Mec4D opened this issue on Feb 26, 2016 ยท 28 posts
bagginsbill posted Sun, 28 February 2016 at 9:04 AM
Sharp shadows from an HDRI require that the image is made to accurately encode the intensity of the sun. Many images I test do not. In fact, most do not. We blame the renderer or the means by which we include the HDRI, but the real culprit is that the HDRI is not physically accurate.
I wrote a long thread showing how to tell if your image is really HIGH dynamic range or just slightly better than JPEG. it has to do with how many f-stops were used in bracketing the photo. Most were only taken with 5 stops dynamic range (32 to 1) which means it's not even close to carrying the total energy that was actually there in the world around the camera.
I have found that boosting the luminance from the sun spot in the image is a way to compensate and get it to produce the correct illumination as a sole light source. However, SuperFly (being a straight path tracer) is not good at finding this hot spot. It takes a very large number of samples to accumulate or integrate the light correctly. It's much simpler to just using a matching infinite light, as is done with most of the sIBL sets.
Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)