Replace leftover Depth-Anything-V2 duplicate with the real PixelModel v5 / VoxelModel v1 demo

#1

This Space was a straight duplicate of Depth-Anything-V2-small that was never replaced with the actual app. Removes the depth estimation code and assets, adds a Gradio app with two tabs: text-to-image via PixelModel v5 and text-to-3D via VoxelModel v1 (rendered as a glb mesh from the raw occupancy grid). Both run on CPU, matching this Space's existing hardware tier. Smoke-tested end to end before push.

BenchLabs changed pull request status to merged
BenchLabs deleted the refs/pr/1 ref

Sign up or log in to comment