comfyui-structured-image-prompt

ComfyUI Structured Image Prompt

A small ComfyUI custom node for building strong image prompts from separate, structured fields:

• style
• camera angle
• lighting
• background
• characters
• action
• clothing
• assets
• quality tags
• negative prompt

The node is deterministic and does not require an LLM or external Python packages.

Node

Display name:

IC Structured Image Prompt

Category:

IC/Prompting

Outputs:

• prompt
• negative_prompt
• character_summary
• warnings

Inputs

The ComfyUI node uses english lower-case widget names and visible field prefixes inside the default text values:

• style
• camera angle
• lighting
• background
• characters
• action
• clothing
• assets
• quality tags
• negative prompt
• prompt format
• include labels
• check character refs
• show debug

ComfyUI can hide multiline widget names in the expanded node body. To keep the fields readable, new nodes start their text values with prefixes like style:, camera angle:, characters:, and negative prompt:. These prefixes are stripped automatically and will not appear in the generated prompt.

Character References

Use a YAML-like character block:

Mira: young rogue mage, short silver hair, confident expression
Oskar: old mechanic, heavy beard, tired eyes

You can then reference characters in action, clothing, and assets:

action:
[Mira] watches the rooftops while [Oskar] repairs a small flying drone

clothing:
Mira: black tactical coat, blue scarf, leather boots
Oskar: worn orange work jacket, welding gloves

assets:
Mira: engraved wand, glowing wrist charm
Oskar: toolbox, brass repair drone

Unknown character references are reported through the warnings output when strict_character_refs is enabled.

Installation

Clone this repository into your ComfyUI custom_nodes folder:

cd C:\path\to\ComfyUI\custom_nodes
git clone https://github.com/YOUR-USER/comfyui-structured-image-prompt.git

Restart ComfyUI. The node will appear under IC/Prompting.

Prompt Formats

The prompt_format input supports:

• natural
• tagged
• cinematic
• sdxl
• flux

natural and cinematic generate clear director-style prompts with explicit sentences for scene action, character details, camera framing, lighting, visual style, and quality cues.

sdxl generates a tag-oriented prompt with SDXL-style quality anchors such as masterpiece, best quality, highly detailed, and visually coherent.

flux generates a natural-language prompt with cleaner descriptive sentences and avoids classic SDXL quality boilerplate.

The tagged format remains comma-separated for tag-heavy workflows.

prompt_prefix and prompt_suffix can be used for model-specific tokens, LoRA triggers, or house style tags.

License

MIT