[ptx] Re: Enblend 1.1

[Andrew C Mihal]

On Tue, 6 Apr 2004, Bernhard Vogl wrote:

> There remains only one special region of interest. You can see it 
> clearly on the right side of the statue. The stitching/blending problems 
> are clearly my fault (I've checked back with the original images). What
> i can't explain are the brightness "steps" that can be seen in the 
> image. The -w parameter has been used but has no impact on this. Also 
> there is no difference in using different output resolutions.

The first thing to check is if the vertical seams exactly match any of the
Region of Interest bounding boxes that enblend outputs in verbose mode. If
so, look in that same blending step to find which images it is blending
together within the problem ROI. Then you can narrow down the problem
until there are only two input images that demonstrate the artifact when
they are blended together. In this case the problem is probably another 
boundary condition bug that I can try to fix.

If the vertical seams do not match any ROI boundaries, then it's a new bug 
that I don't know about yet.

Andrew

---------------------------------
Andrew Mihal
www-cad.eecs.berkeley.edu/~mihal
mihal at eecs.berkeley.edu