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

DPChallenge Forums >> Tips, Tricks, and Q&A >> How Destructive is Rotate?
Pages:  
Showing posts 1 - 16 of 16, (reverse)
AuthorThread
 06/03/2008 10:03:10 PM · #1
Obviously rotations which are 90 degree (or multiple thereof) are not at all destructive, but I wonder about the mapping algorithms that "rotate" pixels. I would think a one degree rotation more destructive than a 45 degree rotation. I go through great pains to not have to rotate. Maybe I'm crazy.

Just wondering if any discussion or research exists on the topic.
 06/03/2008 10:06:24 PM · #2
hmmm good question. Will be interested in the opinions. Personally I've never experienced quality degredation due to rotation - but that is starting from a high-res RAW file. Its been ages since I tried a major rotation on jpg shot.

06/03/2008 10:12:33 PM · #3
On large files i dont think it matters too much - but i know on smaller files it definitely deteriorates the sharpness of the shot. Needless to say i'd probably do any (minor) rotations before starting editing...
 06/03/2008 10:21:26 PM · #4
It really depends on the content of the image. It is definitely destructive, and mostly shows up on patterns with lots of thin, straight lines - especially with structures/cityscapes. Having said that, and now looking at your profile... you are apparently the person it will effect the most :o\

As inshaala mentioned - I'd definitely rotate prior to resizing down.
 06/03/2008 10:29:40 PM · #5
hmmm I've never even thought about that but I assume it could definately have negative effects. I rotate my images a lot to get perfect allignment but now I'm going to be concerned about how it affects the image. I'de love to get a definative answer on how much it degrades the image and if's better to rotate a cartain number of degrees instad of random fractions of a degree.
 06/03/2008 10:55:01 PM · #6
I've figured the same, and I try to do corrections of tilt and skew as one of my first adjustments.
 06/03/2008 11:00:50 PM · #7
Originally posted by bvy:
Obviously rotations which are 90 degree (or multiple thereof) are not at all destructive...


Would you be surprised to learn that (most) 90-degree rotations of JPEG data *are* a little destructive? There are "lossless rotation" functions, however the give-back is that you might lose a few pixels of data at the edges of your image.
 06/03/2008 11:15:35 PM · #8
I've definitely noticed the loss of detail when rotating at other than 90-degree increments. Has the degradation been measured for other rotations though? I wonder. Good question.
 06/03/2008 11:35:46 PM · #9
Just saving a jpeg without changing anything causes degrading. Although at maximum quality it should be very hard to see a difference between the two copies, unless the second one was made years after the first with many copies made in between. Rotating 90 degrees also degrades the image, though minimally. I need to add, it's not the rotation that causes the degradation but the saving of the file in jpeg. If you save in tif or another lossless format there isn't any degrading that occurs.

Message edited by author 2008-06-03 23:39:34.
 06/04/2008 12:05:04 AM · #10
Originally posted by Jac:
Just saving a jpeg without changing anything causes degrading. Although at maximum quality it should be very hard to see a difference between the two copies, unless the second one was made years after the first with many copies made in between. Rotating 90 degrees also degrades the image, though minimally. I need to add, it's not the rotation that causes the degradation but the saving of the file in jpeg. If you save in tif or another lossless format there isn't any degrading that occurs.


Not entirely accurate. I just made a 2000x2000 JPG, high quality of 5 2000x200 black lines and 5 2000x200 white lines intermingled. I saved as a JPG in CS3, but never closed and opened.

The first version was pristine.

Then I copied the file (in CS3 CTRL-SHIFT-D) and rotated this new file 90 degrees. It appeared to lose some of the black on the right edge.

Then I copied the original again and rotated this one 13 degrees (arbitrary # I picked). Where the white and black met, CS3 pixelated some to grey to preserve the appearance. This was barely noticeable at 100%, but at 200% it was evident.

Finally I copied this 13 degree version and rotated it back 13 degrees (Counter-clockwise). It appears the same as the original except now the grey areas between the white and black areas was evident... but again at 200%.

Of course if you did this with real photo data, it would not be a clear straight line example, but the same guessing/adjustments in the algorithm would occur.

None of this involved actually "saving" the JPG, but the rotation issue still occured.

The saving of the JPG does not take effect until you reopen the JPG.

So if you start with a new image in CS3 and save it at 75%, the version on disk is 75% quality. However, if you never close and reopen, but make changes and save again, the version is still 75% (the in-memory version is still "whole"). If you saved, closed, reopened, saved again, then you'd have 75% of 75% - or 56.25% quality.
 06/04/2008 12:46:26 AM · #11
Interesting test results. Thanks for the deepdive HawkeyeLonewolf.

ps. you need a hobby (oh wait... i guess this is your hobby).
 06/04/2008 04:25:01 PM · #12
anything you do to a jpg is destructive, even if it s a simple open and close. If you shoot jpg you have already lost a lot of data before if every reaches your computer. That's why it is always a good reason to shoot RAW and edit RAW copies, or if you don't have software to handle your RAW files, convert them to Tiff, edit in Tiff. Then when you are ready to send to the printer save a jpg copy.

Message edited by author 2008-06-04 16:26:28.
 06/04/2008 09:09:56 PM · #13
Originally posted by roba:
It really depends on the content of the image. It is definitely destructive, and mostly shows up on patterns with lots of thin, straight lines - especially with structures/cityscapes. Having said that, and now looking at your profile... you are apparently the person it will effect the most :o\


You noticed. Thanks. Here's another advantage to shooting in RAW, I suppose.
 06/04/2008 09:25:15 PM · #14
Originally posted by bvy:
Here's another advantage to shooting in RAW, I suppose.

Whether the original camera capture is a RAW or JPEG file should make no difference when it comes to image degradation due to rotation; degradation due to repeated iterations of saving with lossy compression is a separate issue.

Regarding a separate but related issue: when I'm re-sizing, I try to use an "even" reduction ratio. For example, with my Canon, a 25% reduction takes the long side to 704 pixels -- I use that rather than the maximum 720 allowed in Advanced, because I suspect (but don't know) that the algorithms will work "better" if they don't have to calculate values for/from fractional pixels.

Message edited by author 2008-06-04 21:29:26.
 06/04/2008 10:06:45 PM · #15
Originally posted by GeneralE:
Originally posted by bvy:
Here's another advantage to shooting in RAW, I suppose.

Whether the original camera capture is a RAW or JPEG file should make no difference when it comes to image degradation due to rotation; degradation due to repeated iterations of saving with lossy compression is a separate issue.


I was thinking in terms of 90 degree rotations. I can believe that there's some loss (or, at best, change) to a JPG file when rotating it 90 degrees. RAW and TIFF should suffer no degradation though.
06/04/2008 11:36:15 PM · #16
Originally posted by GeneralE:
Originally posted by bvy:
Here's another advantage to shooting in RAW, I suppose.

Whether the original camera capture is a RAW or JPEG file should make no difference when it comes to image degradation due to rotation; degradation due to repeated iterations of saving with lossy compression is a separate issue.

Regarding a separate but related issue: when I'm re-sizing, I try to use an "even" reduction ratio. For example, with my Canon, a 25% reduction takes the long side to 704 pixels -- I use that rather than the maximum 720 allowed in Advanced, because I suspect (but don't know) that the algorithms will work "better" if they don't have to calculate values for/from fractional pixels.


Correct.

Another tip for CS3 users is to always view on-screen at an zoom that is 2x or 1/2 the real size. So 100%, 50%, 25%, 12.5%, 200%, 400% all display accurately.

For some reason CS3 likes to default to 33% when I first open the file and you'll "SEE" jaggies in areas where there really aren't any. So if you make decisions based on that, it won't result in an accurate representation.
Pages:  
Current Server Time: 08/27/2025 03:45:24 PM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 08/27/2025 03:45:24 PM EDT.