Forward inference using a pretrained model

[ayclove]

Falcor 6.0

---

How can I perform inference with a PyTorch model within a RenderPass, for example PathTracer? Referring to the verifyData example in the TestPyTorchPass from the TestPasses, is loading pre-trained PyTorch model weights into the shader code and performing inference within the RenderPass the best approach? Or is there a simpler method?

[derkder]

Hi, have you solved it, Im encountering the same issue. Most of the Falcor+NN project are using tiny-cuda-nn, but I realy want to know if Ive already trained a regular MLP network, how can I implement it to get a render result.