Via Get3D I understand it can take around 30 minutes or so to generate 3D models from text description at this point. What are the expectations on the ability to speed up that time? Is it in the realm of possibility to watch that time decrease as quickly as the image generators did?
Text-guided shape generation takes around 30min as we use a relatively simple approach of finetuning the 3D generator with a directional CLIP loss. The time to generate shapes based on the text description could be largely reduced by training a text-conditioned 3D generative model in the first place.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.