white space compression

[rayjoslyn]

I am compressing a corridor of tiff images to a MG3 image. My area is comprised of a large amount of white space due to the winding corridor. I have found that the white space does not compress well, so I tried an area of interest by creating a shape around my image corridor an using it as a vector overlay. I first tried the mask method but it to created the large image (9GB). The input images area a total of about 20 GB.

I next tried the weight method and this seemed to create an image size that I was expecting, about 1 GB. My question is why can't the mask method compress out the white areas similar to the weight method? And why does white area add to the compressed image size to the extent it does?

Ray

[rparker]

Hello,

What do you mean by "...the white space does not compress well"? Do you mean that it does not stay as a pure white background and leaves behind "speckles", or something different?

If that is the case, you could try compressing the file and despeckling it at the same time by checking the "Despeckle image" checkbox.

[rayjoslyn]

What I mean by not compressing well is that an area that is all 'white' space should compress down to nearly nothing (in file size). When applying a mask to an image it says that it actually is removing the data either inside or outside the mask, therefore this data should be void (white space) data that should not take up considerable space when stored as part of the compressed image, however this is not the case from what I have seen.

Ray

[rparker]

In a MrSID file Generation 3 and earlier, there is no such thing as "no-data" or "void" data. Each pixel must have a numeric value. For the color white, that value is 255 (RGB bands: 255,255,255). For black it is 0 (RGB: 0,0,0). Each pixel is 8-bits of uncompressed space (per band) regardless of the color values.

We are presently looking into developing an improved version of the MrSID format which will handle void or no-data areas better. I'll post more details as I have them.