mirror of https://github.com/vladmandic/automatic
4.2 KiB
4.2 KiB
TODO
Main ToDo list can be found at GitHub projects
Current
Future Candidates
- Modular pipelines and guiders
- Refactor: Sampler options
- Feature: Diffusers group offloading
- Feature: Common repo for
T5andCLiP - Feature: LoRA add OMI format support for SD35/FLUX.1
- Video: Generic API support
- Video: LTX TeaCache and others
- Video: LTX API
- Video: LTX PromptEnhance
- Video: LTX Conditioning preprocess
- WanAI-2.1 VACE(https://github.com/huggingface/diffusers/pull/11582)
- SkyReels-v2(https://github.com/huggingface/diffusers/pull/11518)
- Cosmos-Predict2-Video(https://github.com/huggingface/diffusers/pull/11695)
Blocked items
- Upgrade: unblock
pydanticandalbumentations- see https://github.com/Cschlaefli/automatic
- blocked by
insightface
Under Consideration
- IPAdapter negative guidance
- IPAdapter composition
- STG
- SmoothCache
- MagCache
- HiDream GGUF
- Nunchaku PulID
- Dream0 guidance
- SUPIR upscaler
- Remove: Agent Scheduler
- Remove: CodeFormer
- Remove: GFPGAN
- ModernUI: Lite vs Expert mode
- Canvas
Future Considerations
New models
Pending
External:Unified/MultiModal
External:Image2Image/Editing
External:Video
- WAN2GP
- SelfForcing
- DiffusionForcing
- LanDiff
- HunyuanCustom
- HunyuanAvatar
- WAN-CausVid
- WAN-CausVid-Plus t2v
- WAN-StepDistill
Code TODO
pnpm lint | grep W0511 | awk -F'TODO ' '{print "- "$NF}' | sed 's/ (fixme)//g' | sort
- control: support scripts via api
- fc: autodetect distilled based on model
- fc: autodetect tensor format based on model
- flux: loader for civitai nf4 models
- hypertile: vae breaks when using non-standard sizes
- install: enable ROCm for windows when available
- loader: load receipe
- loader: save receipe
- lora: add other quantization types
- lora: add t5 key support for sd35/f1
- lora: maybe force imediate quantization
- lora: support pre-quantized flux
- model load: cogview4: balanced offload does not work for GlmModel
- model load: add ChromaFillPipeline, ChromaControlPipeline, ChromaImg2ImgPipeline etc when available
- model load: force-reloading entire model as loading transformers only leads to massive memory usage
- model load: implement model in-memory caching
- modernui: monkey-patch for missing tabs.select event
- processing: remove duplicate mask params
- resize image: enable full VAE mode for resize-latent