Ray West
May 31st, 2007, 01:22 PM
An image posted in recent thread caused me to 'get back to basic's. I was seeing jpeg artifacts, some were insisting it was camera blur. Well, jpeg compression is a no-no wrt comparing images at a pixel peeping level. How to prove?
three images below, using more or less the colours in the post to which I allude.
The image was drawn in cs2 with a hard edged brush, using colours eye dropped from the original thread, and is saved as a tiff at http://www.yertiz.com/images/tiff.tif (it may not display in your browser- you can probaly get it by dragging the link to your desk top, and opening the file. (use cs2 if you can))http://www.yertiz.com/images/tiff.tif
The same image was also saved as a couple of jpegs, one at highest accuracy (large file) and the final at about 30%, both in cs2. (not in save for web)
http://www.yertiz.com/images/highjpeg.jpg http://www.yertiz.com/images/highjpeg.jpg
and the third
http://www.yertiz.com/images/lowjpeg.jpg http://www.yertiz.com/images/lowjpeg.jpg
Now, start your pixel peeping! even zooming in at ten times or so, in cs2 on the _tiff_, see how the colours at the boundary change. (I looked at the top edge of black to dark grey, and the top of dark grey to light grey, but anywhere will do). Use the eyedropper, do not trust your eyes, the web and elsewhere is full of examples of optical illusions wrt colour. There is sharpening/anti-aliasing built into cs2. Now something similar will take place in all photo editing software, I guess. Maybe it can be turned off, maybe not.
Bearing that in mind, the jpegs are not even worth looking at, but of course we want to. The larger jpeg file appears more blurred at the edges where the contrast is not high, (lhs). The smaller, well, that's smaller, and there is a load of apparently weird stuff going on there..
I'm hoping this shows that it is virtually impossible to compare camera results, unless you have both cameras, and use the same software, on the same subject in the same conditions. All raw converters perform differently to start with, but I think the foregoing simply demonstrates that at the level of pixel peeping, exactly whose pixels are you peeping?
Generally speaking, the errors? inherent in the software do not cause a problem, or at least you can work around them, in normal photography, but if you are trying to compare camera performance, I think you need to be sure of exactly what it is that you are comparing.
I fully realise this is all a bit quick and dirty, but I think it is the way many folk pixel peep.
Best wishes,
Ray
(If I'm way off beam with this, then I'll stick it in my 'Friday fun day' thread ;-)
three images below, using more or less the colours in the post to which I allude.
The image was drawn in cs2 with a hard edged brush, using colours eye dropped from the original thread, and is saved as a tiff at http://www.yertiz.com/images/tiff.tif (it may not display in your browser- you can probaly get it by dragging the link to your desk top, and opening the file. (use cs2 if you can))http://www.yertiz.com/images/tiff.tif
The same image was also saved as a couple of jpegs, one at highest accuracy (large file) and the final at about 30%, both in cs2. (not in save for web)
http://www.yertiz.com/images/highjpeg.jpg http://www.yertiz.com/images/highjpeg.jpg
and the third
http://www.yertiz.com/images/lowjpeg.jpg http://www.yertiz.com/images/lowjpeg.jpg
Now, start your pixel peeping! even zooming in at ten times or so, in cs2 on the _tiff_, see how the colours at the boundary change. (I looked at the top edge of black to dark grey, and the top of dark grey to light grey, but anywhere will do). Use the eyedropper, do not trust your eyes, the web and elsewhere is full of examples of optical illusions wrt colour. There is sharpening/anti-aliasing built into cs2. Now something similar will take place in all photo editing software, I guess. Maybe it can be turned off, maybe not.
Bearing that in mind, the jpegs are not even worth looking at, but of course we want to. The larger jpeg file appears more blurred at the edges where the contrast is not high, (lhs). The smaller, well, that's smaller, and there is a load of apparently weird stuff going on there..
I'm hoping this shows that it is virtually impossible to compare camera results, unless you have both cameras, and use the same software, on the same subject in the same conditions. All raw converters perform differently to start with, but I think the foregoing simply demonstrates that at the level of pixel peeping, exactly whose pixels are you peeping?
Generally speaking, the errors? inherent in the software do not cause a problem, or at least you can work around them, in normal photography, but if you are trying to compare camera performance, I think you need to be sure of exactly what it is that you are comparing.
I fully realise this is all a bit quick and dirty, but I think it is the way many folk pixel peep.
Best wishes,
Ray
(If I'm way off beam with this, then I'll stick it in my 'Friday fun day' thread ;-)