DCP
Open Navigation

Docs

ENAR

Stable Diffusion XL Base 1.0

SDXL image generation deployment guidance on DCP.

1. What it is

SDXL Base 1.0 (`stabilityai/stable-diffusion-xl-base-1.0`) is a text-to-image diffusion model from Stability AI.

2. What it does

It generates high-quality images from prompts and is suited for creative, marketing, and prototyping workloads.

3. How it compares

  • Versus SD 1.x: better composition and prompt fidelity, with higher VRAM and cold-start cost.
  • Versus LLMs: SDXL is image generation, not text reasoning/chat.

4. Best for on DCP

  • Arabic and English marketing visuals
  • Concept art and product mockups
  • Batch image generation pipelines

5. Hardware requirements on DCP

  • DCP floor: `min_vram_gb: 16` (Tier B diffusion profile)
  • Recommended providers: RTX 3090/4090 or better
  • Template: `stable-diffusion`

6. How to run on DCP

  1. Use template `stable-diffusion`.
  2. Set `params.model` to `stabilityai/stable-diffusion-xl-base-1.0`.
  3. Include `steps`, `width`, `height`, and prompt in the job payload.

7. Licensing and commercial-use notes

SDXL model card publishes CreativeML Open RAIL++-M style terms with usage restrictions. Review the license and use restrictions before public or commercial deployment.

Sources:

  • https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
  • /home/node/dc1-platform/docker-templates/stable-diffusion.json
  • /home/node/dc1-platform/infra/config/arabic-portfolio.json