Pages: 1 [2] 3   Go Down

Author Topic: resizing images  (Read 24107 times)

Stephen Best

  • Guest
resizing images
« Reply #20 on: February 16, 2006, 07:11:01 am »

Quote
Stephen,
I am sure you will see a discernable difference. What I also urge you to try is leaving your test file at 360ppi and sizing the print job using print with preview in Photoshop and then doing the same with Qimage. I am fairly certain you will be impressed by the difference.
[a href=\"index.php?act=findpost&pid=58274\"][{POST_SNAPBACK}][/a]

I have a Mac though I did have a brief look at Qimage in Virtual PC. Most of my own prints are downsampled from larger files, and if upsampled (to 24x30) only by about 5%. The only time I use Print with Preview to resize is for quick and nasties :-).

One practical application of the difference between 240ppi and 360ppi is a single pixel border I add to my own images (part of the pre-print Action). I'm after the lightest/thinest border I can get to just hold in the image (sky areas etc).
Logged

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #21 on: February 16, 2006, 07:30:30 am »

Stephen,
Returning to Qimage again, there is a feature which automatically turns on a very thin black border all round the image which I use in exactly the same way.
Logged
Andrew Richards [url=http://www.andrewri

Stephen Best

  • Guest
resizing images
« Reply #22 on: February 16, 2006, 07:59:35 am »

Quote
Stephen,
Returning to Qimage again, there is a feature which automatically turns on a very thin black border all round the image which I use in exactly the same way.
[a href=\"index.php?act=findpost&pid=58278\"][{POST_SNAPBACK}][/a]

Ack! You mean everybody's doing this? :-).

I'm surprised not to see some photos of North Wales on your site. Some lovely country there.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
resizing images
« Reply #23 on: February 16, 2006, 08:11:21 am »

Everything printed from Qimage, big or small, is at least 360ppi, isn't it? In fact, sometimes an image that is already 360ppi will be uprezzed to 720ppi by Qimage. Not sure why. Possibly because it's actually 360.1ppi and Qimage only uprezzes.

Maybe there's some adjustment in some preferences option I've missed, but it's never been of much concern to me. A bit of pixel-peeping is fine, but trying to find any meaningful difference between 240 and 360ppi, or even between GF and bicubic in normal size prints that haven't been interpolated by a really huge degree, is pixel-peeping carried to the extreme.
« Last Edit: February 16, 2006, 08:24:19 am by Ray »
Logged

Dale_Cotton

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 588
    • http://daystarvisions.com
resizing images
« Reply #24 on: February 16, 2006, 11:02:43 am »

Quote
Everything printed from Qimage, big or small, is at least 360ppi, isn't it? In fact, sometimes an image that is already 360ppi will be uprezzed to 720ppi by Qimage. Not sure why. Possibly because it's actually 360.1ppi and Qimage only uprezzes.[{POST_SNAPBACK}][/a]
FYI: [a href=\"http://www.ddisoftware.com/qimage/quality/]QImage ppi handling[/url].
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #25 on: February 16, 2006, 12:17:24 pm »

Quote
..........trying to find any meaningful difference between 240 and 360ppi, in normal size prints that haven't been interpolated by a really huge degree, is pixel-peeping carried to the extreme.
[a href=\"index.php?act=findpost&pid=58281\"][{POST_SNAPBACK}][/a]

I don't know what size range you include in "normal", but if normal includes A4 and A3, based on the extensive amount of printing I've done of all kinds of photographs at various resolutions from 180 to 451 on an Epson 4000 and then an Epson 4800, I agree with you. However, it matters whether one is discussing resizing or resampling. For resizing, looking for quality differences accross a range of 240 to 360 or 360 and above is largely unproductive pixel-peeping. However, if one is talking resampling by large amounts from low starting points, methodology and tools matter. In fact, highly-regarded professionals who have tested this stuff extensively - be it for production, educational or software development purposes - generally recommend to avoid resampling unless there is absolutely no choice, and then to minimize it.

I would at the same time add some comment on the question of the Printer driver's "native resolution", because this is not as straightforward a concept as one would like to believe it should be, hence it is simplistic to theorize that converging an image's PPI with the "native resolution" of the printer, or using integer divisors is somehow superior to not doing so.

Based on what I have learned through discussion with professionals who have deep knowledge of these matters, when one talks about an inkjet printer's "native resolution" of say 360, inkjet printers lay down ink dots, not pixels. The print head itself has ink nozzles 1/360th of an inch apart. The 720, 1440, and 2880 numbers are obtained by the stepper motor moving the printhead by those smaller increments and the printer's weaving algorythms. Through processes of dithering that are corporate secrets of the printer manufacturers, pixels get converted into ink dots. Basically, these processes overlay the image pixels on a 360-cell-per-inch grid, and for each cell, calculates how to turn the pixels into dots of ink.

What all this really means is that there is no choice but to simply use one's eyes, look at prints made at various PPI and printer settings on various media - and see within the range of stuff one normally does whether any of this (short of major resampling) makes a serious difference. My own experience mentioned above suggests that it rarely does, so I am back to Ray's point about pixel-peeping.
« Last Edit: February 16, 2006, 12:19:29 pm by MarkDS »
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Lisa Nikodym

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1705
    • http://www.stanford.edu/~melkor/lisa_pictures/lisa_pictures.html
resizing images
« Reply #26 on: February 16, 2006, 12:55:21 pm »

Quote
QUOTE(nniko @ Feb 15 2006, 11:12 PM)
There is no discernable difference between 240 ppi and 360 ppi.



Try this. Create a new image at 240ppi, white canvas, draw a 300x300 black pixel square and add some text in the middle, say Helvetica 3pt. Create a second image, this time at 360ppi, draw a 450x450 pixel square (same corresponding size) and add the same text, again at 3pt. Print both out on Enhanced Matte (or similar) at 1440dpi unidirectional. Compare. My eyes aren't great these days but I can see a difference both in the weight of box and the resolution of the text. Without a loupe. The same would apply for any micro details in your files ... assuming the information is there in the first place.

My experiments weren't with text and drawn boxes, because that's not what I (and the vast majority of others here) are printing; my experiments were with a typical landscape photo.  I'm not surprised that you might see a difference between 240 & 360 when you're looking at printed text, but that's pixel-peeping.  For a normal landscape photo, I saw no difference, and that's what matters to me.

Quote
I would at the same time add some comment on the question of the Printer driver's "native resolution", because this is not as straightforward a concept as one would like to believe it should be, hence it is simplistic to theorize that converging an image's PPI with the "native resolution" of the printer, or using integer divisors is somehow superior to not doing so.

I never made any claims about any theory behind the "printer native resolution" issue.  I don't give a !@#$ about theory, only about results.   All I did was report that, in practice, with a typical landscape photo, it *did* make a difference (a quite small one, but more noticeable than the 240/360 issue).

Perhaps we are making mountains out of molehills, arguing about whether something is "small" vs. "none", when, in reality, it's not worth the time worrying about it one way or the other.

Lisa
Logged
[url=http://www.stanford.edu/~melkor/lis

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #27 on: February 16, 2006, 01:21:32 pm »

Quote
I never made any claims about any theory behind the "printer native resolution" issue.  I don't give a !@#$ about theory, only about results.   All I did was report that, in practice, with a typical landscape photo, it *did* make a difference (a quite small one, but more noticeable than the 240/360 issue).

Perhaps we are making mountains out of molehills, arguing about whether something is "small" vs. "none", when, in reality, it's not worth the time worrying about it one way or the other.

Lisa
[a href=\"index.php?act=findpost&pid=58313\"][{POST_SNAPBACK}][/a]

I wasn't talking theory either - I was talking about the practical realities of how printers work, in order to make two points on which we both seem to be agreed: (i) what matters is what normal photographs look like on paper, and (ii) the quality differences we're talking about here aren't worth the time worrying about.
« Last Edit: February 16, 2006, 01:22:19 pm by MarkDS »
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #28 on: February 16, 2006, 02:19:28 pm »

Dale,
Thanks for pointing out that article on Qimage. I do remember reading it the first time, just before I gave it a try. I did not give it much notice at the time because Qimage is so cheap, I just was not prepared to give much credence to the claims. However, I am a total convert now and I really think that Photoshop is a poor application to print out of. Not convinced? Let me show you a real-world photograph:

That is the whole image. Now two identically sized A4 prints and a sectional flat-bed scan of each, the first sized in PS CS2 print with preview. Remember, both prints are made from the same uncropped 16-bit Tiff from a 1DS MKII at 300ppi, so plenty of resolution for a good A4 print. Both also made with the same output profile. I could also show you examples from prints at native resolutions (i.e. no adjustment of size to fit the print) and the same sorts of problems are manifest, they just happen to be particularly obvious in this example.

Next sized in Qimage.

Note how the Photoshop print is less sharp and displays very obvious artefacts.
Personally, resizing all the time for printing is just a pain in the derriere. Leave the original images in their original highest quality state, do the photo-editing in Photoshop and the printing (at any size) out of Qimage.
Logged
Andrew Richards [url=http://www.andrewri

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #29 on: February 16, 2006, 04:36:26 pm »

Drew, thanks for sharing this. It is interesting. I have two points about it though:

(1) One wonders why, with all the brilliant minds at Adobe, and Photoshop having gone through 9 versions over the past 20 years, they have not created resampling algorythms at least as good as QImage's. Perhaps there is no real explanation - just shows there will always be a niche for brilliant alternatives outside the four walls of the big corporations. Same reason why PixelGenius has sharpening algorythms that far surpass anything in Photoshop.

(2) On my monitor your full image is about 68 square inches. The blow-ups you then show are approximately 3 square inches of the original image. So there is an enlargement ratio of about 22.7x. That makes me wonder whether one would see any real difference of results between the two resampling tools on the full 68 sq.in. photograph without a loupe be it on the monitor or printed.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #30 on: February 16, 2006, 05:44:44 pm »

Mark,
In response to your two specific questions I would say firstly, that after nine versions of Photoshop, it is still far from perfect. Adobe Bridge is a huge improvement over the file browser in CS, but that really is not saying much. The file browser in CS was completely rubbish. I must have cussed endlessly while it cached thumbnails for the first time and tied up Photoshop CS itself. There is no way it can have been properly tested with digital files of any sort of decent resolution. Even with Bridge, I find the handling of IPTC data clunky and on my sytems, it is a bit crash prone. I think the handling of printing is poor. I am not impressed by the 32 bit high dynamic range feature. I still get results more pleasing to me with layers and masks. The picture package feature again is absolutely not a patch on Qimage. Photomerge is a joke, especially if you compare it with the venerable Panotools with PTGui as the graphical front end. PTGui (was that Sergio's recommendation?) is absolutely brilliant. Very accurate and almost perfectly blended 16bit  panoramas. I am also concerned that Adobe will take its eye off the ball needlessly developing a product to compete with Aperture (I am referring to Lightroom of course), when there is still a lot that could be improved in its main product for photographers i.e. Photoshop.
Secondly, my full image is not at all representative of the original file in that it has been downsampled and converted to sRGB for web display. The flatbed scans are just that, selective scans of portions of A4 inkjet prints of the full image in Adobe RGB 1998. If I were to make selective crops of the full size file, you would of course see much more detail, as below:
Logged
Andrew Richards [url=http://www.andrewri

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #31 on: February 16, 2006, 06:02:40 pm »

Drew, I agree with all that - added to which Bridge is also clunky and slows Photoshop. But I have to tell you, I went to a Lightroom Seminar yesterday afternoon here in Toronto given by Michael Reichmann - it was a three hour session on that program, and the program is GOOD. I'm on Windows so I can't use it yet, and it is still under development, but if you are on Mac, download it and try it. Miles ahead of Bridge, fast, easy, and it has some tools that are so cool they will probably end-up in future versions of Photoshop.

But on my second point, yes, that image you pointed me to is crystal clear - very good, no argument about it. But the question I was asking is different. If you were to re-rez the image in Photoshop (using the Image Size dialogue box, not Print with Preview) and then go back to the same file in its previous state and repeat the exercise in QImage, would one see a difference between the two full frame A4 prints with no further magnification? To put a finer point on it, I think if the term ever got into a dictionary, the dictionary definition of pixel-peeping would be where one is looking for differences between pixels under conditions that one would not normally look at the photographs. I don't look at photographs with a loupe or at 22x magnification. I just look at full photos, at normal viewing distance for the size of the image. So under these conditions, have you tested for and do you see an eye-popping difference between the same complete image (not a crop) at A3 or A4 size resampled with Photoshop "Image Size" tool versus resampled with QImage?
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Stephen Best

  • Guest
resizing images
« Reply #32 on: February 16, 2006, 06:48:04 pm »

Quote
My experiments weren't with text and drawn boxes, because that's not what I (and the vast majority of others here) are printing; my experiments were with a typical landscape photo.  I'm not surprised that you might see a difference between 240 & 360 when you're looking at printed text, but that's pixel-peeping.  For a normal landscape photo, I saw no difference, and that's what matters to me.

OK, if the printer is physically capable of resolving this difference to the naked eye (with the simple test I gave) and you're not seeing any differences from your D70 (or whatever) files, what  does this tell you ... other than it probably makes little difference with your typical files/workflow?
« Last Edit: February 16, 2006, 06:49:08 pm by Stephen Best »
Logged

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #33 on: February 16, 2006, 07:36:08 pm »

Mark,
I will do as you suggest and give Lightroom a spin.
What I have done is make another A3+ print from the original file using print with preview (so that the correct profile can be applied) and no scaling of the print. In other words a 300ppi image sent to the printer without any change in image size. I have compared this with the same file sent to the printer for an A3+ print using Qimage. I have  scaled the image slightly in Qimage to fit the media better, so that the final printed image is actually slightly larger than the Photoshop print. Result? The Photoshop print still shows artefacts, admittedly slightly less severe than on the A4 print, but definitely still there. The Qimage print is perfect. Are the artefacts visible to the naked eye without the need of a loupe. Absolutely, yes
Logged
Andrew Richards [url=http://www.andrewri

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #34 on: February 16, 2006, 08:02:34 pm »

Andrew, it is good of you to do that in aid of a discussion forum. I should do more of this testing myself, but what holds me back is the enormous amount of printing I've just completed over the past few months from a major photo-shoot, resizing my images to A4 or A3 in Photoshop (which almost always means reducing linear dimensions without resampling pixels, i.e. I'm resizing pixels rather than adding or subtracting them), then printing the images using the Epson driver and profile; honestly, I'm getting such clear, clean results that I'm having trouble relating to the problem, notwithstanding all the obviously objective evidence you have been producing. As I've mentioned above, there are definitely issues to deal with when severely cropping a 1Ds (1st or MKII) image and then trying to expand the remainder to A3, for example starting from something like 120PPI and aiming at 240 or more. Anyhow, this has been a very interesting discussion, and there is food for thought here in case of real tough nuts to crack out of the printer.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

jani

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1624
    • Øyet
resizing images
« Reply #35 on: February 17, 2006, 05:32:06 am »

Perhaps if Andrew would tell us which printer he has, we could see if there's a difference in the printers Mark, Lisa and Andrew use?
Logged
Jan

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #36 on: February 17, 2006, 05:46:55 am »

Jani,
I used an Epson 7600 to make the prints in this instance, but I could also have shown examples from an Epson 4000 and a 2100 converted for black and white printing. I think there must be some problem with the handling of certain sizes of image out of photoshop, because the prints scaled up out of print with preview from film scans, which have small linear dimensions, but very high resolution look fine to me. There are no such issues with Qimage and I obviously like the very consistent way it handles print jobs.
Logged
Andrew Richards [url=http://www.andrewri

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
resizing images
« Reply #37 on: February 17, 2006, 08:48:09 am »

Jan, Andrew,

In the class of printers we're all using, the printer model clearly isn't a relevant factor - it is most likely the resampling algorythm. Andrew, I think part of the problem using Print with Preview to resample is that it can only use the default resampling algorythm that you specified in your Photoshop Preferences (General Tab, Interpolation). Hence if your preference there is specified as "Bicubic" you are depriving yourself of the technical advantages of Bicubic Smoother and Bicubic Sharper, which you can independently select according to specific image requirements if you were to activate your resampling using the "Image Size" dialogue box instead of "Print with Preview". (Likewise, by the way, "Transform" functions in Photoshop use only the default interpolation method specified in General Preferences.) Adobe developed the Smoother and Sharper variants to address limitations of Bicubic alone. This may partly explain the quality difference you see between Photoshop and QImage.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

jani

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1624
    • Øyet
resizing images
« Reply #38 on: February 17, 2006, 09:33:44 am »

Quote
Jan, Andrew,

In the class of printers we're all using, the printer model clearly isn't a relevant factor
Well, it could be that the printer drivers were significantly better for a new printer model as compared to a very old one; since Andrew didn't state what he used, and the qimage website refers to Photoshop 6, I was considering the possibility that there might be differences related to that.

That appears not to be the case.
Logged
Jan

drew

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 477
    • http://www.andrewrichards.net
resizing images
« Reply #39 on: February 17, 2006, 09:43:27 am »

Mark,
I really do not know if Photoshop actually does resample out of print with preview or if it just does what the Qimage web info page says it does, which is just to chuck the job over to the Epson print driver and get it to fit the page. Either way it still does not explain what is happening when the image is sent to the printer without any change in size. Also, I find it very hard to believe that there really is any big difference between bicubic and bicubic smoother and bicubic sharper, certainly not enough to explain what I am seeing. Also, I take what you say about setting up in preferences, but often my printing is as much about scaling up as scaling down. As the thread has already indicated, sometines you will see these problems and sometimes you will not. Generally I find them whenever there is a very sharply defined and smooth line or curve, but in addition, having found this problem, I also find that the Qimage prints are very subtley better in other respects. Unless you have the two different versions in front of you, a lot of the time you will be satisfied with what photoshop does, especially if it is a landscape shot.
On the other side of the coin, if you are a PC user, a Qimage license is peanuts. Even if you are a Mac user, you can buy a PC with plenty of memory capable of running Qimage, network it to your Mac and run it through the switchable input of your Mac monitor and still have change out of the cost of a full photoshop license. If you subtract the cost of virtual PC, it would be even less.
Logged
Andrew Richards [url=http://www.andrewri
Pages: 1 [2] 3   Go Up