Skip to content

Commit 9c1b6d3

Browse files
committed
update cosmos
Signed-off-by: Vladimir Mandic <[email protected]>
1 parent ce5cdff commit 9c1b6d3

File tree

2 files changed

+17
-3
lines changed

2 files changed

+17
-3
lines changed

CHANGELOG.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# Change Log for SD.Next
22

3-
## Update for 2025-08-27
3+
## Update for 2025-08-28
44

55
- **Models**
66
- **Chroma** final versions: [Chroma1-HD](https://huggingface.co/lodestones/Chroma1-HD), [Chroma1-Base](https://huggingface.co/lodestones/Chroma1-Base) and [Chroma1-Flash](https://huggingface.co/lodestones/Chroma1-Flash)
@@ -16,6 +16,7 @@
1616
- additional hints to ui, thanks @Artheriax
1717
- add video support to gallery, thanks @CalamitousFelicitousness
1818
- additional artwork for reference models in networks, thanks @liutyi
19+
- improve ui hints display
1920
- **Offloading**
2021
- enable offload during pre-forward by default
2122
- improve offloading of models with multiple dits
@@ -28,8 +29,7 @@
2829
- fix handling of pre-quantized `flux` models
2930
- fix `wan` use correct pipeline for i2v models
3031
- fix `qwen-image` with hires
31-
- fix `omnigen-2`
32-
- fix ui hints display
32+
- fix `omnigen-2` failure
3333

3434
## Update for 2025-08-20
3535

modules/video_models/models_def.py

Lines changed: 14 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -320,4 +320,18 @@ def __str__(self):
320320
te_cls=transformers.T5EncoderModel,
321321
dit_cls=diffusers.CogVideoXTransformer3DModel),
322322
],
323+
'nVidia Cosmos': [
324+
Model(name='nvidia Cosmos Predict2 2B I2I',
325+
url='https://huggingface.co/nvidia/Cosmos-Predict2-2B-Text2Image',
326+
repo='nvidia/Cosmos-Predict2-2B-Video2World',
327+
repo_cls=diffusers.Cosmos2VideoToWorldPipeline,
328+
te_cls=transformers.T5EncoderModel,
329+
dit_cls=diffusers.CosmosTransformer3DModel),
330+
Model(name='nvidia Cosmos Predict2 2B I2I',
331+
url='https://huggingface.co/nvidia/Cosmos-Predict2-2B-Text2Image',
332+
repo='nvidia/Cosmos-Predict2-2B-Video2World',
333+
repo_cls=diffusers.Cosmos2VideoToWorldPipeline,
334+
te_cls=transformers.T5EncoderModel,
335+
dit_cls=diffusers.CosmosTransformer3DModel),
336+
],
323337
}

0 commit comments

Comments
 (0)