⚡ REFERENCE IMAGE → PRODUCTION-READY PROMPTS

Prompt Studio Suite v3.2 is a reference-image analysis and multimodal prompt-generation workflow for ComfyUI. Load a reference image, optionally add creative direction, and generate a production-ready still-image prompt or structured MiniMax H3 video script.

The still-image engine can tailor its prompt-writing strategy for Generic, FLUX, Krea 2, or SDXL.

---

### ◈ KEY CAPABILITIES

🔍 Reference Image & Aspect-Ratio Analysis

Analyzes the reference image while automatically detecting its aspect ratio and feeding composition context into the prompt-generation engines.

🎨 Still Prompt Studio — RECREATE / DIRECT

Generate detailed still-image prompts in two ways:

- RECREATE: Leave Still Direction blank to reverse-engineer the reference as faithfully as possible.

- DIRECT: Add Still Direction to modify the scene while retaining useful visual context from the reference.

The still engine analyzes subjects, appearance, pose, wardrobe and materials, environment, composition, lighting, color, atmosphere, camera characteristics, optics, depth of field, and other visually relevant details.

🎯 Target Model Prompting

Set the Target Model to:

- Generic — balanced, model-agnostic natural-language prompting

- FLUX — coherent, literal descriptions emphasizing spatial relationships, materials, lighting, and camera perspective

- Krea 2 — visually focused art direction emphasizing composition, mood, lighting, texture, color, and photographic/cinematic intent

- SDXL — structured, keyword-aware prompting with explicit visual and technical detail

These presets change how the LLM writes the prompt rather than simply appending a model name.

🎬 MiniMax H3 Video Scripting

The independent video track creates structured multimodal production scripts containing:

- integrated_multimodal_description — numbered shots, visual action, camera direction, movement, continuity, and timing

- overall_soundscape — synchronized diegetic sound and environmental audio

- non_diegetic_music — external score, mood, and musical pacing

🎛️ Independent Still & Video Direction

Still Direction and Video Direction are completely separate.

You can recreate the reference image for the still track while giving entirely different direction to the video track, direct both independently, or leave either direction blank.

🔀 Fast Track Switching

Powered by rgthree Fast Bypasser so the still and video engines can be enabled or disabled without rewiring the workflow.

🔒 Centralized API-Key Input

OpenRouter credentials are entered through a dedicated API-key node.

Never upload or distribute a workflow after entering your personal API key without first removing it.

---

### 📦 REQUIRED NODES

Install these extensions through ComfyUI Manager before running Prompt Studio:

- ComfyUI-Custom-Scripts (pythongosssss)

- ComfyUI-KJNodes

- rgthree-comfy

- openrouter_node

An OpenRouter account/API key is required. API usage may incur charges depending on the models used and your OpenRouter account.

---

### 🚀 QUICK SETUP

1. Load prompt_studio_v3.json into ComfyUI.

2. Enter your OpenRouter API key in OpenRouter API Key.

3. Load an image into Reference Image Input.

4. For still-image prompting, enter Generic, FLUX, Krea 2, or SDXL in Target Model.

5. Leave Still Direction blank for RECREATE, or add instructions for DIRECT mode.

6. Add independent Video Direction if generating a MiniMax H3 video script.

7. Enable the desired track under Studio Toggle Switches.

8. Click Queue Prompt.

---

### ◈ LINKS

GitHub Repository:

https://github.com/phibbyrizz/ComfyUI-Prompt-Studio

If you find Prompt Studio useful and want to support continued development:

[Support Phibby](https://ko-fi.com/phibby)