Reference entries for the bf16 repo and the sdnq uint4 quant load the
modular pipeline through the standard dispatch. Image tabs run the
model in still mode with audio off; the video tab keeps its own
overrides through the shared per-generation hook. Detailer is not
supported and is disabled with a warning.
Krea 2 is a 12.9B single-stream flow-matching DiT trained from scratch, using a Qwen3-VL-4B text encoder and the Qwen-Image VAE. The transformer is vendored as a diffusers ModelMixin whose module tree mirrors the checkpoint, so weights load with no key conversion; the pipeline ports the reference encode, flow-matching denoise, and VAE decode. The text encoder is shared at runtime via the existing dedup registry, so Base and Turbo reuse one Qwen3-VL-4B copy.
Covers text-to-image, image-to-image, native LoRA, and the single-file UNET override. Also completes SD.Next's partial Qwen-Image VAE support (5D decode input and TAESD preview mapping) that K2 shares.
sd_hijack_accelerate.py: bias.to(weight.dtype) discarded its result (Tensor.to is not in-place), so the conv still received the mismatched bias. sd_hijack_te.py: os.environ.get returns a str when MAX_SEQUENCE_LENGTH is set, so max(int, str) raised TypeError (uncaught, before the try); coerce with int(). sd_detect.py: pipeline was referenced in 'if callable(pipeline)' but only assigned when cls is not None, raising UnboundLocalError for a model_index.json without _class_name.
Co-Authored-By: Claude <noreply@anthropic.com>
Diffusers-native port of the 9.3B flow-matching DiT: dual-transformer asymmetric CFG, a 13-layer Qwen3-VL tap encoder deduped with VQA and prompt-enhance, the Flux.2 VAE, and a logit-normal schedule. Loads a published bf16 repo with SDNQ at load.
- Relax sd_detect to match 'anima' without requiring 'cosmos' in name
- Use hf_hub_download for custom pipeline.py and adapter modules
- Register custom modules in sys.modules for Diffusers trust_remote_code
- Pass trust_remote_code=True to from_pretrained
- Map AnimaTextToImage to 'cosmos' model type for TAESD preview support
Anima replaces the Cosmos T5-11B text encoder with Qwen3-0.6B + a
6-layer LLM adapter and uses CONST preconditioning instead of EDM.
- Add pipelines/model_anima.py loader with dynamic import of custom
AnimaTextToImagePipeline and AnimaLLMAdapter from model repo
- Register 'Anima' pipeline in shared_items.py
- Add name-based detection in sd_detect.py
- Fix list-format _class_name handling in guess_by_diffusers()
- Wire loader in sd_models.py load_diffuser_force()
- Skip noise_pred callback injection for Anima (uses velocity instead)
- Add output_type='np' override in processing_args.py
Add support for FLUX.2 Klein distilled models (4B and 9B variants):
- Add pipeline loader for Flux2KleinPipeline
- Add model detection for 'flux.2' + 'klein' patterns
- Add pipeline mapping in shared_items
- Add shared Qwen3ForCausalLM text encoder handling:
- 4B variants use Z-Image-Turbo's Qwen3-8B
- 9B variants use FLUX.2-klein-9B's Qwen3-14B
- Add reference entries for distilled (4B, 9B) and base models
- Update diffusers commit for Flux2KleinPipeline support
- Add GLM-Image (zai-org/GLM-Image) model detection and loading
- Custom pipeline loader with proper component handling:
- ByT5 text encoder (cannot use shared T5 due to different hidden size)
- Vision-language encoder (9B AR model)
- DiT transformer (7B)
- Fix EOS token early stopping in AR generation
- Add AR token generation progress tracking with terminal progress bar
- Fix uninitialized audio variable in processing
- Add TAESD support for GLM-Image (using f1 variant)
Vladimir Mandic
CalamitousFelicitousness
QualiaRain
Oleksandr Liutyi
awsr