OTOY Forums

Hi,

I am confused.

I love the Live viewer in C4D Octane, and how it works with clickable AOVs like Noise and so on. But it seems that the correct way to do a final render is to render in the Picture Viewer, where there are no AOVs or special Octane magic.
I tried to do a final render in the Live Viewer and I got an image size of about 900x900 px, not the 3840x3840 px that I have entered in the render settings.
If so, is there a way to add more samples to the Picture Viewer if the entered one is not enough?

Is this how it is or have I missed something?

Would like to render in the Live Viewer but in full size and autosave when render is done to a Layered EXR or PSD, as it works in the Picture viewer.

Use Live Viewer for previewing, making creative decisions, sampling, etc., and the Picture Viewer for rendering; as the Picture Viewer has several issues with regarding color management and multi-pass support, there is only so much that can be seen of the outputs created by Octane. That said, Octane is creating those files and layers, as you likely know. The best use of Picture Viewer for checking Octane output is motion blur. For everything else, Live Viewer is the way to go. Always render to a file using Render Settings > Octane Renderer > Render AOV group tab for finals.

Use the Lock icon in the Live Viewer window to see actual pixels at final render resolution (as set in Render Settings > Output.)

Thanks for that answer, cleared it out some.

This is my problem, I am rendering a simple scene with a small "window" or hole for the classic Window Light look. And the render times I get with the portal are monstrous. Up tp 40 to 60+ hours.

These are my settings, what am I doing wrong?

Your Max. Samples value is set to 8000. I rarely go over 2048 samples, and I usually start at 512 samples. Enable Adaptive Sampling (your screen grab has clipped off the value, so I can't tell) and use the AI Denoiser (from the Octane Camera tag). The only issue then is if the AI Denoiser is doing the job you need. It works well most of the time; sometimes it will require more samples. In your case, it might, due to the glass vase. I also typically run the GI Clamp at 10.

Hope this helps.

Thanks, adaptive sampling is on. Yes it’s the caustics that is taking so long.
I will also increase the GI clamp, and test again.

By the way, does extra AOVs take extra time?

Yes, caustics will definitely add to the time. You might want to try the PMC kernel if you have many caustics. It is slower typically than the Pathtracing Kernel, but can look better with caustics. That said, the AI denoiser is not supported by the PMC kernel, so you must increase samples in order to denoise.

In most cases, AOVs should not cause an appreciable increase in render time, AFAIK.

The Dispersion make allot of the problems I now see. But it also added to the realism.. I need more GPUs.. haha

haffy, Coherent ratio to around 0.3 + parallel samples @ 32 speeds up things a bit

You could maybe re-render the glass with dispersion with region render.

Thanks for all the help !