Hello everybody,
I encountered some strange behavior with the Output of the zDepth-Channel of Coronas Multipass. To start at the beginning:
I read an article about
When to NOT Linearize Your Textures by Mike Boers. I tried to rebuild his setup in a similar manner with the only difference, that I rendered my greyscale-map through the zdepth-channel, the setup looked like this:
The flying planes range from 0 to 100 cm in height. The resulting map I wanted to use for the displacement looked like this:
So far so good. Now I did some testing: I tried to displace a flat plane of same size with this map (EXR) and the bitmap-loader. No matter what setting in the input (embedded, linear, sRGB), the result was always like this:
I would have expected this kind of curve if I told corona to override my linear EXR with 2.2 gamma (if doing this now the curves becomes even steeper of course). The only way to correct the gradient to a correct (linear) stepping was to use an override gamma value of 0.45. So I thought there must be something wrong with the output and in fact, when I color pick the different values of my 32-bit EXR in photoshop they do not represent my intended linear 10%-stepping:
I was sure that the EXR output from corona would be linear, even for the multipass channels. To complete my confusion I then noticed, that when color picking in the VFB it is giving exact linear values after all!
Is there someone who could explain to me why this transition is happening? The background to these experiments is that I usually create my displacement maps by simply rendering them from modeled geometry (since c4d texture baking is not working with corona as far as I know). Do I have to override the gamma value by 0.45 from now on or is there another way?
Just for displaying purposes the comparison of the 10%-step-gradients in linear and sRGB next to each other.
