It your review of the Nikon Coolpix 990, you state the approximate compression ratios for the 990 as 1:6, 1:12, and 1:20, in the manual that comes with the 990 Nikon states the values as 1:4, 1:8, and 1:16, who is correct? Further do all camera makers use the same values or do they vary widely with each manufacturer?

-- Roy Boorman (rboorman@earthlink.net), May 07, 2000

Answers

Compression varies by picture. I'm guessing that the IR reviewers actually measure the compression based upon their test shots. A different set of test images would result in different compression.

Maybe there are standard test images that manufacturers use, but I doubt it. The ratios you posted for the Nikon (4:1, 8:1, 16:1) are all powers of two... maybe Nikon was aiming for these compression levels?

I don't think there are standard compression values that all manufacturers use.

-- Greg Philmon (gphilmon@yahoo.com), June 07, 2000.

Roy,

Compression ratio's vary with the image captured according to the following parameter's. Resolution and Color Bit Depth

JPEG compression applies a mathematical algorithm to an image called a DCT (Discrete Cosine Transform) this algorithm performs averaging about pixels in the image based on their color, the compression ratio is determined by the level of detail desired in the color palette originally describing the initial image.( This technique is a form of "color sampling") By varying the scope of the averaging algorithm about arbitrary pixel locations, different image size/image quality values are attained. You can actually see the regions of averaging in a JPEG image that has been compressed to a high degree, they show up as repeated rectangular regions of similar colors. The algorithm represents say a region with 10 shades of blue by a region with only 3, thus saving storage space on the 7 shades no longer represented. This is what contributes to the "lossy" quality of JPEG compression, once its compressed the extra image information formally represented by those missing 7 shades of blue are gone for good. ( There are some other tricks that JPEG compression uses to save even more space beside color sampling but it forms the major contribution to an images reduced image size.)The compression ratio is simply a quotent of the "after" image size divided by the "before" image size. This being the case, your first question is really irrelevant, who is correct varies with the initial images resolution and local color gamut ( the range of color values within a particular image is its gamut) since this varies from picture to picture compression ratio's are repesentational only of those original samples. As far as I know, there are no standard JPEG image/gamut target images used by many companies to determine compression ratio's so they each come up with their own. So your second question is thus answered. The beginnings of a cross comparison between different compression estimates however, can be made if image size and gamut information is given for the image targets in question. ( Which IR didn't seem to do in its Review ...Tsk Tsk ! )

-Igo4ward

-- David SL (igo4ward@netzero.net), July 04, 2000.