jinx/ComfyUI
1
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-08-02 23:14:49 +08:00
Code Issues Activity

Add PixArt model support (#6055)

Browse Source 
* PixArt initial version

* PixArt Diffusers convert logic

* pos_emb and interpolation logic

* Reduce  duplicate code

* Formatting

* Use optimized attention

* Edit empty token logic

* Basic PixArt LoRA support

* Fix aspect ratio logic

* PixArtAlpha text encode with conds

* Use same detection key logic for PixArt diffusers
This commit is contained in:
City
2024-12-20 21:25:00 +01:00
committed by GitHub
parent 418eb7062d
commit bddb02660c
13 changed files with 1079 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
comfy/sd.py
View File

@@ -27,6 +27,7 @@ import comfy.text_encoders.sd2_clip
import comfy.text_encoders.sd3_clip
import comfy.text_encoders.sa_t5
import comfy.text_encoders.aura_t5
import comfy.text_encoders.pixart_t5
import comfy.text_encoders.hydit
import comfy.text_encoders.flux
import comfy.text_encoders.long_clipl
@@ -604,6 +605,8 @@ class CLIPType(Enum):
MOCHI = 7
LTXV = 8
HUNYUAN_VIDEO = 9
PIXART = 10
def load_clip(ckpt_paths, embedding_directory=None, clip_type=CLIPType.STABLE_DIFFUSION, model_options={}):
clip_data = []
@@ -696,6 +699,9 @@ def load_text_encoder_state_dicts(state_dicts=[], embedding_directory=None, clip
elif clip_type == CLIPType.LTXV:
clip_target.clip = comfy.text_encoders.lt.ltxv_te(**t5xxl_detect(clip_data))
clip_target.tokenizer = comfy.text_encoders.lt.LTXVT5Tokenizer
elif clip_type == CLIPType.PIXART:
clip_target.clip = comfy.text_encoders.pixart_t5.pixart_te(**t5xxl_detect(clip_data))
clip_target.tokenizer = comfy.text_encoders.pixart_t5.PixArtTokenizer
else: #CLIPType.MOCHI
clip_target.clip = comfy.text_encoders.genmo.mochi_te(**t5xxl_detect(clip_data))
clip_target.tokenizer = comfy.text_encoders.genmo.MochiT5Tokenizer