesn't offer anything new to experienced users who know how to use existing tools (like PS) properly.  It seems it's USP is convenience, simplicity and price.  Not that there's anything wrong with that!  Smiley

Sal

You really need to take some time to use Deep Focus Sharpening rather than just making guesses based on web examples.  It really is revolutionary!  The only reason you think a radius of 10 is ridiculous is because you can't do it with USM without ruining the image!  With DFS, you can get the same level of sharpening as USM but you don't have to handicap yourself with tiny radii just because of the halo artifacts.  That opens the door to new levels of sharpening (and de-fogging) because you are not so limited in what you can do.

BTW, the second example (of the rose) is at radius 4.

http://www.ddisoftware.com/qimage-u/tech-dfs.htm

P.S. You really have to run a lot of (real) images through an interpolator to get an idea of how well it works: running a single edge through it is really not going to tell you a lot about it.  All the methods have their pros and cons.  Some work better on hard edges.  Some are better at rendering fine "random" detail like leaves or grass at a distance.  Some have problems with circular patterns.  You really have to interpolate a lot of images and pick the interpolator with the best overall balance.  FWIW, the ones that make the sharpest edges often produce results that can look a little "fake" because they break the relationship between sharpness and detail, making some edges much sharper than they really should be.  Finally (sorry, I keep editing to add more because there's a lot involved)... keep in mind that with today's digital cameras, interpolation by super large amounts is less common and nearly all of the interpolation algorithms can do a good job up to about 2x which is likely all you'll ever use.  What is more important is to match the printer's native resolution and to have a sharpening algorithm that can produce the best final output sharpening.  With the high resolution photos of today's cameras, a good sharpening algorithm is more useful than interpolation algorithms, hence the importance of something like DFS that can sharpen without halo artifacts.

Mike
« Last Edit: January 21, 2013, 11:02:21 AM by mchaney » Logged
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #11 on: January 21, 2013, 11:25:23 AM »
ReplyReply

You really need to take some time to use Deep Focus Sharpening rather than just making guesses based on web examples.  It really is revolutionary!  The only reason you think a radius of 10 is ridiculous is because you can't do it with USM without ruining the image!  With DFS, you can get the same level of sharpening as USM but you don't have to handicap yourself with tiny radii just because of the halo artifacts.  That opens the door to new levels of sharpening (and de-fogging) because you are not so limited in what you can do.

BTW, the second example (of the rose) is at radius 4.

http://www.ddisoftware.com/qimage-u/tech-dfs.htm

P.S. You really have to run a lot of (real) images through an interpolator to get an idea of how well it works: running a single edge through it is really not going to tell you a lot about it.  All the methods have their pros and cons.  Some work better on hard edges.  Some are better at rendering fine "random" detail like leaves or grass at a distance.  Some have problems with circular patterns.  You really have to interpolate a lot of images and pick the interpolator with the best overall balance.  FWIW, the ones that make the sharpest edges often produce results that can look a little "fake" because they break the relationship between sharpness and detail, making some edges much sharper than they really should be.  Finally (sorry, I keep editing to add more because there's a lot involved)... keep in mind that with today's digital cameras, interpolation by super large amounts is less common and nearly all of the interpolation algorithms can do a good job up to about 2x which is likely all you'll ever use.  What is more important is to match the printer's native resolution and to have a sharpening algorithm that can produce the best final output sharpening.  With the high resolution photos of today's cameras, a good sharpening algorithm is more useful than interpolation algorithms, hence the importance of something like DFS that can sharpen without halo artifacts.

Mike

I don't need to.  I've spent a lot of time (many years) sharpening prints without QI, and I have no halos in the final prints, they are sharp and not artificial looking.  If QI makes it easier for folks to get the same results great. 

This really seems more like a case of one guitar amplifier volume knob goes to 10 and the other 11.  Smiley In the QI sample the deep focus sample looks exactly like USM if USM was set to a radius of .6 and and amount of 80.  If anything the USM requires much less aggressive setting to get the same result if we really think the settings mean anything. 

Also, my output is ALWAYS for large prints.  A radius/strength with USM that shows what you would call small halos at 100% views are not visible in the inkjet print.  That's what's most important.  I always edit my image at full resolution (native for my printer) and carefully sharpen images selectively and soft proof the effects before printing.  For me QI would be an unneeded addition to my workflow that offers no real image improvement.  But for users who find this confusing or actually think setting up for a print in PS or Lightroom is complicated, QI seems to be a very good option.

Cheers.

Sal
Logged
mchaney
Newbie
*
Offline Offline

Posts: 15


« Reply #12 on: January 21, 2013, 12:06:48 PM »
ReplyReply

I don't need to.  I've spent a lot of time (many years) sharpening prints without QI, and I have no halos in the final prints, they are sharp and not artificial looking.  If QI makes it easier for folks to get the same results great.  

This really seems more like a case of one guitar amplifier volume knob goes to 10 and the other 11.  Smiley In the QI sample the deep focus sample looks exactly like USM if USM was set to a radius of .6 and and amount of 80.  If anything the USM requires much less aggressive setting to get the same result if we really think the settings mean anything.  

Also, my output is ALWAYS for large prints.  A radius/strength with USM that shows what you would call small halos at 100% views are not visible in the inkjet print.  That's what's most important.  I always edit my image at full resolution (native for my printer) and carefully sharpen images selectively and soft proof the effects before printing.  For me QI would be an unneeded addition to my workflow that offers no real image improvement.  But for users who find this confusing or actually think setting up for a print in PS or Lightroom is complicated, QI seems to be a very good option.

Cheers.

Sal

Good analogy, but completely wrong.  Let's follow that analogy.  It's more like one guitar amp that, internally, is connected to a phone speaker.  Another one that is connected to a 100 watt high fidelity audio speaker.  On the former, you can only turn up the volume to about 1 before it starts to crackle.  On the latter, you can turn the volume all the way up to 10 without distortion.  This analogy is EXACTLY what we are talking about here!  And... you would say "I don't need to hear it... I've been working with an amp connected to a phone speaker for years".  The QU example also had the exact same radius 10, strength 100 and no: it looks nothing like USM at radius .6 and strength 80.  The reason it looks sharp without halos is that the algorithm itself does not produce halos: you can see that in the example because both versions have the same level of sharpening (USM and DFS) with the only difference being the USM version is ruined by halos.  DFS simply does as good a job as USM, but with NO halos whatsoever.  If you prefer halos in your work, then there's not much I can say to convince you.  If you want a much higher quality sharpening algorithm that can sharpen much farther into the details without artifacts, QU has the solution.  Granted, not every photo will need higher levels of sharpening but for those that do, you can do the job without worrying about halos.

Mike
« Last Edit: January 21, 2013, 12:11:12 PM by mchaney » Logged
jrsforums
Sr. Member
****
Offline Offline

Posts: 744


« Reply #13 on: January 21, 2013, 12:16:25 PM »
ReplyReply

I don't need to.  I've spent a lot of time (many years) sharpening prints without QI, and I have no halos in the final prints, they are sharp and not artificial looking.  If QI makes it easier for folks to get the same results great. 

This really seems more like a case of one guitar amplifier volume knob goes to 10 and the other 11.  Smiley In the QI sample the deep focus sample looks exactly like USM if USM was set to a radius of .6 and and amount of 80.  If anything the USM requires much less aggressive setting to get the same result if we really think the settings mean anything. 

Also, my output is ALWAYS for large prints.  A radius/strength with USM that shows what you would call small halos at 100% views are not visible in the inkjet print.  That's what's most important.  I always edit my image at full resolution (native for my printer) and carefully sharpen images selectively and soft proof the effects before printing.  For me QI would be an unneeded addition to my workflow that offers no real image improvement.  But for users who find this confusing or actually think setting up for a print in PS or Lightroom is complicated, QI seems to be a very good option.

Cheers.

Sal

Sal,

I am glad your workflow works for you.

One restrictions of your method is that you must work at "full resolution" for each size print you do.  Redoing it if you change size.

I believe, one of the primary benefits of Lightroom and Qimage printing is that you sharpen for the screen,  until it looks the way you want.  Then, when printing, LR and Qimage interpolate and sharpen to the size selected and resolution need by the printer.  If needed, I can even change cropping and it will be handled.

While LR and Qimage both do it, Qimage does it better.

While I am sure you are an expert at PS output, I did it for years and it is not difficult....just tedious...and really not necessary to get equivalent...maybe better...output with Qimage.

John
Logged

John
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #14 on: January 21, 2013, 12:21:55 PM »
ReplyReply

Good analogy, but completely wrong.  Let's follow that analogy.  It's more like one guitar amp that, internally, is connected to a phone speaker.  Another one that is connected to a 100 watt high fidelity audio speaker.  On the former, you can only turn up the volume to about 1 before it starts to crackle.  On the latter, you can turn the volume all the way up to 10 without distortion.  This analogy is EXACTLY what we are talking about here!  And... you would say "I don't need to hear it... I've been working with an amp connected to a phone speaker for years".  The QU example also had the exact same radius 10, strength 100 and no: it looks nothing like USM at radius .6 and strength 80.  The reason it looks sharp without halos is that the algorithm itself does not produce halos: you can see that in the example because both versions have the same level of sharpening (USM and DFS) with the only difference being the USM version is ruined by halos.  DFS simply does as good a job as USM, but with NO halos whatsoever.  If you prefer halos in your work, then there's not much I can say to convince you.  If you want a much higher quality sharpening algorithm that can sharpen much farther into the details without artifacts, QU has the solution.  Granted, not every photo will need higher levels of sharpening but for those that do, you can do the job without worrying about halos.

Mike

Mike, I have no halos and I sharpen as much as I ever need.  Your analogy is even goofier than mine.  Sorry, I just don't buy it.

Cheers.

Mike, I have no halos and I sharpen as much as I ever need so this is all moot. 

Your analogy is even goofier than mine.  Smiley

Sorry, I just don't buy it, but I'm glad it's helpful for you.

Cheers.

Sal
Logged
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #15 on: January 21, 2013, 12:44:13 PM »
ReplyReply

Sal,

I am glad your workflow works for you.

One restrictions of your method is that you must work at "full resolution" for each size print you do.  Redoing it if you change size.

I believe, one of the primary benefits of Lightroom and Qimage printing is that you sharpen for the screen,  until it looks the way you want.  Then, when printing, LR and Qimage interpolate and sharpen to the size selected and resolution need by the printer.  If needed, I can even change cropping and it will be handled.

While LR and Qimage both do it, Qimage does it better.

While I am sure you are an expert at PS output, I did it for years and it is not difficult....just tedious...and really not necessary to get equivalent...maybe better...output with Qimage.

John

John, that's not a restriction at all, I choose to work at full resolution.  I do a great deal of work on my art prints and use a wide variety of editing tools and plug-ins.   I prefer to have all these editing changes made with real pixels at full resolution rather than editing and up-sampling many layers of work.  I don't want someone else's algorithm deciding how much sharpening I want, in which part of the image, and changing it based on arbitrary print size parameters.  I keep my original RAW files and make my edits at full-res at the largest size I would ever print.  Any thing smaller get's down-sampled (slightly) with no loss of IQ.  All of your examples, again, speak to convenience rather than ultimate print quality assuming knowledgeable printmakers.  If QI can guess better than a specific user's result would be without it, then it's well worth the money for them.  One person's "tedious" might be another person's passion, fun, and ultimate control over the medium.  Thanks.

Sal
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3749


« Reply #16 on: January 21, 2013, 12:49:14 PM »
ReplyReply

You really need to take some time to use Deep Focus Sharpening rather than just making guesses based on web examples.  It really is revolutionary!  The only reason you think a radius of 10 is ridiculous is because you can't do it with USM without ruining the image!  With DFS, you can get the same level of sharpening as USM but you don't have to handicap yourself with tiny radii just because of the halo artifacts.  That opens the door to new levels of sharpening (and de-fogging) because you are not so limited in what you can do.

Hi Mike,

I have to agree that the halo free DFS is a very nice enhancement, especially because it can also be used at the final (native) printer resolution after resampling. So, as long as there are few artifacts in the resampled image data, DFS can not create or make artifacts more visible. Whether one will use the large radii used in the demo examples, is an open question (although it could indeed be used as a haze reducer). I probably would personally limit myself to smaller radii, depending on the degree of upsampling.

"Fusion' resampling does indeed produce very nice results on subject matter at various angles, but for very much (> 3x) larger format output I would probably prefer one of the other algoritms when there are sharp edges and lines in otherwise featureless image areas.

Quote
P.S. You really have to run a lot of (real) images through an interpolator to get an idea of how well it works: running a single edge through it is really not going to tell you a lot about it.  All the methods have their pros and cons.  Some work better on hard edges.  Some are better at rendering fine "random" detail like leaves or grass at a distance.  Some have problems with circular patterns.  You really have to interpolate a lot of images and pick the interpolator with the best overall balance.

I agree, and therefore 'Fusion' produces overall good results for many types of subject matter, but at the same time it is also useful to be aware of the potential risks of one's choices before they happen and escape one's attention. It's just like an important letter/mail you just sent, that spontaneously 'develops' typos, seconds after it's on its way to the destination.

Quote
FWIW, the ones that make the sharpest edges often produce results that can look a little "fake" because they break the relationship between sharpness and detail, making some edges much sharper than they really should be.

Yes, I always warn for the mental disconnect between edge sharpness and material structure detail. A program like PhotoZoom Pro does allow to tune the edge sharpness down, for that reason. Using such a tool also requires a roundtrip to another application (or Plugin if one uses Photoshop for printing), whereas Qimage handles everything (including using the correct settings for e.g. reprints), and for direct printing doesn't require to generate any new images because it resamples on-the-fly.

Quote
Finally (sorry, I keep editing to add more because there's a lot involved)... keep in mind that with today's digital cameras, interpolation by super large amounts is less common and nearly all of the interpolation algorithms can do a good job up to about 2x which is likely all you'll ever use.  What is more important is to match the printer's native resolution and to have a sharpening algorithm that can produce the best final output sharpening.  With the high resolution photos of today's cameras, a good sharpening algorithm is more useful than interpolation algorithms, hence the importance of something like DFS that can sharpen without halo artifacts.

Indeed, the importance of resampling to the native printer resolution is major, and being able to automatically add halo free output sharpening at that final resolution is great. Thanks for your great software solution, it's a real productivity tool which also produces output quality that's hard to beat.

Cheers,
Bart
Logged
jrsforums
Sr. Member
****
Offline Offline

Posts: 744


« Reply #17 on: January 21, 2013, 12:58:41 PM »
ReplyReply

John, that's not a restriction at all, I choose to work at full resolution.  I do a great deal of work on my art prints and use a wide variety of editing tools and plug-ins.   I prefer to have all these editing changes made with real pixels at full resolution rather than editing and up-sampling many layers of work.  I don't want someone else's algorithm deciding how much sharpening I want, in which part of the image, and changing it based on arbitrary print size parameters.  I keep my original RAW files and make my edits at full-res at the largest size I would ever print.  Any thing smaller get's down-sampled (slightly) with no loss of IQ.  All of your examples, again, speak to convenience rather than ultimate print quality assuming knowledgeable printmakers.  If QI can guess better than a specific user's result would be without it, then it's well worth the money for them.  One person's "tedious" might be another person's passion, fun, and ultimate control over the medium.  Thanks.

Sal

We all make compromises.  One of yours is to down sample rather than redo at specific output size.....I guess for "convenience".  I assume you are satisfied with this "guess" vs specific results.

Look....I have no doubt you are an excellent printer, with years of experience behind you.  You guess....with an experienced eye....at how much sharpening is needed for a given print at a given print size.  But, I am sure, even you are not absolutely sure until you actually print...every "knowledgeable printmaker" I know has said this.

Many, such as you, seem to love the process as much, maybe more, than the result.  I do not think, however, that you should look down at those who can get just as good (maybe, maybe not, better) results, but with a much easier process.  Just because we don't love your process does not mean we want any less in ultimate print quality.

John
Logged

John
mchaney
Newbie
*
Offline Offline

Posts: 15


« Reply #18 on: January 21, 2013, 01:04:15 PM »
ReplyReply

Mike, I have no halos and I sharpen as much as I ever need so this is all moot. 

Your analogy is even goofier than mine.  Smiley

Sorry, I just don't buy it, but I'm glad it's helpful for you.

Cheers.

Sal

There's plenty of validity to "use whatever works" so I understand and I won't dispute what works for you.  I'll just make these final thoughts for you to consider.

The fact remains that DFS can simply be pushed a lot harder without artifacts.  Some may need that ability.  Others (like you) maybe not.  But if you don't, it's probably only because you've learned to work within the limitations of USM.  I'm sure, like the rest of us, there are times where you would like to use a larger radius but you don't because you know you can't without causing USM halos.  The larger radius is very useful for that occasional soft photo, landscape where you want high contrast, image where you want to remove fog with a large radius sharpening, or a very large print where you need larger radii based on image resolution versus print size (something that, as John pointed out, Qimage does automatically or, as you may prefer, be adjusted manually).

But again, to my analogy of the speakers.  We're looking at a technology in DFS that allows you to push harder, just like a real audio speaker can be pushed harder than a phone earpiece.  If all you've ever heard is a phone speaker, you may not think you need anything better and you'll work within those limitations which is why I suggested just trying it rather than guessing what it can do.  I know what it can do.  You can put any radius/strength that you like into USM, even smaller ones, and as soon as you switch to DFS, the sharpening remains nearly identical except when you click DFS, the halos disappear.  When presented with two algorithms, one with an obvious flaw (USM = halos) and another without (DFS = halos completely gone), I'm not sure why anyone would want to use the inferior one.  And once you see the difference and you can switch back and forth between USM and DFS, you begin to see that even radius values in the .5 to 1.0 range have artifacts (not necessarily all halos as pointed out below) because when you switch to DFS, you can see the same level of sharpening without the brightening around edges and even inside "holes".  Even radius .5 to 1.0 can cause specular highlights in small areas where, for example, light shining through tiny holes in/between leaves can turn what should be a blue point of light (blue sky behind the leaves) into white.  After working with hundreds of images under both methods, you begin to spot a lot of deficiencies like this (even though some may be small) in the USM algorithm.

Mike
Logged
Sal Baker
Full Member
***
Offline Offline

Posts: 140


« Reply #19 on: January 21, 2013, 01:22:42 PM »
ReplyReply

We all make compromises.  One of yours is to down sample rather than redo at specific output size.....I guess for "convenience".  I assume you are satisfied with this "guess" vs specific results.

Look....I have no doubt you are an excellent printer, with years of experience behind you.  You guess....with an experienced eye....at how much sharpening is needed for a given print at a given print size.  But, I am sure, even you are not absolutely sure until you actually print...every "knowledgeable printmaker" I know has said this.

Many, such as you, seem to love the process as much, maybe more, than the result.  I do not think, however, that you should look down at those who can get just as good (maybe, maybe not, better) results, but with a much easier process.  Just because we don't love your process does not mean we want any less in ultimate print quality.

John
John, you're making a lot of assumptions.  Of course I guess at getting optimum quality, this is an art not a science.  I only sharpen at the last stage before printing, after downsampling if I need an occasional small print.  I save my selection mask which follows any size change.  We all learn as we go along but I don't want a software guessing for me.  I rarely need to print smaller, but if the size change is dramatic I'll sometimes make completely different decisions on the which areas receive sharpening, not just the amount.  Why would I not want this flexibility?

Where did I "look down" on anyone?  You seem to be looking down on me because I simply say I can get as good of results without QI.  I never said QI gives poorer results, I clearly said I don't need it as it would get in the way of my workflow and I have no halo or sharpness problems that need fixing, but thanks for the suggestions. 

Sal
« Last Edit: January 21, 2013, 01:30:02 PM by Sal Baker » Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad