@vitous1k and @michalis: thank you, i will test it as soon as i have a few minutes spare time.
EDIT: So, i made a test. Simple scene, environment texture mirror ball hdr (same as before),
rendered 100 samples progressive, 1 time env. tex. color and 1 time env. tex. non-color data.
compared in the blender UV/image editor:
same rendertime (aproximate)
every single pixel of both images have the same R/G/B/A CM R/G/B H/S/V/L value
identical histogram / waveform / vectorscope etc.
Saved as openexr rgba full:
same filesize (every single byte was the same)
Saved as 16 bit rgba tiff and analyzed with imagej
same filesize
everything was the same: histogram color count, highest and lowest value …
Conclusion:
either blender has a bug in calculating color/non-color data or it doesn’t matter what to use.
GodOfBigThings- I had forgotten about that site, thank you! You are absolutely right, it is entirely adequate to use the hdri for the environment lighting only, and the sun lamp for the main light source (especially when in hard/non-diffused light situations). Thank you for the tips, and I will certainly be giving the pano stitching tools another try!
varkenvarken- A great resource, thank you! I had wished I’d known how to setup a scene in such a manner, and now I know. Fantastic!
JoHal- I’ve tried what you suggested, and the brightest points in my .EXR map have RGB values of 3.7813. Obviously that is many, many times lower than what your map contained.
I know I correctly listed the values, and the stop increments, in Photomatix when I assembled the .EXR file. I don’t see how I could have done anything wrong at that point in the workflow. My only guess is that the mirror ball you’re using was shot correctly, with many more stops of data. Can you tell us where you found it? I would like to take a look at it.
I’m using 300 samples/per pixel most of the time, and bounces are set to a minimum of 3, and a maximum of 7.
michalis and Vitos1k- I didn’t know that; I’ve tried it as well though, and render times and results appear identical to me. Is there an advantage I’m overlooking?
This is true for random ubiased approach, but with Multiple Importance Sample option (previous Sample as Lamp) turned on in World Settings you can get really nice, low noise and sharp environment lighting.