Mr Sid file sizes and auto-stretching in Microstation

[Brandogrammetry]

Ever since getting GeoExpress 8 I've noticed that my file sizes for Sid files are exactly the same in a lot of cases. Today I ran 25 orthos and half of them came out at 5,274kb. I would expect every Sid to be different file sizes since they vary greatly on the amount of black(background) area they have.

The other reason I bring this up is because the images that were much smaller than 5,274kb are auto-stretching in Microstation V8i.

Has anyone experienced this before? Is there a reason why most of my files have the same file sizes?

[rparker]

The file size is most dependent on the following attributes:

width (in pixels)
height
Band number (RGB is three bands)
Bits (typically 8-bit images but 16 is also popular).

These determine the "raw" size of the image.

When you compress in GeoExpress, you actually choose the output file size. If you choose a compression ratio of 20:1, the raw image size will be divided by 20.

[Brandogrammetry]

All my input files are 6000x6000 pixel, 8-bit TIFFs with a file size of 110,622 KB. At a 20:1 compression ratio only half of them come out to 5,274 KB. The others are smaller. Why is this?

[Brandogrammetry]

I'm interested to know why that is although I'm not too concerned with the file size, the main problem is that the 2 smallest files (6KB and 130KB) are much brighter than the surrounding SIDs. They appear like the non-background pixels have been increased in brightness to possibly stretch the histogram since 95% of the image is black. Although, when I open the SIDs in ArcMap or Global Mapper I don't see this effect. For some reason it only displays like this in Microstation. I'm not sure which software is creating this effect, but the file sizes seemed to point to the probability that the images themselves could be the problem.

Sorry, I meant to expand on describing this issue in my original post.

[rparker]

The files could have slightly different metadata. In any case, GeoExpress attempts to hit your requested file size but does not always do so for reasons I do not understand at this time.

However, what you see is not related to the file size, but rather, as you suggest, the dynamic range of the image and how well various software adjusts to a dynamic range which is shifted to one end of the spectrum.

In GeoExpress, there is a color balance tool which allows you to adjust the histogram of the image. Perhaps adjusting this yourself and color balancing the image would help.

[Brandogrammetry]

The problem is that I would have to do that individually to many sids since they aren't all effected the same way. That would be extremely time consuming. The Microstation forum pointed out that they saw dynamic range metadata only in the problematic SID files. It appeared like this...

IMAGE__DYNAMIC_RANGE_LEVEL=98.000000

IMAGE__DYNAMIC_RANGE_WINDOW=197.000000

The other files didn't contain this data. Do you know why GeoExpress would include this data in some files, but not in others? The problem may be that Microstation automatically uses this metadata to stretch the image instead of ignoring it like other programs.