yamind-image-ultra-image

name: yamind-image-ultra-image description: "Free local image generation via ComfyUI + Stable Diffusion. Use when: (1) Generating images from text prompts, (2) Image-to-image editing, (3) Upscaling images, (4) Style transfer, (5) Creating book covers, posters, illustrations, (6) Video generation, (7) Batch image processing. No API keys needed - runs locally. Falls back to Pollinations.ai cloud if ComfyUI unavailable."

Ultra Image - ComfyUI Integration

Free, local image generation using ComfyUI and Stable Diffusion models.

Quick Start

from scripts.comfyui_client import generate_image

result = generate_image(
    prompt="A mystical forest at sunset, digital art",
    model="sdxl",
    output_dir="~/Pictures/generated"
)

Available Models

Capabilities

Text-to-Image

python scripts/comfyui_client.py --prompt "cyberpunk city" --model sdxl

Image-to-Image

Load an image and modify it with a prompt. Uses img2img workflow.

Upscaling

Use ESRGAN/SwinIR models to upscale images 2x-4x.

Batch Generation

Generate multiple variations with different seeds.

Style Transfer

Apply artistic styles using LoRA or style models.

ControlNet

Guide generation with depth, pose, or edge maps.

ComfyUI API

Check Status

curl http://127.0.0.1:8188/system_stats

Queue Prompt

import json, urllib.request
workflow = json.load(open("workflow.json"))
resp = urllib.request.urlopen(
    "http://127.0.0.1:8188/prompt",
    json.dumps({"prompt": workflow}).encode()
)
print(json.loads(resp.read())["prompt_id"])

WebSocket Progress

import websocket
ws = websocket.WebSocket()
ws.connect("ws://127.0.0.1:8188/ws")
while True:
    msg = ws.recv()
    print(msg)  # Progress updates

Fallback: Pollinations.ai

If ComfyUI unavailable, automatically fallback to Pollinations.ai (free, no API key):

from scripts.comfyui_client import pollinations_fallback
result = pollinations_fallback("abstract art", 1024, 1024, "~/Pictures", "output.png")

Workflow Templates

Templates are in assets/:

• text_to_image_sdxl.json - SDXL text-to-image
• text_to_image_flux.json - Flux text-to-image
• image_to_image_sdxl.json - Image-to-image editing
• upscale_esrgan.json - Image upscaling

To create custom workflows:

1. Build workflow in ComfyUI GUI
2. Enable Dev mode (Settings → Dev)
3. Export as API JSON
4. Save to assets/

Sampling Parameters

Recommended Samplers

• Quality: dpmpp_2m_sde, dpmpp_3m_sde
• Speed: euler, euler_ancestral
• Turbo models: euler_ancestral

Integration with YaMac

Telegram

/do generate image: a sunset over mountains

Model Routing

Configured via ultra-image backend in ~/yamac-core/config/codex-model-routing.json.

Skill Routing

Auto-selected for tasks containing: image, generate, art, cover, poster, illustration.

References

• references/comfyui-capabilities.md - Full capability reference
• references/comfyui_workflows.md - Workflow examples
• references/cloud_providers.md - Cloud fallback options

Requirements

• ComfyUI Desktop installed (or running server)
• Models downloaded to ComfyUI models folder
• Python with Pillow (installed)

Troubleshooting

ComfyUI not running

• Open ComfyUI.app from Applications
• Or start server: python main.py in ComfyUI directory

Model not found

• Download from HuggingFace or Civitai
• Place in ~/ComfyUI/models/checkpoints/

Out of memory

• Use --lowvram flag
• Use smaller models (Turbo variants)
• Reduce resolution