-
latinorum :)
-
Update model card
-
Add poses/dl3dv/test: VGGT camera poses, DL3DV test split
-
Add poses/re10k/test: VGGT camera poses, RealEstate10K test split
-
Update model card
-
Add stage2-diffusion/dl3dv-video-ft.ckpt: Stage-2 video fine-tune of dl3dv-base (10-frame sequences, temporal blocks only), 100k steps
-
Add stage2-diffusion/dl3dv-upsample-ft.ckpt: Stage-2 f=8->4 fine-tune of dl3dv-base (2x upsampled latents), 50k steps
-
Add stage2-diffusion/dl3dv-base.ckpt: Stage-2 base, DL3DV 256x448, init from re10k-base, 140k steps
-
Add stage2-diffusion/re10k-video-ft.ckpt: Stage-2 video fine-tune of re10k-base (10-frame sequences, temporal blocks only), 100k steps
-
Add stage2-diffusion/re10k-upsample-ft.ckpt: Stage-2 f=8->4 fine-tune of re10k-base (2x upsampled latents), 50k steps
-
Add stage2-diffusion/re10k-base.ckpt: Stage-2 base, RealEstate10K 256^2, init from SD 2.1, 200k steps
-
Add stage1-vae/dl3dv.ckpt: Stage-1 multi-view VAE, DL3DV, 100k steps
-
Add stage1-vae/re10k.ckpt: Stage-1 multi-view VAE, RealEstate10K, 100k steps
-
initial commit
