comfyui-mcp
Version:
Local-first, agent-native control plane for ComfyUI — MCP server + autonomous sidebar agent that drives your live graph in natural language on ANY LLM: Claude/ChatGPT/Gemini on your subscription (no API key), free local models via Ollama (fully offline),
31 lines (30 loc) • 1.37 kB
YAML
name: ideogram-txt2img
display_name: Ideogram 4 — Text to Image
family: ideogram
kind: txt2img
description: >-
Ideogram 4 (Ideogram Ultra) text-to-image, running LOCALLY from the Comfy-Org
fp8 DiT weights (NOT the Ideogram cloud API). Pairs a Qwen3-VL text encoder
with a Gemma4 prompt-builder LLM (JSON Prompt Builder subgraph) and a dual
conditional/unconditional UNet guider for strong prompt adherence and crisp
TEXT RENDERING. Single-purpose txt2img graph sliced from the ideogram monolith
(one pipeline, no group toggles), with optional KJ image sharpening and an
off-by-default Power Lora Loader.
vram: "<8GB"
workflow: workflow.json
skill: ideogram-txt2img
launch_args: []
sources:
model: https://huggingface.co/Comfy-Org/Ideogram-4
mirror: https://huggingface.co/Aitrepreneur/FLX
notes:
- Run the generated installer from your ComfyUI root (folder with custom_nodes/
and models/).
- Local weights only — this does NOT call the Ideogram cloud API; the fp8
Ideogram-4 DiT is loaded via UNETLoader and sampled with SamplerCustomAdvanced.
- Sliced + un-bypassed from the monolith's "TEXT TO IMAGE" group.
- Ideogram excels at literal text/typography — write the exact string in quotes.
post_install:
- Restart ComfyUI (or ComfyUI-Manager -> Install missing custom nodes for any
node Python deps).
- Load workflow.json.