stable-diffusion-webui/requirements.txt

98 lines
2.4 KiB
Plaintext
Raw Normal View History

-e .
# See: https://github.com/CompVis/taming-transformers/issues/176
# -e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers # required by ldm
# Note: taming package needs to be installed with -e option
-e git+https://github.com/CompVis/taming-transformers#egg=taming-transformers
invisible-watermark==0.1.5
taming-transformers-rom1504==0.0.6 # required by ldm
# Note: K-diffusion brings in CLIP 1.0 as a dependency automatically; will create a dependency resolution conflict when explicitly specified together
git+https://github.com/openai/CLIP.git@main#egg=clip
git+https://github.com/crowsonkb/k-diffusion.git
# git+https://github.com/hlky/k-diffusion-sd#egg=k_diffusion
# Dependencies required for Stable Diffusion UI
pynvml==11.4.1
omegaconf==2.2.3
# Note: Jinja2 3.x major version required due to breaking changes found in markupsafe==2.1.1; 2.0.1 is incompatible with other upstream dependencies
# see https://github.com/pallets/markupsafe/issues/304
Jinja2==3.1.2 # Jinja2 is required by Gradio
# Environment Dependencies for WebUI (gradio)
gradio==3.4.1
# Environment Dependencies for WebUI (streamlit)
streamlit==1.13.0
streamlit-on-Hover-tabs==1.0.1
streamlit-option-menu==0.3.2
streamlit_nested_layout==0.1.1
streamlit-server-state==0.14.2
streamlit-tensorboard==0.0.2
hydralit==1.0.14
hydralit_components==1.0.10
stqdm==0.0.4
2022-10-15 08:34:44 +03:00
decorest==0.1.0
# txt2vid
stable-diffusion-videos==0.5.3
diffusers==0.4
librosa==0.9.2
# img2img inpainting
streamlit-drawable-canvas==0.9.2
# Img2text
ftfy==6.1.1
fairscale==0.4.4
regex
Scene-to-Image Prompt Layering System (#1179) # Summary of the change - new Scene-to-Image tab - new scn2img function - functions for loading and running monocular_depth_estimation with tensorflow # Description (relevant motivation, which issue is fixed) Related to discussion #925 > Would it be possible to have a layers system where we could do have foreground, mid, and background objects which relate to one another and share the style? So we could say generate a landscape, one another layer generate a castle, and on another layer generate a crowd of people. To make this work I made a prompt-based layering system in a new "Scene-to-Image" tab. You write a a multi-line prompt that looks like markdown, where each section declares one layer. It is hierarchical, so each layer can have their own child layers. Examples: https://imgur.com/a/eUxd5qn ![](https://i.imgur.com/L61w00Q.png) In the frontend you can find a brief documentation for the syntax, examples and reference for the various arguments. Here a short summary: Sections with "prompt" and child layers are img2img, without child layers they are txt2img. Without "prompt" they are just images, useful for mask selection, image composition, etc. Images can be initialized with "color", resized with "resize" and their position specified with "pos". Rotation and rotation center are "rotation" and "center". Mask can automatically be selected by color or by estimated depth based on https://huggingface.co/spaces/atsantiago/Monocular_Depth_Filter. ![](https://i.imgur.com/8rMHWmZ.png) # Additional dependencies that are required for this change For mask selection by monocular depth estimation tensorflow is required and the model must be cloned to ./src/monocular_depth_estimation/ Changes in environment.yaml: - einops>=0.3.0 - tensorflow>=2.10.0 Einops must be allowed to be newer for tensorflow to work. # Checklist: - [x] I have changed the base branch to `dev` - [x] I have performed a self-review of my own code - [x] I have commented my code in hard-to-understand areas - [x] I have made corresponding changes to the documentation Co-authored-by: hlky <106811348+hlky@users.noreply.github.com>
2022-10-02 20:23:37 +03:00
timm==0.6.7
tqdm==4.64.0
tensorboard==2.10.1
# Other
retry==0.9.2 # used by sd_utils
python-slugify==6.1.2 # used by sd_utils
piexif==1.1.3 # used by sd_utils
accelerate==0.12.0
albumentations==0.4.3
einops==0.3.1
facexlib>=0.2.3
imageio-ffmpeg==0.4.2
imageio==2.9.0
kornia==0.6
loguru
opencv-python-headless==4.6.0.66
open-clip-torch==2.0.2
pandas==1.4.3
pudb==2019.2
pytorch-lightning==1.7.7
realesrgan==0.3.0
test-tube>=0.7.5
timm==0.6.7
torch-fidelity==0.3.0
transformers==4.19.2 # do not change
wget
# Optional packages commonly used with Stable Diffusion workflow
# Upscalers
basicsr==1.4.2 # required by RealESRGAN
gfpgan==1.3.8 # GFPGAN
realesrgan==0.3.0 # RealESRGAN brings in GFPGAN as a requirement
git+https://github.com/CompVis/latent-diffusion
Scene-to-Image Prompt Layering System (#1179) # Summary of the change - new Scene-to-Image tab - new scn2img function - functions for loading and running monocular_depth_estimation with tensorflow # Description (relevant motivation, which issue is fixed) Related to discussion #925 > Would it be possible to have a layers system where we could do have foreground, mid, and background objects which relate to one another and share the style? So we could say generate a landscape, one another layer generate a castle, and on another layer generate a crowd of people. To make this work I made a prompt-based layering system in a new "Scene-to-Image" tab. You write a a multi-line prompt that looks like markdown, where each section declares one layer. It is hierarchical, so each layer can have their own child layers. Examples: https://imgur.com/a/eUxd5qn ![](https://i.imgur.com/L61w00Q.png) In the frontend you can find a brief documentation for the syntax, examples and reference for the various arguments. Here a short summary: Sections with "prompt" and child layers are img2img, without child layers they are txt2img. Without "prompt" they are just images, useful for mask selection, image composition, etc. Images can be initialized with "color", resized with "resize" and their position specified with "pos". Rotation and rotation center are "rotation" and "center". Mask can automatically be selected by color or by estimated depth based on https://huggingface.co/spaces/atsantiago/Monocular_Depth_Filter. ![](https://i.imgur.com/8rMHWmZ.png) # Additional dependencies that are required for this change For mask selection by monocular depth estimation tensorflow is required and the model must be cloned to ./src/monocular_depth_estimation/ Changes in environment.yaml: - einops>=0.3.0 - tensorflow>=2.10.0 Einops must be allowed to be newer for tensorflow to work. # Checklist: - [x] I have changed the base branch to `dev` - [x] I have performed a self-review of my own code - [x] I have commented my code in hard-to-understand areas - [x] I have made corresponding changes to the documentation Co-authored-by: hlky <106811348+hlky@users.noreply.github.com>
2022-10-02 20:23:37 +03:00
## for monocular depth estimation
tensorflow==2.10.0
# Unused Packages: No current usage but will be used in the future.
# Orphaned Packages: No usage found
Scene-to-Image Prompt Layering System (#1179) # Summary of the change - new Scene-to-Image tab - new scn2img function - functions for loading and running monocular_depth_estimation with tensorflow # Description (relevant motivation, which issue is fixed) Related to discussion #925 > Would it be possible to have a layers system where we could do have foreground, mid, and background objects which relate to one another and share the style? So we could say generate a landscape, one another layer generate a castle, and on another layer generate a crowd of people. To make this work I made a prompt-based layering system in a new "Scene-to-Image" tab. You write a a multi-line prompt that looks like markdown, where each section declares one layer. It is hierarchical, so each layer can have their own child layers. Examples: https://imgur.com/a/eUxd5qn ![](https://i.imgur.com/L61w00Q.png) In the frontend you can find a brief documentation for the syntax, examples and reference for the various arguments. Here a short summary: Sections with "prompt" and child layers are img2img, without child layers they are txt2img. Without "prompt" they are just images, useful for mask selection, image composition, etc. Images can be initialized with "color", resized with "resize" and their position specified with "pos". Rotation and rotation center are "rotation" and "center". Mask can automatically be selected by color or by estimated depth based on https://huggingface.co/spaces/atsantiago/Monocular_Depth_Filter. ![](https://i.imgur.com/8rMHWmZ.png) # Additional dependencies that are required for this change For mask selection by monocular depth estimation tensorflow is required and the model must be cloned to ./src/monocular_depth_estimation/ Changes in environment.yaml: - einops>=0.3.0 - tensorflow>=2.10.0 Einops must be allowed to be newer for tensorflow to work. # Checklist: - [x] I have changed the base branch to `dev` - [x] I have performed a self-review of my own code - [x] I have commented my code in hard-to-understand areas - [x] I have made corresponding changes to the documentation Co-authored-by: hlky <106811348+hlky@users.noreply.github.com>
2022-10-02 20:23:37 +03:00