Closed AmanKishore closed 1 year ago
@AmanKishore Hi, in fact we already use a similar acceleration technique (check torch-ngp) for the nerf part. The major time bottleneck is the diffusion model now.
Oh that's great!
@ashawkey you might be interested in this https://twitter.com/realDanFu/status/1580641495991754752
Curious if https://github.com/KAIR-BAIR/nerfacc could accelerate training and inference for dreamfusion!