Using a latent diffusion image model for implicit 3D rendering.
Code coming ASAP. Small demo here.
Sneaker model weights can be downloaded here. First-stage model here.
The model is able to generate novel views from existing and non-existing sneakers in high-resolution.
The also works on high-resolution generated sneakers from other models. The following is made from a Midjourney creation.
