/stable-diffusion

Primary LanguageJupyter NotebookOtherNOASSERTION

Update: v1.0.6 even superfaster superbigger-res updated ultimate GUI mode edition, with k_diffusion samplers and CodeFormer and Upscalers (only in gradio), with xformers flash attention support

Optimized Stable Diffusion

Open In Colab

THE ARTROOM GUI IS OUT

See GUI our new 1-click windows app

The hlky's webui version is out

The peacasso GUI version is out

The gradio UI now has every feature on one page. See usage below.

The flash attention Windows support has arrived! To try it out, pip install xformers (on linux) or pip install https://github.com/neonsecret/xformers/releases/download/v0.14/xformers-0.0.14.dev0-cp39-cp39-win_amd64.whl (on windows, python 3.9)

The project is completely open-source and free, and is only maintained by me. If you want to support me, I have a ko-fi

Warning: this requires gradio >= 3.3, be sure to install it or update it.

New update feature: added codeformer, will install itself automatically

To keep up with the newest updates, make sure to run the pip install --upgrade -r requirements.txt to get all the newest dependencies

The superfast and low-vram mode have been updated. The latest results are: 2048x2048 on 8 gb vram and 3200x3200 on 24 gb.

Below you can see the speed/resolution comparison table.

resolution steps speed_mp classic time vram low vram mode
512x512 50 default 1.5 minutes 4 no
512x512 50 default 31 seconds 8 no
512x512 50 default 28 seconds 10 no
512x512 50 default 15 seconds 24 no
1024x1024 50 default 15 minutes 4 no
1024x1024 50 default 2.3 minutes 8 no
1024x1024 50 default 3 minutes 10 no
1024x1024 50 default 70 seconds 24 no
2048x2048 50 default 25 minutes 8 no
2048x2048 50 default 20 minutes 10 no
2048x2048 50 default 15 minutes 24 no
512x4096 50 default 2 minutes 24 no
3840x2176 50 default 40 minutes 24 no
3200x3200 50 default 60 minutes 24 no

gpus used: gtx 1050 ti, rtx 3070, colab gpu, rtx 3090 (huge thanks to @therustysmear for helping me in these tests)

soft_limiter parameter limits vram usage, so that you can use your pc while generating images. 100% though allows the max sped.

How to generate so high-res images?

The default mode already allows to generate as high-res as possible images, however, if you encounter OOM errors or want to go higher in resolution, disable it:

Example cli command with txt2img and high-res mode:

python optimizedSD/optimized_txt2img.py --prompt "an apple" --config_path optimizedSD/v1-inference_lowvram.yaml --H 512 --W 512 --seed 27 --n_iter 1 --n_samples 1 --ddim_steps 50

Example gradio command:

python optimizedSD/neongradio_ultimate.py

Example gradio low-vram command:

python optimizedSD/neongradio_ultimate.py --config_path optimizedSD/v1-inference_lowvram.yaml

the --config_path optimizedSD/v1-inference_lowvram.yaml argument enables a low-vram mode which allows to generate bigger-resolution images at the slight cost of the speed.

Description

This repo is a modified version of the Stable Diffusion repo, optimized to use less VRAM than the original by sacrificing inference speed.

To achieve this, the stable diffusion model is fragmented into four parts which are sent to the GPU only when needed. After the calculation is done, they are moved back to the CPU. This allows us to run a bigger model while requiring less VRAM.

Also I invented the sliced atttention technique, which allows to push the model's abilities even further. It works by automatically determining the slice size from your vram and image size and then allocating it one by one accordingly. You can practically generate any image size, it just depends on the generation speed you are willing to sacrifice.

Installation

You can clone this repo and follow the same installation steps as the original (mainly creating the conda environment and placing the weights at the specified location).
So run:
conda env create -f environment.yaml
conda activate ldm

Additional steps for AMD Cards

After activating your conda environment, you have to update torch and torchvision wheels which were built with ROCm support (only on linux):

pip3 install --upgrade torch torchvision --extra-index-url https://download.pytorch.org/whl/rocm5.1.1

Docker

Alternatively, if you prefer to use Docker, you can do the following:

  1. Install Docker , Docker Compose plugin, and NVIDIA Container Toolkit
  2. Clone this repo to, e.g., ~/stable-diffusion
  3. Put your downloaded model.ckpt file into ~/sd-data (it's a relative path, you can change it in docker-compose.yml)
  4. cd into ~/stable-diffusion and execute docker compose up --build

This will launch gradio on port 7860 with txt2img. You can also use docker compose run to execute other Python scripts.

Usage

img2img

  • img2img can generate 512x512 images from a prior image and prompt on a 4GB VRAM GPU in under 20 seconds per image on an RTX 2060.

  • The maximum size that can fit on 6GB GPU (RTX 2060) is around 576x768.

  • For example, the following command will generate 20 512x512 images:

python optimizedSD/optimized_img2img.py --prompt "Austrian alps" --init-img ~/sketch-mountains-input.jpg --strength 0.8 --n_iter 2 --n_samples 10 --H 512 --W 512

txt2img

  • txt2img can generate 512x512 images from a prompt on a 2GB VRAM GPU in under 25 seconds per image.

  • For example, the following command will generate 20 512x512 images:

python optimizedSD/optimized_txt2img.py --prompt "Cyberpunk style image of a Telsa car reflection in rain" --H 512 --W 512 --seed 27 --n_iter 2 --n_samples 10 --ddim_steps 50

inpainting

  • Inpainting can fill masked parts of an image based on a given prompt. It can inpaint 512x512 images while using under 2GB of VRAM.

  • To launch the gradio interface for inpainting, run python optimizedSD/inpaint_gradio.py. The mask for the image can be drawn on the selected image using the brush tool.

  • The results are not yet perfect but can be improved by using a combination of prompt weighting, prompt engineering and testing out multiple values of the --strength argument.

  • Suggestions to improve the inpainting algorithm are most welcome.

img2img interpolation

  • img2img_interpolate.py creates an animation of image transformation using a text prompt

  • To launch the gradio interface for inpainting, run python optimizedSD/img2img_interpolate.py. The mask for the image can be drawn on the selected image using the brush tool.

  • The results are not yet perfect but can be improved by using a combination of prompt weighting, prompt engineering and testing out multiple values of the --strength argument.

Using the Gradio GUI

  • You can also use the built-in gradio interface for img2img, txt2img & inpainting instead of the command line interface. Activate the conda environment and install the latest version of gradio using pip install gradio,

  • Run the ultimate UI using python optimizedSD/neongradio_ultimate.py. All features available on one tab.

  • img2img has a feature to crop input images. Look for the pen symbol in the image box after selecting the image.

Arguments

--seed

Seed for image generation, can be used to reproduce previously generated images. Defaults to a random seed if unspecified.

  • The code will give the seed number along with each generated image. To generate the same image again, just specify the seed using --seed argument. Images are saved with its seed number as its name by default.

  • For example if the seed number for an image is 1234 and it's the 55th image in the folder, the image name will be named seed_1234_00055.png.

--n_samples

Batch size/amount of images to generate at once.

  • To get the lowest inference time per image, use the maximum batch size --n_samples that can fit on the GPU. Inference time per image will reduce on increasing the batch size, but the required VRAM will increase.

  • If you get a CUDA out of memory error, try reducing the batch size --n_samples. If it doesn't work, the other option is to reduce the image width --W or height --H or both.

--n_iter

Run x amount of times

  • Equivalent to running the script n_iter number of times. Only difference is that the model is loaded only once per n_iter iterations. Unlike n_samples, reducing it doesn't have an effect on VRAM required or inference time.

--H & --W

Height & width of the generated image.

  • Both height and width should be a multiple of 64.

--turbo

Increases inference speed at the cost of extra VRAM usage.

  • Using this argument increases the inference speed by using around 1GB of extra GPU VRAM. It is especially effective when generating a small batch of images (~ 1 to 4) images. It takes under 25 seconds for txt2img and 15 seconds for img2img (on an RTX 2060, excluding the time to load the model). Use it on larger batch sizes if GPU VRAM available.

--precision autocast or --precision full

Whether to use full or mixed precision

  • Mixed Precision is enabled by default. If you don't have a GPU with tensor cores (any GTX 10 series card), you may not be able use mixed precision. Use the --precision full argument to disable it.

--format png or --format jpg

Output image format

  • The default output format is png. While png is lossless, it takes up a lot of space (unless large portions of the image happen to be a single colour). Use lossy jpg to get smaller image file sizes.

--unet_bs

Batch size for the unet model

  • Takes up a lot of extra RAM for very little improvement in inference time. unet_bs > 1 is not recommended!

  • Should generally be a multiple of 2x(n_samples)

Weighted Prompts

  • Prompts can also be weighted to put relative emphasis on certain words. eg. --prompt tabby cat:0.25 white duck:0.75 hybrid.

  • The number followed by the colon represents the weight given to the words before the colon. The weights can be both fractions or integers.