skills.rest
artokun
artokunCOMMUNITY
Explore all skills by this author →
Website𝕏@artokunGitHub

ernie-image

Builds ComfyUI workflows for Baidu ERNIE-Image text-to-image generation and refinement.

715|111|Updated Feb 15, 2026
Search Tags:#image generation#text-to-image#Workflow#gguf#ComfyUI#baidu#ernie-image
One-click install
npx skills add https://github.com/artokun/comfyui-mcp --skill ernie-image
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ernie-image
Source: https://github.com/artokun/comfyui-mcp/tree/main/plugin/skills/ernie-image
Command: npx skills add https://github.com/artokun/comfyui-mcp --skill ernie-image

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Configuring Baidu ERNIE-Image in ComfyUI requires knowing the exact model files, node types, sampler settings, and wiring conventions, and the shipped pack graphs are large, group-toggled, and confusing to read. This Skill provides verified model lists, node dependencies, sampler settings, and ready-to-use API-format workflows so you can generate images with ERNIE-Image without reverse-engineering the source graphs.

Core Features & Use Cases

  • Text-to-Image Workflows: Complete API-format ComfyUI graphs for ERNIE-Image-Turbo with correct GGUF loaders, Ministral-3-3B text encoder, Flux 2 VAE, and AuraFlow shift settings.
  • Image Refinement Pipelines: Denoise-based image-to-image refine and ERNIE × Z-Image Turbo two-pass combo pipelines with post-processing (film grain, sharpening, upscale).
  • Model & Node Installation Guidance: Exact download URLs, quant selection by VRAM, and the custom node packs required (ComfyUI-GGUF, rgthree-comfy, KJNodes, and others).
  • Use Case: You want a poster with precise multilingual text rendering. Use this Skill to build an ERNIE-Image-Turbo workflow at 1920×1088 with 9 steps, cfg 1, and a prompt quoting the exact text to render.

Quick Start

Use the ernie-image skill to build a ComfyUI text-to-image workflow that renders a vintage travel poster with the title text "KYOTO" at the top.

Frequently Asked Questions about ernie-image

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build an ERNIE-Image text-to-image workflow in ComfyUI?▼

Load the ERNIE-Image-Turbo GGUF with UnetLoaderGGUF, the ministral-3-3b encoder with CLIPLoader set to type flux2, and the flux2-vae with VAELoader. Apply ModelSamplingAuraFlow with shift 3.1, then run KSampler at 8-9 steps, cfg 1, euler sampler, simple scheduler.

What is ERNIE-Image good for compared to Qwen-Image-Edit or Flux Kontext?▼

ERNIE-Image excels at text-to-image with precise multilingual text rendering, posters, signage, and manga multi-panel layouts. It does not perform instruction-grounded editing; use Qwen-Image-Edit or Flux Kontext for tasks like changing objects in an existing photo.

Which GGUF quant should I download for my GPU VRAM?▼

The installer guidance recommends Q5_K_S for GPUs under 8 GB, Q6_K for 8 to 12 GB, and Q8_0 for 12 to 16 GB or more. The Ministral-3-3B encoder and Flux 2 VAE add a few extra GB on top of the UNet.

Can ERNIE-Image edit an existing photo based on instructions?▼

No. ERNIE's image-to-image mode is plain denoise-based refinement at low denoise values around 0.35 to 0.4, which restyles or refines a single source image. It does not follow edit instructions like changing or removing objects.

Why does CLIPLoader fail when loading the ministral text encoder?▼

The CLIPLoader type must be set to flux2 for ministral-3-3b, not qwen_image or lumina2. The lumina2 type belongs to the Z-Image Qwen3 encoder bundled in the combo pack, not to ERNIE.

What custom nodes are required for ERNIE-Image workflows in ComfyUI?▼

Required packs include ComfyUI-GGUF for GGUF loaders, rgthree-comfy for Power Lora Loader and group bypassers, ComfyUI_essentials for ImageResize+, plus ComfyUI-Easy-Use, KJNodes, wlsh_nodes, and comfyui-vrgamedevgirl for post-processing nodes.

Related Skills

Explore More Design & Creative Skills→
jackspacejackspace
16

comfyui-workflow-helper

Guide ComfyUI workflow creation, node configuration, and optimization.

Community
Advanced
erfianugraherfianugrah
1

comfyui

Generate images by submitting ComfyUI workflows through a localhost llm-compose proxy.

Community
Intermediate
CesarG503CesarG503

threejs-postprocessing

Apply post-processing effects to Three.js scenes with EffectComposer.

Community
Intermediate
agentspace-soagentspace-so
31

flux-kontext

Edit single images with Flux 1 Kontext Pro via RunComfy CLI.

Official
Intermediate