Upscale Workflow

Paid

ComfyUI workflow for high-fidelity image upscaling

4.7

New AI ToolsContact

Inputs: image, textOutputs: image

Type

Saas

Company

Runtime44

Visit Website

About Upscale Workflow

An open-source ComfyUI workflow for upscaling images using a double sampling process. Developed by Runtime44, this workflow adapts the principles of their Mage platform to the ComfyUI environment. It separates subjects from backgrounds using masks, applies a first sampling pass (often with SD 1.5) to generate detail and noise, then a second pass to refine and fix artifacts. Supports multiple diffusion models (SD1.5, SDXL), upscale model chaining (e.g., 2x, 3x, 4x), and optional ControlNet Tile for consistency. Requires several custom node packages (Impact Pack, rgthree, Comfyroll, etc.) and specific upscale models. Results may differ from the original Mage service due to differences in pipeline and parallel processing.

Key Features

Double sampling process (exaggerated first pass, refining second pass)

Mask-based separation of subjects from background

Supports SD1.5 and SDXL diffusion models

Upscale model chaining (2x, 3x, 4x) to achieve higher upscale factors without dedicated large models

Optional ControlNet Tile for consistency

Partial adaptation of Runtime44's Mage pipeline to ComfyUI

Pros & Cons

Pros

Produces high-quality upscaled images with good faithfulness to original
Flexible upscale chain allows using multiple small models instead of one large model
Supports both SD1.5 and SDXL, giving users model choice
Open-source and free to use (requires ComfyUI and dependencies)
Includes mask-based separation for targeted processing

Cons

Requires installation of numerous custom nodes and models (heavy setup)
Not a standalone tool – requires ComfyUI to run
Results may differ from the original Mage service due to different pipeline
GPU-intensive; requires sufficient VRAM for double sampling and upscale models
Limited documentation beyond basic README (no detailed troubleshooting guide)

Best For

Upscaling images while preserving original detail and faithfulnessRefining noisy or low-quality images through sequential diffusion passesGenerating high-resolution images from lower-resolution inputsIsolating and enhancing subjects separately from background during upscaling

Alternatives to Upscale Workflow

PlugSugar

Automate conversations, answer questions with Web Search plugin, and customize ChatGPT experience using powerful AI plugins.

MedARC

Monitor patient progress, streamline processes and generate insights for healthcare providers to improve patient care.

SumUp

Experience Unmatched Sound Quality with AmbienceSoundPro!

Thinking Toolbox

Assess thinking skills, practice critical thinking, and access curated resources to improve cognitive abilities.

Respage

Automate lead acquisition, interact with potential leads, and capture lead information and preferences.

Travel Plan AI

Your personal AI guide for unforgettable journeys.

FAQ

How does this workflow differ from Runtime44's Mage?

Both aim for high-quality upscaling faithful to the original, but the ComfyUI workflow is a partial adaptation. Mage is optimized for parallel processing on multiple GPUs with a different diffusion pipeline, so results may differ.

What dependencies are needed to run this workflow?

You need ComfyUI and custom nodes: runtime44_comfyui_nodes, ComfyUI-Impact-Pack, rgthree-comfy, ComfyUI_Comfyroll_CustomNodes, ComfyUI-Custom-Scripts, comfyui_controlnet_aux, and ComfyUI-KJNodes. Install via ComfyUI Manager or manually.

Which upscale models are recommended?

The workflow can use any upscale model that fits the scaling factor. The developers recommend a combination like 2x RealESRGAN_plus and 4x foolhardy_Remacri, available on OpenModelDB.

Can I use this workflow for 8x upscaling?

Yes, the upscale chain distributes the load across multiple nodes (up to four). You can use a 2x model and chain it instead of needing an 8x model.