Eliora Infinite Canvas

Eliora Infinite CanvasLocal-first · llama.cpp + ComfyUI

ComfyUI Eliora Miragea

Palette

Board

0cards

0links

Local AI · llama.cpp

llama.cpp ComfyUI

Default path. Start llama-server on 127.0.0.1:8080. Use Ask Local or Ctrl+Enter.

Endpoint

Model

Key

System

Local Image · ComfyUI

GPU images & video. Start ComfyUI on 127.0.0.1:8188. Use Send Comfy or Ctrl+Shift+Enter.

Endpoint

Local UNET (diffusion_models) Image models only

Connect ComfyUI to list your diffusion_models.

Presets · CLIP/VAE · VRAM

Quick preset

CLIP override

VAE override

Weight dtype (VRAM)

Pick a UNET above. Overrides help large-GPU local runs. Full graphs: Open ComfyUI.

Negative

Width

Height

API workflow JSON

Inject canvas context

Open ComfyUI

Empty workflow = auto text-to-image.

AI Mode

Local by default. Switch to cloud only when away.

Workflow Import

Type

Show

Add ComfyUI workflow JSON via Import JSON.

Cloud API · BYOK

Eliora does not proxy. Keys stay on this device.

BYOK only — OpenRouter / OpenAI-compatible. Keys stay on device. No China-specific services.

Chat endpoint

Model

API Key (BYOK)

System

File

Select a card

100% Select No selection Saved

Ready

Log

カードを置く