DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Tips, Tricks, and Q&A >> Resolution vs Quality
Pages:  
Showing posts 1 - 25 of 42, (reverse)
AuthorThread
04/05/2004 02:19:07 AM · #1
theoretically speaking...(i.e. made-up resolutions here)

if i took a photo at 100*100 and later shrunk it to 10*10 (thus keepign same rati0), would the quality of the shrunk photo be worse, same or better than if I had originally taken the picture at 10*10?

And what software is good at shrinking pictures?

thanks
04/05/2004 02:24:05 AM · #2
taking a photo at a given size will always be better than taking a larger photo and then shrinking it in software. photoshop is the defacto for this and everything else to do with photo editing.

when you shrink an image, the software has to guess which pixels to disguard in order to maintain the content of the image in the reduction. photoshop is really good at guessing this, to the point where 99 times out of 100 you wouldn't be able to tell the difference. but since taking a photo at that size guarentees you that the pixels you need will be there, no guessing involved, it is always the surefire method.
04/05/2004 02:28:44 AM · #3
darcy is correct, but as stated, you will not notice the difference, so you should always shoot in the highest quality available to you.
04/05/2004 02:52:05 AM · #4
I disagree ... I think shooting in the highest resolution and then letting good software scale the photo will be better than shooting at lower resolution in the first place. If you shoot at a lower resolution than your camera is capable of shooting, you're just asking your on-camera software to average/figure it out for you and that software is probably worse than specialized software in Photoshop or some other photo editing software that has multiple algorithms to run. Moreover, shooting at the full resolution of your camera gives you a high res photo to keep for printing and you can scale a copy down for posting on the web.
04/05/2004 03:04:38 AM · #5
i could be wrong, but i don't think digital camera's take all photos at their max resolution and then downscale using some software algorithm to the resolution you selected. i always assumed it just grabbed the shot with the number of pixels you asked for.

what you're describing sounds like the opposite of digital zoom. if that were the case, it would be terrible! i hope that's not the case.

i agree that shooting at the highest possible resolution is the best bet for ensuring maximum usability of the photo. that way, you can use it to downscale in software and post on the web, but also print and hang on your wall. still, technically speaking i think if you wanted the best quality and only wanted it small, shooting small is better than scaling small - technically speaking of course.
04/05/2004 03:09:17 AM · #6
But if I use photoshop and there's only 1% difference between the two then it's really to my advantage to take the bigger photos right?

And is photoshop elements a 'shrunk' version of Photoshop?
04/05/2004 05:28:25 AM · #7
Originally posted by Delphi00:

But if I use photoshop and there's only 1% difference between the two then it's really to my advantage to take the bigger photos right?

And is photoshop elements a 'shrunk' version of Photoshop?


Elements is the baby version of PS7. Most of the extra things in PS7 that Elements doesn't have can be done with a little extra work.

If you plan to print photos always use the highest settings that your camera can handle. If space is a problem try to buy more memory. If you need to manipulate a photo or fix something it's to your benefit to have a larger/bigger file to work from.
04/05/2004 05:50:14 AM · #8
I've noticed a strange quirk in resizing... I've tried making photos smaller in Photoshop using every available resizing method, and have found that IrfanView will consistenly resize better with less obvious pixellation. I use IrfanView now for all my resizing.
04/05/2004 06:53:12 AM · #9
Originally posted by darcy:

i could be wrong, but i don't think digital camera's take all photos at their max resolution and then downscale using some software algorithm to the resolution you selected. i always assumed it just grabbed the shot with the number of pixels you asked for.



It's the CCD/CMOS that captures the pixels, and the physical number of pixels is always the same. So to get a image in lower resolution the camera must calculate pixels from several physical pixels from the CCD/CMOS
04/05/2004 09:56:37 AM · #10
Originally posted by BobsterLobster:

I've noticed a strange quirk in resizing... I've tried making photos smaller in Photoshop using every available resizing method, and have found that IrfanView will consistenly resize better with less obvious pixellation. I use IrfanView now for all my resizing.


Have you tried the new bicubic algorithms in photoshop CS ?

Used to be I would resize then sharpen again, but now with the bicubic sharper option, I haven't been bothering.

Lanczos interpolation is pretty good too - but it isn't available in CS. (QImage) Not sure what Infranview uses.
04/05/2004 10:21:05 AM · #11
Originally posted by Gordon:

Originally posted by BobsterLobster:

I've noticed a strange quirk in resizing... I've tried making photos smaller in Photoshop using every available resizing method, and have found that IrfanView will consistenly resize better with less obvious pixellation. I use IrfanView now for all my resizing.


Have you tried the new bicubic algorithms in photoshop CS ?

Used to be I would resize then sharpen again, but now with the bicubic sharper option, I haven't been bothering.

Lanczos interpolation is pretty good too - but it isn't available in CS. (QImage) Not sure what Infranview uses.


IrfanView 3.85 has multiple filters for resampling: Lanczos, Triangle, Mitchell, Bell, B-Spline & Hermite. Lanczos seems to work best. Irfanview also makes a relatively fast image browser and is free :-)
Has someone compared the GIMP with Photoshop fro resizing?

04/05/2004 10:30:49 AM · #12
My personal experience is as follows. I have found that I get the best results from shooting in the highest resolution possible then scaling down. I have found that I get the best results scaling down when I scale down the image in small increments then sharpen a small amount in the end. Scaling down seems to preserve more of the major detail in the image but usually eliminates the high ISO noise.

Greg
04/05/2004 10:55:50 AM · #13
Originally posted by Delphi00:

theoretically speaking...(i.e. made-up resolutions here)

if i took a photo at 100*100 and later shrunk it to 10*10 (thus keepign same rati0), would the quality of the shrunk photo be worse, same or better than if I had originally taken the picture at 10*10?

And what software is good at shrinking pictures?

thanks


That very much depends, if both pictures are taken with the same camera, then using software interpolation would be better. If the two pictures are taken with a different camera then not using interpolation would probably be better. In all this you should not forget however that most digital cameras a limited by the lenses.

When you use smaller than full resolution, your camera will re-sample the image using its inbuilt algorithms. This will most likely be nearest neighbour interpolation -- very poor.
One of the problems is that even the best heuristics algorithms are always slightly wrong, as they have to guess. The main thing is that the processing power of your computer is so much greater that using your computer will almost always provide a significantly better image.
04/05/2004 12:58:23 PM · #14
i'm glad people who actually know something about this stepped in! ;)

i always assumed that the presets in a digital camera were there because they were actual resolutions that it could record in. knowing now that the camera interpolates to produce lower resolutions, i would have to join the side saying do it in software. photoshop cs has some great new interpolation features.

if, however, you had the choice between a camera with a max resolution of 800x600 and one that shot at a max of 1600x1200, and you spefically wanted an 800x600 image... assuming both cameras had the same lenses and such, you would technically get the best quality from the lower resolution camera.
04/05/2004 02:16:01 PM · #15
I used to do stair/ step interpolation for enlarging, but don't even bother doing that now, with the new algorithms in CS. I've heard about people doing it for shrinking images, but I've never seen any samples that showed improvment doing it.

I use bicubic sharper for reduction and bicubic softer for enlargment, both single step to the final size and haven't had any complaints about the print quality I've had from these files.
04/05/2004 02:52:00 PM · #16
Originally posted by Gordon:

I've heard about people doing it for shrinking images, but I've never seen any samples that showed improvment doing it.


The first experience I had with this problem was on my Transportation entry:
' . substr('//images.dpchallenge.com/images_challenge/91/thumb/20057.jpg', strrpos('//images.dpchallenge.com/images_challenge/91/thumb/20057.jpg', '/') + 1) . '

The grass in this shot, after resizing straight to the final resolution, showed an excessive amount of jaggedness(?). It was very frustrating, and one of those times where I was submitted in the last hour and didn't have time to find a solution. Later, I figured or found out (don't remember which) that step resizing in smaller increments made a huge difference. Here's a stepped-down version of that same shot, resized in 10% increments:
' . substr('//images.dpchallenge.com/images_portfolio/7151/thumb/68386.jpg', strrpos('//images.dpchallenge.com/images_portfolio/7151/thumb/68386.jpg', '/') + 1) . '

There's still a little bit in some spots, but you can see a huge difference in the blade towards the middle of the shot that bends back towards the upper right.

There may be better solutions (and some day I may be able to look into CS), but this usually works pretty well. I have an action set up that resizes the current image by 10%, so It's just a few clicks of key till I get the right size.
04/05/2004 02:57:53 PM · #17
Confused - did you post the right samples ? They look like very differently processed files, with really strong halo'ing from sharpening in the step resized case ?
04/05/2004 07:36:17 PM · #18
Yes, the right sample, but both were created some time ago, and you're right, they definitely weren't processed the same, and I don't really remember the difference at this point (but at the very least, different levels and sharpening). Fortunately, I had the original available, so I opened it up and just took the straight image and resized it both ways - no edits, no sharpening. Here are the new results:

The Coaster - Direct Resize, No Edits
' . substr('//images.dpchallenge.com/images_portfolio/7151/thumb/68426.jpg', strrpos('//images.dpchallenge.com/images_portfolio/7151/thumb/68426.jpg', '/') + 1) . '

The Coaster - Stepped Resize, No Edits
' . substr('//images.dpchallenge.com/images_portfolio/7151/thumb/68427.jpg', strrpos('//images.dpchallenge.com/images_portfolio/7151/thumb/68427.jpg', '/') + 1) . '

Message edited by author 2004-04-05 19:36:41.
04/05/2004 07:56:57 PM · #19
VERY IMPRESSIVE!! This is no trivial technique, but something well worth learning. Reaffirms my faith in user-based forums!
Thanks for sharing!
04/05/2004 08:14:07 PM · #20
Honestly, I'd still say the stepped resized one looks a lot worse, with still pretty strong haloing around the edges of the reeds, particularly on the diagonals. But it is good to find out what works for you and do the practical tests.

Message edited by author 2004-04-05 20:14:40.
04/05/2004 08:21:13 PM · #21
Well, it has halos, but not jaggies, and I'd personally have small halos than jaggies.

I never realized what a difference resizing in steps could make!

Message edited by author 2004-04-05 20:21:26.
04/05/2004 08:46:46 PM · #22
The halos look a lot like unsharp masking artifacts and are present in both images, just a lot more noticeable in the multi-step image.
Anyone have a cure for the halos?
04/05/2004 09:42:31 PM · #23
Originally posted by ElGordo:

The halos look a lot like unsharp masking artifacts and are present in both images, just a lot more noticeable in the multi-step image.
Anyone have a cure for the halos?


Basically like you said - don't sharpen so much.
04/05/2004 10:11:33 PM · #24
ScottK, I like the way the step resizing eliminated the jaggies, but I also think the haloing is substantial. Not quite sure why that is. Did you sharpen/unsharp mask both the same and only once at the end? Alternatively, did you save the stepped resize image as .jpg in between any of the steps? If so, the halos may be from the jpg recompression algorithm being used each time you saved.
04/06/2004 02:25:09 AM · #25
Interesting. I totally don't see the haloing, either at work or at home, as displayed. I took both of these and loaded into PS and blew them up to 800%, expecting to be justified in not seeing any evidence - but there it was. The only things I can think to explain it is either monitors (I have same brand, similar models in both places) or my being color blind. I did, however, see it on the first stepped resize I posted above, but that had probably been sharpened before and/or after resizing.

ElGordo got it right in that it is on both images, just more on the stepped one. Also, I blew up the original, full sized image, and the halos are there as well. So maybe this is just a good image to demonstrate the technique with the jaggies, but happens to be a bad image as far as haloing.

And there was no PS sharpening either before or after resizing, just what was done by the camera. I just tried sharpening both of them a little, and on the one hand it seems to "equalize" the effect between the two versions, and the haloing does become just slightly noticable to me at regular magnification.

I'm going to play a little more with it - maybe a different resampling algorythm will work better? I'll let you know if I figure a way around the haloing. Also, I'll see if I can find some other images that don't have halos to start with, and see how they work.

Message edited by author 2004-04-06 02:34:45.
Pages:  
Current Server Time: 12/01/2020 02:18:47 PM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Prints! - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2020 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Proudly hosted by Sargasso Networks. Current Server Time: 12/01/2020 02:18:47 PM EST.