mirrors/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-09-12 04:27:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

Initial Hunyuan3Dv2 implementation.

Browse Source 
Supports the multiview, mini, turbo models and VAEs.
This commit is contained in:
comfyanonymous
2025-03-19 16:19:50 -04:00
parent 3b19fc76e3
commit 11f1b41bab
9 changed files with 1225 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
comfy/latent_formats.py
View File

@@ -456,3 +456,13 @@ class Wan21(LatentFormat):
latents_mean = self.latents_mean.to(latent.device, latent.dtype)
latents_std = self.latents_std.to(latent.device, latent.dtype)
return latent * latents_std / self.scale_factor + latents_mean
class Hunyuan3Dv2(LatentFormat):
latent_channels = 64
latent_dimensions = 1
scale_factor = 0.9990943042622529
class Hunyuan3Dv2mini(LatentFormat):
latent_channels = 64
latent_dimensions = 1
scale_factor = 1.0188137142395404