When rendering an image with 16bit integer quantization using mantra and then load the .exr in COPS, it reads as an 16bit float image.

Houdini seems to presume that whenever you use the .exr format, it is meant to be 16bit float. You can of course set the bit depth manually in the file COP to 16bit integer, but I'm unsure whether that is destructive or not.

Am I completely misunderstanding openexr and does the format not have support for 16bit integer? Or is 16bit integer and 16bit float so close to each other that distinguishing between them is just a waste of time?

Thanks.

OpenEXR supports 16bit float, 32bit int and 32bit float, but not 16bit int. Houdini's implementation of OpenEXR in 8.x and lower only supports 16bit float, so regardless of which data type you use, it'll be converted to 16 bit float during the image save.

Incidentally, even though their website ( http://www.openexr.com/about.html) [openexr.com] states that they support 16bit int, they actually support only 32bit unsigned int (I checked with the source code). I'll let them know.

Thanks, that cleared it up.

D.

After notifying the webmaster, the webpage listed above has been updated to 32bit int.