These pictures serve to illustrate the differences in bit depth. Notice how the dithering gets markedly worse as a smaller palette is used. Of course, this picture of a rose would be better displayed as a JPEG, but it serves well to show how the graphics degrade as bit depth decreases.
![]() 8 bit gif, 19K |
![]() 7 bit gif, 15K |
![]() 5 bit gif, 9K |
![]() 3 bit gif, 5K |
The moral of the story: Use your eye to determine minimum image quality Vs file size.
Compare the file sizes in these logos. Note that this is a fairly simple graphic with no gradients or subtle shade differences. None of the jpeg compressions can come close to the 3K gif version. Be sure to notice the distortion of the highly compressed jpeg. This is not a bug, it's how jpeg works, and brilliantly illustrates how gif is better at these sorts of graphics.
![]() gif, 3k |
![]() jpeg, high compression, 6k |
![]() jpeg, medium comp, 12K |
![]() jpeg, low comp, 21K |
The moral of the story: gif is better than jpeg for images with lots of repeated colors (logos, icons and files with few colors).
JPEG for photos! Note that in this case, gif cannot compress file size as much as jpeg, even if low compression is used.
Get your nose up to the screen and see how gif "splotches" the middle penguin's white feathers compared to the low compression jpeg. And the gif image is 5K larger!
Now compare the three levels of jpeg; again, use the middle penguin's white feathers as comparison. The high compression creates the same kinds of blocky artifacts we saw in the logo example. It is not usable. But compare the middle and low compression. Is the low compression image worth the extra 9K?
The degree of compression used with jpeg is always a judgment call - size vs. fidelity.
![]() gif, 26K |
![]() jpeg, high compression, 4K |
![]() jpeg, medium compression, 12K |
![]() jpeg, low compression, 21K |
Moral of the story: jpeg is much better than gif for complex images such as photos. Use your eye to determine minimum image quality.
Last modified Monday, February 19, 2007 10:33 PM