feat: use sdcpp-webui as embedded webui

[leejet]

.\bin\Release\sd-server.exe --diffusion-model  ..\models\diffusion_models\z_image_turbo_bf16.safetensors --vae ..\models\vae\ae.sft  --llm ..\models\text_encoders\qwen_3_4b.safetensors --diffusion-fa --offload-to-cpu -v --cfg-scale 1.0

Visit http://127.0.0.1:1234/

[mags0ft]

Unfortunately, I get this output:

[ERROR] common.cpp:277  - error: unknown argument: --cfg-scale 1.0
stable-diffusion.cpp version unknown, commit c41c5de

Removing the argument and visiting the website gives me a plain, blank page saying

Stable Diffusion Server is running

Is this correct?! Using the master:cuda Docker build.

My configuration is:

services:
  stable-diffusion-cpp:
    image: ghcr.io/leejet/stable-diffusion.cpp:master-cuda
    container_name: stable-diffusion-cpp
    restart: always
    ports:
      - "3211:1234"
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility
    volumes:
      - ./models:/models
      - ./output:/output
    runtime: nvidia
    entrypoint:
      - ./sd-server
      - --diffusion-model
      - /models/z-image/z-image-turbo-Q6_K.gguf
      - --vae
      - /models/z-image/vae.safetensors
      - --llm
      - /models/.cache/models--unsloth--Qwen3-VL-8B-Thinking-GGUF/snapshots/894774171339940a7a7f162a39d8143fb5154cdd/Qwen3-VL-8B-Thinking-Q6_K.gguf
      - -v
      - -l
      - 0.0.0.0
      - -p
      - "1234"
      - --offload-to-cpu
      - --diffusion-fa

[mags0ft]

Okay, I was able to build the index.html file (for anyone too tired to do it themselves, here is a build, but don't take my word on it, may be outdated by the time you read this!) and attached it to the web server by using the --serve-html-path /.../index.html command line argument. Now I get the web UI, but it says Service unavailable.

Upon closer investigation, this seems to be occuring due to this error message:

filesystem error: status: Operation not permitted [./proc/1/map_files/5d40ded2f000-5d40ded5c000]

Weird.

[mags0ft]

Wait, why is it trying to mess with /proc/1/?

Maybe memory mapping...