Re: [PanoTools-devel] Re:PTMender and Pano12.dll Updates

[Max L. <max...@ve...>]

> 1. Images created using cropped files (during the first stage, the
>    actual creation of the panorama files) are not identical to the
>    ones created using the non-cropped files. Look at the following
>    file:
> 
> http://turingmachine.org/~dmg/temp/differenceCroppedTest.psd
> 
> This PSD file has 2 layers (the  uncropped is the background, the
> cropped is the first layer). The second layer is using difference
> blending. The result should be pure back. The third layer has
> exagerated contrast to show the difference. It is a subtle difference,
> but it is there.

Interesting.  Did you use the fast transform logic?  If so, that might explain 
the difference.  Its interpolation routines depend on start and end points that 
could be different when dealing with cropped images versus full sized images.  
If not, then I'm not sure what might cause this.  In any case, the differences 
are so minor (and certainly visible to the naked eye), that I don't think it is 
much of an issue.

> 2. ComputingStitchingMasks is still memory hungry. I'll fix it in my
>    next patch (ROWSPERSTRIP > 1). It is not using cropped files yet
>    (if nobody else is looking into this I'll fix it).

Sounds good.

> 3. Photoshop is not able to read the cropped files properly. Is this a
>    bug of a "feature" of the tiffs we are generating? If it can't then
>    we need to add an option where they will be resized back to the
>    entire size at the end.

Photoshop doesn't know about (as far as I'm aware) "cropped TIFFs".  Photoshop 
just reads the physical dimensions of the TIFF fils and ignores the offset 
data. Software has to be coded to read this offset data and deal with it 
intelligently.  Enblend does this, but Photoshop doesn't.

Max