jaidaken/ComfyUI
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Support loading Wan/Qwen VAEs with different in/out channels. (#11405)

Browse Source
This commit is contained in:
comfyanonymous
2025-12-18 14:45:33 -08:00
committed by GitHub
parent e8ebbe668e
commit e4fb3a3572
2 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
comfy/sd.py
View File

@@ -546,7 +546,8 @@ class VAE:
self.downscale_index_formula = (4, 8, 8)
self.latent_dim = 3
self.latent_channels = 16
ddconfig = {"dim": dim, "z_dim": self.latent_channels, "dim_mult": [1, 2, 4, 4], "num_res_blocks": 2, "attn_scales": [], "temperal_downsample": [False, True, True], "dropout": 0.0}
self.output_channels = sd["encoder.conv1.weight"].shape[1]
ddconfig = {"dim": dim, "z_dim": self.latent_channels, "dim_mult": [1, 2, 4, 4], "num_res_blocks": 2, "attn_scales": [], "temperal_downsample": [False, True, True], "image_channels": self.output_channels, "dropout": 0.0}
self.first_stage_model = comfy.ldm.wan.vae.WanVAE(**ddconfig)
self.working_dtypes = [torch.bfloat16, torch.float16, torch.float32]
self.memory_used_encode = lambda shape, dtype: (1500 if shape[2]<=4 else 6000) * shape[3] * shape[4] * model_management.dtype_size(dtype)