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

Omnigen2 model implementation. (#8669)

Browse Source
This commit is contained in:
comfyanonymous
2025-06-25 16:35:57 -07:00
committed by GitHub
parent 7a13f74220
commit ec70ed6aea
13 changed files with 152295 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
comfy/supported_models.py
View File

@@ -18,6 +18,7 @@ import comfy.text_encoders.cosmos
import comfy.text_encoders.lumina2
import comfy.text_encoders.wan
import comfy.text_encoders.ace
import comfy.text_encoders.omnigen2
from . import supported_models_base
from . import latent_formats
@@ -1181,6 +1182,36 @@ 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, CosmosT2IPredict2, CosmosI2VPredict2, Lumina2, WAN21_T2V, WAN21_I2V, WAN21_FunControl2V, WAN21_Vace, WAN21_Camera, Hunyuan3Dv2mini, Hunyuan3Dv2, HiDream, Chroma, ACEStep]
class Omnigen2(supported_models_base.BASE):
unet_config = {
"image_model": "omnigen2",
}
sampling_settings = {
"multiplier": 1.0,
"shift": 2.6,
}
memory_usage_factor = 1.65 #TODO
unet_extra_config = {}
latent_format = latent_formats.Flux
supported_inference_dtypes = [torch.float16, torch.bfloat16, torch.float32]
vae_key_prefix = ["vae."]
text_encoder_key_prefix = ["text_encoders."]
def get_model(self, state_dict, prefix="", device=None):
out = model_base.Omnigen2(self, device=device)
return out
def clip_target(self, state_dict={}):
pref = self.text_encoder_key_prefix[0]
hunyuan_detect = comfy.text_encoders.hunyuan_video.llama_detect(state_dict, "{}qwen25_3b.transformer.".format(pref))
return supported_models_base.ClipTarget(comfy.text_encoders.omnigen2.LuminaTokenizer, comfy.text_encoders.omnigen2.te(**hunyuan_detect))
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, Omnigen2]
models += [SVD_img2vid]