Would be more than great to have a 32Bit Floating Point Rendering additionally to the HDR rendering with 16Bit only!
I understand that it might slow down the performance, but at least for the rendering this would be essential for precision of gradients and later compositing <3
The internal processing is 32 bit but the output buffers are 16 because there’s been no real compelling reason to do 32 yet.
What issues are you having with gradients and/or HDR content that the current render pipeline is not capable of? It would be good to see some examples.