http://www.bluelavalamp.net/hugin/large_crop.jpg
http://www.bluelavalamp.net/hugin/small_crop.jpg
As you can see the large crop is darker then the small crop.
Here's the output of diff on the two pto files:
diff test-large_crop.pto test-small_crop.pto
3c3
< p f13 w30576 h21630 v62 E-0.266483 R0 S9484,15985,6919,11887
n"TIFF_m c:LZW"
---
> p f13 w30576 h21630 v62 E-0.266483 R0 S11724,13994,10005,11066 n"TIFF_m c:LZW"
Is this normal behavior? Would enfuse output a different result for a
section of an image based upon the crop of the image?
> Is this normal behavior? Would enfuse output a different result for a
> section of an image based upon the crop of the image?
This could be a bug, but I would expect a crop to be different in some
way to the same region of a larger stitch:
When you stitch a 'cropped' view, Hugin decides which photos are
relevant and ignores the rest, but for the larger stitch all images
are used.
Any output pixel produced by enblend/enfuse can be a combination of
all input pixels from all photos if the maximum number of blending
levels is chosen. If you have just 8 blending levels then photos 2^8
pixels (256 pixels) beyond any frame will influence content in the
frame.
[I've been considering creating a gigapixel stitcher 'wrapper': It
would generate the grid tiles directly rather than stitching the whole
thing at once, but the tiles would each need to be oversized by the
number of blending levels and this border would have to be discarded
after stitching]
--
Bruno
Thanks