Ömer Karışman

Hi! Thanks for the awesome library! It would be great if one could drop in to add and drop out to remove objects to and from the collection view.

Question about speed and memory consumption

2

This paragraph from the paper confuses me: `As shown in Fig. 3, we use ControlNet to control each level of the U-net. Note that the way we connect the ControlNet...

Replicate Cog

1

Here is a modified version of the code to work with [Replicate/Cog](https://github.com/replicate/cog) This helps with running it containerised via Docker locally or host on remote VMs. This is the same...

Hello, thank you for sharing the code with the paper! I took the liberty to implement sample.py on replicate.com to easily use it. https://replicate.com/okaris/progen2 I can send a PR for...

Are there any plans to release the portrait segmentation algorithm as well?

Coming from the diffusers repo about a discussion about how timesteps are chosen for img2img tasks, I was wondering about the reasons for filtering the timesteps based on strength value...

See the problem description here: https://github.com/Mikubill/sd-webui-controlnet/issues/2113

Timesteps are not calculated correctly for img2img pipelines.

27

### Describe the bug I have arrived at this conclusion after experimenting with the LCM models/schedulers, and I believe it applies to all schedulers and img2img pipelines. For longer diffusion...

Use scheduler strength option for set_timesteps

2

# What does this PR do? Changes the SD and SDXL pipelines to use the strength parameter of the scheduler when possible Fixes # (issue) #7651 ## Who can review?...

[https://github.com/huggingface/hf_transfer](https://github.com/huggingface/hf_transfer) Just needs two lines of code 1) `- "hf_transfer"` to `cog.yaml` `python_packages` 2) `os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"` to the top of `train.py` and/or `predict.py`