Does Prompt Director work with Stable Diffusion?

Yes. Copy the positive prompt into prompt and the negative prompt into negative_prompt in Automatic1111, Forge, ReForge, or another compatible Stable Diffusion WebUI.

Does Prompt Director work with ComfyUI?

Yes. Use the positive and negative outputs in the matching CLIP Text Encode nodes and apply the recommended sampler, scheduler, steps, CFG, and dimensions.

For humans and agents

Turn an idea or reference image into a generation-ready prompt package.

Build positive and negative prompts for ComfyUI, Stable Diffusion WebUI, anime models, and realistic checkpoints—or reverse an existing image into a recreation prompt with vision + Prompt Director.

New · Image → prompt

Reference image URL Optional direction (how to restyle / change it)

Vision step costs 2 flexible credits per image (Pro/Builder monthly credits apply first — no free FAL calls). Prompt packaging is free; Advanced Optimizer is +1 credit when it succeeds.

Describe the image (or paste a vision description)

Medium Aspect ratio Subjects Framing

Advanced target settings optional

Model family Interface Conditioning Checkpoint

LoRAs and trigger words

Optional style direction Optional things to avoid

Standard is free. Advanced uses two-stage DeepSeek optimization plus a critic audit and costs 1 credit only when both stages succeed.

✦

Your prompt package will appear here

Choose anime or realistic output, describe the scene, and generate copy-ready prompts for humans or JSON for agents.

Works with your workflow

ComfyUI and Stable Diffusion ready

ComfyUI

Paste the outputs into positive and negative CLIP Text Encode nodes, then use the recommended KSampler settings and dimensions.

Stable Diffusion WebUI

Works with Automatic1111, Forge, and ReForge: use prompt, negative_prompt, DPM++ 2M, and Karras.

REST API

Use the same director from an agent

curl -X POST https://agentmediatools.com/api/prompt-director \
  -H "Content-Type: application/json" \
  -d '{"request":"A cinematic cyberpunk heroine","medium":"anime","aspect_ratio":"portrait","subject_count":1}'

# Reverse an image into a generation package (2 credits after free daily describe)
curl -X POST https://agentmediatools.com/api/prompt-director/from-image \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com/ref.png","medium":"anime","direction":"same pose, soft studio light"}'

Common questions

Prompt Director FAQ

Does it work with Stable Diffusion?

Yes. The prompt text is model-portable and the result includes explicit Stable Diffusion WebUI field and sampler mappings.

Does it work with ComfyUI?

Yes. Use the positive and negative prompts in CLIP Text Encode nodes and copy the recommended KSampler settings.

Can I turn a photo into a generation prompt?

Yes. Paste a public image URL and use Image → prompt package. Vision describe costs 2 flexible credits per image (no free FAL calls; Pro monthly credits apply). Agents: POST /api/prompt-director/from-image or MCP image_to_prompt.

Can agents and third-party apps use it?

Yes. Call POST /api/prompt-director over REST, POST /api/prompt-director/from-image for reverse prompts, or use the MCP tools direct_image_prompt / image_to_prompt.