mirrors/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-09-10 11:35:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

Basic initial support for cosmos predict2 text to image 2B and 14B models. (#8517)

Browse Source
This commit is contained in:
comfyanonymous
2025-06-13 04:05:23 -07:00
committed by GitHub
parent c6529c0d77
commit 251f54a2ad
7 changed files with 1021 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

40
comfy/supported_models.py
View File

@@ -908,6 +908,44 @@ class CosmosI2V(CosmosT2V):
out = model_base.CosmosVideo(self, image_to_video=True, device=device)
return out
class CosmosT2IPredict2(supported_models_base.BASE):
unet_config = {
"image_model": "cosmos_predict2",
"in_channels": 16,
}
sampling_settings = {
"sigma_data": 1.0,
"sigma_max": 80.0,
"sigma_min": 0.002,
}
unet_extra_config = {}
latent_format = latent_formats.Wan21
memory_usage_factor = 1.6 #TODO
supported_inference_dtypes = [torch.bfloat16, torch.float16, torch.float32] #TODO
def get_model(self, state_dict, prefix="", device=None):
out = model_base.CosmosPredict2(self, device=device)
return out
def clip_target(self, state_dict={}):
pref = self.text_encoder_key_prefix[0]
t5_detect = comfy.text_encoders.sd3_clip.t5_xxl_detect(state_dict, "{}t5xxl.transformer.".format(pref))
return supported_models_base.ClipTarget(comfy.text_encoders.cosmos.CosmosT5Tokenizer, comfy.text_encoders.cosmos.te(**t5_detect))
class CosmosI2VPredict2(CosmosT2IPredict2):
unet_config = {
"image_model": "cosmos_predict2",
"in_channels": 17,
}
def get_model(self, state_dict, prefix="", device=None):
out = model_base.CosmosPredict2(self, image_to_video=True, device=device)
return out
class Lumina2(supported_models_base.BASE):
unet_config = {
"image_model": "lumina2",
@@ -1139,6 +1177,6 @@ class ACEStep(supported_models_base.BASE):
def clip_target(self, state_dict={}):
return supported_models_base.ClipTarget(comfy.text_encoders.ace.AceT5Tokenizer, comfy.text_encoders.ace.AceT5Model)
models = [LotusD, Stable_Zero123, SD15_instructpix2pix, SD15, SD20, SD21UnclipL, SD21UnclipH, SDXL_instructpix2pix, SDXLRefiner, SDXL, SSD1B, KOALA_700M, KOALA_1B, Segmind_Vega, SD_X4Upscaler, Stable_Cascade_C, Stable_Cascade_B, SV3D_u, SV3D_p, SD3, StableAudio, AuraFlow, PixArtAlpha, PixArtSigma, HunyuanDiT, HunyuanDiT1, FluxInpaint, Flux, FluxSchnell, GenmoMochi, LTXV, HunyuanVideoSkyreelsI2V, HunyuanVideoI2V, HunyuanVideo, CosmosT2V, CosmosI2V, Lumina2, WAN21_T2V, WAN21_I2V, WAN21_FunControl2V, WAN21_Vace, WAN21_Camera, Hunyuan3Dv2mini, Hunyuan3Dv2, HiDream, Chroma, ACEStep]
models = [LotusD, Stable_Zero123, SD15_instructpix2pix, SD15, SD20, SD21UnclipL, SD21UnclipH, SDXL_instructpix2pix, SDXLRefiner, SDXL, SSD1B, KOALA_700M, KOALA_1B, Segmind_Vega, SD_X4Upscaler, Stable_Cascade_C, Stable_Cascade_B, SV3D_u, SV3D_p, SD3, StableAudio, AuraFlow, PixArtAlpha, PixArtSigma, HunyuanDiT, HunyuanDiT1, FluxInpaint, Flux, FluxSchnell, GenmoMochi, LTXV, HunyuanVideoSkyreelsI2V, HunyuanVideoI2V, HunyuanVideo, CosmosT2V, CosmosI2V, CosmosT2IPredict2, CosmosI2VPredict2, Lumina2, WAN21_T2V, WAN21_I2V, WAN21_FunControl2V, WAN21_Vace, WAN21_Camera, Hunyuan3Dv2mini, Hunyuan3Dv2, HiDream, Chroma, ACEStep]
models += [SVD_img2vid]