Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo
I'm trying to construct high dynamic range environment maps using Hugin and enblend. Ideally I'd like to combine the multiple exposure levels into a single high dynamic range format before I even start stitching them (just less images to work with). I'm able to load either Floating Point tiff files or RGBE files into hugin and it will stitch them and apparently produce floating point tiff files as output. In addition enblend seems happy to read the floating point tiff files however, when it stitched them there were areas where very large, very strange artifacts were created. The artifacts looked like some sort of attempt at blending so I'm guessing that even though enblend can read floating point tiff files it does not seem to handle the blending correctly. Is this correct or am I just doing something wrong?
I have never tested floating point images, so this is probably a bug. Could you post or email small versions of your files? I can try to track down the problem.
I have come accross a similar issue. Please have a look at http://www.milan-knizek.net/cms/index.php?q=gallery&g2_itemId=1947&g2_imageViewsIndex=1
There is a large black artifact in the blended image, while the optimised images from hugin do not have it.
In the same pano, there were artifacts on various parts of the blended image, mainly in low or mid tones (grass, brick wall).
If you want, you can download the (resized) optimised tiff files from
Running "enblend -o test.tif *small.tif" results in the artifacts.
I am using enblend 2.5, with the 3.0 version I did not succeed on blending hdr tiffs at all (result was a completely black image).
I have just uploaded a patch that fixes the emtpy output issue of enblend 3.0
Following the proposal of Bruno Postle to apply gamma of 2.2 to the images before running enblend helped. It seems a good idea to apply gamma of 0.45 to the blended image to decompress the hdr scale a bit.
One can try this workaround on the test images, however the gamma of 2.2 must be applied in CinePaint (Image / Color / Levels), since CinePaint keeps the alpha channel.
Pablo: thanks for the patch.
Hi there guys,
I've been having the same problem blending HDR images with enblend 3.0. It simply says the bounding box is 0x0 large... leaving me with one big black image in the end.
I hope the patch from Pablo does the trick in fixing this, though I can't test it myself at the moment seeing that finding updated windows executables for enblend is near impossible... Is there a developer for windows, or do you just build the major releases for it?
I'll try out the patch when I can pull myself away from CS3 one day and reboot to suse...
Y'all have a nice day now,
Thanks again for the patch Pablo.