So I finally got around to posting this on here,
I am a fairly young 3D content creator, currently studying a bachelor of animation. I’m interested in many, many, many aspects of CGI & art in general, though as I have a 3D environmental modelling course this semester I thought I would try some environmental/archvis work. From the beginning of this project I had decided I wanted to work all linear 32bit images. I have experience with photography, and in particular HDR photography (for IBL & just for viewing). I originally wanted to composite in photoshop, however its handling of 32bit files was a bit problematic to start off, so I took the opportunity to learn some nuke. Couple things caught me off guard (like terminology such as ‘shuffle’)
Things I learned:
32bit HDR images are great
Nuke is great (seemed a bit slow at first, but you forget that when you see how powerful it is)
How much I had to ‘soften’ my render to make it look somewhat photographic was astounding!
Treat the render as a photo (tonemapping in luminance HDR, don’t be afraid to overexpose,blur,introduce aberrations)
I rendered out 9 passess for the primary foreground; Diffuse colour/direct/indirect, Glossy colour/direct/indirect, Translucent colour/direct/indirect.
Doing this wasn’t completely necessary, however I was able to apply some noise reduction to specific passes (DifDir and DifInd) plus it was a good exercise in node wrangling.
After nuke, I took the .exr into a free program LuminanceHDR for tonemapping
output to 16bit tifs
and finally into photoshop for more touch ups + output to .jpg for web upload.
So, what do you think? Any ideas on improvement?
I think one of the areas could have gotten a bit better at was compositing and maybe a bit more control over colour palette.