Off the top of my head, you could brute force it and physically imitate a rolling shutter: render 1080 individual 1x1920 pixel frames (or whatever fits your resolution) across the time of each frame exposure using an saw-tooth oscillating camera. Assembling the single pixel high frames in post ...

Hello

Is there a way to simulate rolling shutter like in Arnold... but in Octane? I'm being asked to provide renders by my ML team that need simulated data that looks like it has been taken with real devices like phones.

If there is a way to do it, please let me know. Alternatively, I could do it ...

The last solution with the null material solved the issue with the ability to control the color so elegantly.

Thank you for your help on this, I appreciate it very much.

Cheers

Another solution that might be closer to what you are trying to do: use a null material on the proxy objects. This way they won't show up in the main beauty render, but they will still show up in the Object layer color AOV.
In this case there is no need for render layers, you can save the beauty ...

Would this work: render the scene once on render layer 1 (normal geo) and save the beauty pass, then render the scene on render layer 2 (proxy geos) and save the Object layer color AOV.

Solution 2, simpler but maybe there are too many objects: don't use render layers or proxy objects, go into the ...

Thank you for your response. Unfortunately English is not my first language and I'm not sure if I understood your formulation.

I will try and explain the issue more clearly.

We would like to render out Beauty passes of objects that are complex to teach AI to recognize those objects in real life ...

Thank you for the response.

It is available and it would do the job of customizing the color, but it doesn't solve the visibility problem. I need to have a beauty pass without the masking objects visible and an info pass where some things are occluded for the AI to group them into one label. That ...

^^ BUMP ^^

Anyone?

Hello all

I am working on setting up a pipeline for ML training with synthetic data sets generated with C4D and Octane.

The beauty shots are all good but when I need to go and set up a the INFO passes as AOVs I have an issue.

I'm masking out some parts of the geometry, in this case the wheels of ...