Note 11.6k images with cloth - human pairs including masks and poses.

Note Large scale synthetic image prompt pairs. Generated using Stable Diffusion

Note 2.8M real world image - text pairs. Text is *not* from a VLM or a captioning model though

Note Text to 3D dataset filtered from Objectverse. Viewer doesn't work but has 365k pairs iirc.

Note A collection of diverse prompts across multiple domains.