Dockerfile

ARG BASE_IMAGE
FROM ${BASE_IMAGE}

SHELL ["/bin/bash", "-o", "pipefail", "-c"]
ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONUNBUFFERED=1 \
    PIP_NO_CACHE_DIR=on \
    SHELL=/bin/bash

RUN mkdir -p /sd-models

# Add SDXL models and VAE
# These need to already have been downloaded:
#   wget https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/realisticVisionV51_v51VAE.safetensors
COPY realisticVisionV51_v51VAE.safetensors /sd-models/realisticVisionV51_v51VAE.safetensors

# Create and use the Python venv
RUN python3 -m venv /venv

# Clone the git repo of Stable Diffusion WebUI Forge and set version
ARG FORGE_COMMIT
RUN git clone https://github.com/lllyasviel/stable-diffusion-webui-forge.git && \
    cd /stable-diffusion-webui-forge

# Install the dependencies for Stable Diffusion WebUI Forge
ARG INDEX_URL
ARG TORCH_VERSION
ARG XFORMERS_VERSION
WORKDIR /stable-diffusion-webui-forge
ENV TORCH_INDEX_URL=${INDEX_URL}
ENV TORCH_COMMAND="pip install torch==${TORCH_VERSION} torchvision --index-url ${TORCH_INDEX_URL}"
ENV XFORMERS_PACKAGE="xformers==${XFORMERS_VERSION} --index-url ${TORCH_INDEX_URL}"
RUN source /venv/bin/activate && \
    ${TORCH_COMMAND} && \
    pip3 install -r requirements_versions.txt && \
    pip3 install ${XFORMERS_PACKAGE} &&  \
    deactivate

# Install the dependencies for the built-in extensions
RUN source /venv/bin/activate && \
    pip3 install -r extensions-builtin/sd_forge_controlnet/requirements.txt && \
    pip3 install -r extensions-builtin/forge_legacy_preprocessors/requirements.txt && \
    pip3 install insightface && \
    pip3 uninstall -y onnxruntime && \
    pip3 install onnxruntime-gpu && \
    pip3 install protobuf==3.20.0 && \
    deactivate

COPY forge/cache-sd-model.py ./
RUN source /venv/bin/activate && \
    python3 -c "from launch import prepare_environment; prepare_environment()" --skip-torch-cuda-test && \
    deactivate

# Cache the Stable Diffusion Models
# SDXL models result in OOM kills with 8GB system memory, need 30GB+ to cache these
    #RUN source /venv/bin/activate && \
    #    python3 cache-sd-model.py --no-half-vae --no-half --xformers --use-cpu=all --ckpt /sd-models/realisticVisionV51_v51VAE.safetensors && \
    #    deactivate

# Copy Stable Diffusion WebUI Forge config files
COPY forge/relauncher.py forge/webui-user.sh forge/config.json forge/ui-config.json /stable-diffusion-webui-forge/

# ADD SDXL styles.csv
ADD https://raw.githubusercontent.com/Douleb/SDXL-750-Styles-GPT4-/main/styles.csv /stable-diffusion-webui/styles.csv

# Install CivitAI Model Downloader
ARG CIVITAI_DOWNLOADER_VERSION
RUN git clone https://github.com/ashleykleynhans/civitai-downloader.git && \
    cd civitai-downloader && \
    git checkout tags/${CIVITAI_DOWNLOADER_VERSION} && \
    cp download.py /usr/local/bin/download-model && \
    chmod +x /usr/local/bin/download-model && \
    cd .. && \
    rm -rf civitai-downloader

# Remove existing SSH host keys
RUN rm -f /etc/ssh/ssh_host_*

# NGINX Proxy
COPY nginx/nginx.conf /etc/nginx/nginx.conf

# Set template version
ARG RELEASE
ENV TEMPLATE_VERSION=${RELEASE}

# Set the venv path
ARG VENV_PATH
ENV VENV_PATH=${VENV_PATH}

# Copy the scripts
WORKDIR /
COPY --chmod=755 scripts/* ./

# Start the container
SHELL ["/bin/bash", "--login", "-c"]
CMD [ "/start.sh" ]