Differences between tiff and "fine JPEG"

greenspun.com : LUSENET : Imaging Resource Discussion : One Thread

I looked at the Nikon950 pictures sample (in the "fine JPEG mode" and the "tif mode" in 1600x1200 resolution).

I didn't notice any differences between the uncompressed 5.7Mb file and the 850 Kb best JPEG compressed mode. This is true for test pattern image or closed portrait or landscape pictures. So why the uncompressed mode does exist?

Am I wrong or does someone see differences between these two modes (on the Nikon950 or another camera).

Thanks Francois

-- Francois DAVID (FjhDAVID@aol.com), March 19, 1999

Answers

Well implemented JPEG should appear virtually lossless to the naked eye at compression ratios of approximately 4-1 or lower. This is the target compression ratio of the Nikon Fine Mode.

IF you start getting up to ratios of 8-1 or more, you should be able to tell the difference on most subject matter.

-- Doug Green (dougjgreen@yahoo.com), March 19, 1999.


Explaining why you couldn't see the differences between compressed and uncompressed images kind of becomes a problem in physics, but I'm going to take a shot at it. I think you had a difficult time noting the differences between the compressed image and the uncompressed image because you tried to view the errors on a computer screen at a fairly high resolution.

Contrary to the complaints of anyone who thinks 640x480 is too low a resolution to view a decent image, millions of people watch television quite comfortably every day at a considerably lower resolution with no real complaints of aliasing (another word for typical jpeg errors). The trick is that projected mediums often perform a bit of what is known as anti-aliasing due to the nature of how the image is produced. I believe the anti-aliasing occurs because of the way the light from adjacent pixels is combined and projected as well as some effects produced by the gap between pixels created by the shadow mask and the fact that television is typically viewed from a much greater distance than computer screens. I think by looking at the image at a higher resolution (1600x1200) you effectively reduced the size of each pixel and made the errors harder to spot. Jpeg errors that are visable in an image that is compressed about 6.7 to 1 should be easier to see at 640x480, but are still hard to spot in many cases.

The best way to show the worth of the uncompressed image is to download both images and print them at larger sizes. The printed output will probably make it easier to spot the problems, since you can blow the pictures up so that the pixels printed are much larger than they would appear at 1600x1200 onscreen resolution. Most graphics programs let you zoom in to perform a similar operation onscreen, but are still subject to the "built-in" anti-aliasing that the onscreen display provides.

If anyone doubts that this anti-aliasing effect exists, simply zoom-in to blow up a picture to the point where it becomes blocky onscreen and then move away from the screen. The results will probably suprise you. You'll also notice a similar effect with printed output.

I hope that makes some sense?

-- Gerald Payne (gmp@francorp.francomm.com), March 19, 1999.


Moderation questions? read the FAQ