Mine eyes - a saving test

Holly Snyder · 01-21-2005, 09:49 AM

Mike and Hanna,

When you're comparing the image that's been saved multiple times vs. the original image, enlarge the images to 300+ percent and look at one small area of the images. A drop of water perhaps. You should see a difference. I'm not surprised that you don't see a difference in images saved multiple times as a maximum quality jpg, and then posted on the forum, as they're only displaying 72 dpi.

Just to prove to yourself that jpg is a lossy format, try saving the same image, 20 times, at the lowest quality jpg setting, and there should be a very noticeable difference (without having to magnify the image to see it). Very nice closeups, by the way.

Holly

Mike McCarty · 01-21-2005, 10:30 AM

Holly,

Here are two images patched together. The one on the right was saved 21 times the other only cropped down from the original. I didn't save the one as anything other than the resolution that it began it's life with.

I would think that if you took two identical images, reduced one to the lowest resolution and then compared it to the other, without any saving abuse at all, that would be enough to make for a stark comparison.

Michele Rushworth · 01-21-2005, 10:52 AM

Looks pretty clear: when choosing the highest quality (ie not much compression) the jpeg file format seems to be more than adequate for our needs in storing reference photos to paint from.

Holly Snyder · 01-21-2005, 10:57 AM

The thing is that you're displaying them at only 72 dpi on the forum, so any loss in image quality will probably not be viewable. If you look at those same images in your graphics program, zoomed in significantly (300+ percent) so that you can see the enlarged individual pixels, can you see a difference?

While jpg is lossy, it is an excellent algorithm, so that when an image is saved at the lowest compression (highest image quality), the difference vs. the original isn't that noticeable. But try the same test with the highest jpg compression (lowest image quality), and the difference will be much more noticeable.

Holly

Mike McCarty · 01-21-2005, 11:18 AM

Holly,

In rereading your post I don't think I did exactly what you suggested. The images above are just severely cropped. I can zoom in as far as I want in my program and when I do they still look pretty similar.

This one is cropped about as much as I dare (expecting the ghost of Ansel Adams to appear).

I don't have any particular stake in proving this one way or the other, it's just something I got curious about.

Quote:

But try the same test with the highest jpg compression (lowest image quality), and the difference will be much more noticeable.

It seems from what you say that if it is your intention to screw up the image you certainly can. Or, it may actually be screwed up but I am unable to see it. However, if you pay attention the information can be maintained sufficiently. Certainly to the extent that mine eyes are able to utilize it. This gets into the realm of "a wink is as good as a nod to a blind man."

Holly Snyder · 01-21-2005, 01:11 PM

You peaked my curiousity Mike, so I decided to spend too much time and do another test.

The first image is a section of your original image enlarged to 1600% and saved as a jpg at maximum quality. In the second, I began from the first image, and drew a T, pixel by pixel, each time saving it as a jpg at maximum quality. There's hardly any image degradation, just a very slight color change of some of the pixels.

The third image began as the first, but saved as low quality -obvious deterioration. The fourth began from the third, and drew a T in the same location as in image two, pixel by pixel, each time saving it as a jpg at low quality. Again more deterioration.

Images one and two are much closer than I would have suspected. It may be hard to see, but if you look at the top row, fifth and sixth pixels from the right, you'll see a color change.

I guess the upshot is that jpgs, at max quality, are certainly fine to paint from!

Ok, back to work!

Holly