2 Comments

Do you have any inkling as to when tools like this will be available in ways like DALLE, such that regular people could used text prompts to have 3D objects created for them for use in VR?

Seems like progress is being made fast, but there is still some way to go in terms of performance and also practicality

Expand full comment

We will get into more recent papers, but it seems like the accuracy of these methods is getting really good. I would say they are a bit slow in terms of training/inference time, but overall I think the potential applications to VR are possible and widespread.

In terms of doing text-to-3D, I'm not 100% sure. Recall that the methods we are overviewing use solely feed-forward network architectures. This is drastically different from a lot of the work in text-image generation (i.e., more complex architectures/techniques are used to encode both text and images). However, totally possible these two areas could get bridged soon!

Expand full comment