ComfyUI-RedNodeStudio Introduction
ComfyUI-RedNodeStudio brings image creation and finishing into one workspace in ComfyUI. Instead of managing a large graph of connected nodes for every step, you can choose a model, write a prompt, generate an image, refine it, and save it from a single panel.
The extension is designed for more than starting a picture from scratch. You can work from an existing image, paint an area you want to change, build a new scene around a person from a photo, add detail, upscale, and apply a finished look. Its workspace shows the stages of a run in order, so you can see what is enabled and what needs attention without tracing wires across a canvas.
The Run page showing a render, pipeline stages, and VRAM use
Watch the full RedNode Studio tutorial for a visual walkthrough of the models and workspace tabs.
How ComfyUI-RedNodeStudio Works
Think of the Workspace as a compact image studio. You give it a creative direction, choose the equipment for the job, and decide which finishing steps to use. The workspace organizes those steps behind one interface while ComfyUI performs the image processing.
A typical new-image session looks like this:
- Choose a model setup. The Models Setup page looks for the model and supporting files in your existing folders. Select a model family to build a ready-to-use setup, or load another model manually.
- Describe your image. Write a prompt directly, or use separate boxes for elements such as style, subject, surroundings, and lighting.
- Choose the stages you need. The workspace shows the run in order. Enabled stages appear green; disabled stages appear grey. If something needs fixing, the workspace labels it, and you can select that stage to inspect it.
- Generate and review. The Run page shows the image as it develops, the time spent at each stage, and a chart of graphics-card memory use.
- Refine and save. You can edit or paint over the result, apply finishing effects, and organize the files you want to keep.
You do not have to use every stage for every picture. For example, a quick concept image may only need a model and prompt, while a portrait might also benefit from identity guidance, detailing, and a finishing look.
The workspace overview and prompt controls
ComfyUI-RedNodeStudio Features
Prompts and Camera Direction
The Prompts tab lets you build a description in parts, including style, subject, surroundings, and light. It also offers wildcards and saved presets for reusing or varying ideas. A Frame box and camera controls help you describe how the scene should be composed.
In the Camera tab, you can position the subject and camera on a stage, then choose a lens and camera height. The workspace uses those choices to help form the prompt and control applicable camera LoRAs. For example, changing the camera height gives you a way to direct the viewpoint without rewriting the whole scene description.
Prompt boxes for building an image description
Editing and Painting
The Tools and Re-render areas let you work on an image you just generated or one from the gallery. The original README describes uses such as turning an illustration into a photograph, changing the viewing angle, swapping in a face, and upscaling.
For a local change, use the Paint tab: mark the region you want to alter, describe what should appear there, and generate. The painted region changes while the rest of the image serves as its context. The result becomes the next image you can paint on, which makes it possible to refine an area in stages.
Re-render controls for working from an existing image
The Paint tab for changing a selected region
Identity and Subject Images
Hero Creator turns a source photo into a clean, front-facing headshot by cropping the head above the clothing and rendering a new portrait. You can place that headshot in the Subject gallery and use Krea 2 Identity to guide a new scene featuring the same person. This is useful when you want to change the setting or outfit while keeping the subject recognizable.
Hero Creator source, crop, and headshot
An identity-guided image in a different scene
Detail, Upscaling, and Finishing
The Detailer presents its passes as a list instead of requiring you to wire each one into a graph. The workspace also includes image-to-image work with multiple passes, a searchable LoRA stack with groups, and moodboards for visual direction.
Post FX provides 22 effects arranged in camera order, covering adjustments from exposure and skin to film stocks and lens effects. You can begin with one of 18 ready-made looks, then adjust the finishing choices to suit the picture rather than treating the preset as the final result.
Post FX controls and ready-made looks
Run Monitoring and Memory
The Run page shows a live picture, stage-by-stage timings, and a VRAM chart. VRAM is the memory on your graphics card; watching its use can help you understand why a demanding workflow is difficult to run. You can set a VRAM limit, and the workspace estimates what the run needs and manages it against that limit.
The workspace also offers live previews, picture history with rerun, and comparisons between stages, so you can inspect how a result changed during processing.
Saving and Workspace Layout
The Save page can organize pictures using details such as date, preset, seed, model, or image size. You can separate drafts from keepers and review a session within ComfyUI. Saved files carry their prompt, settings, and model and LoRA hashes, helping you identify what produced an image later.
You can place the tab rail on either side, collapse it to icons, or hide tabs you do not use. Workspace presets include Basic, New image, Image to image, Edit, Paint, and Finishing. The Generate button remains at the top as you move between tabs. A local assistant can explain settings, and control panels can drive other nodes without visible wires.
The Save page for organizing drafts and keepers
The tab rail positioned on the right
ComfyUI-RedNodeStudio Models
Models are the image-making engines you choose for a run. RedNode Studio can prepare setups for these model families:
| Model or family | How it fits into the workspace |
|---|---|
| Krea 2, including Krea 2 Turbo | Available for image creation; Krea 2 Identity is also used with a Hero Creator headshot to guide the same person's appearance in a new scene. |
| Qwen Image 2.1 | Available as a one-click model setup for image work in the workspace. |
| Z-Image, including Z-Image Turbo | Available as a one-click model setup. |
| Anima | Available as a one-click model setup and shown in the workspace examples. |
| SDXL, Illustrious, and Pony | Supported as model-family setups within the same workspace. |
| Other models | Can be loaded manually rather than through one of the listed one-click setups. |
For supported families, Models Setup finds the model, text encoder, VAE, and relevant settings by name in your own folders. You can keep multiple setups and switch between them. The supplied context does not compare their artistic styles, speed, or hardware requirements, so the best choice for a particular picture is something to test with your own prompts and available files.
Models Setup showing files found for each model family
Troubleshooting ComfyUI-RedNodeStudio
A model setup says something needs attention. What should you check?
- Open the Workspace overview and select the stage marked as needing attention.
- Open Models Setup and check which files it found for the model family you selected.
- Confirm that the model and any supporting files needed by that setup are present in your ComfyUI folders.
- Consult the full guide for the files and settings associated with that model.
A run uses more graphics memory than you expected. What can you do?
- Open the Run page and watch the VRAM chart while the workflow runs.
- Review the active stages and any additional passes you have enabled.
- Set a VRAM limit in the Workspace so it can manage the run against the memory available on your card.
Where can you see which part of a run is taking time?
The Run page displays each stage and its timing. You can also use its live picture and stage comparisons to see what changed as the image progressed.
Can you edit a picture without starting over?
Yes. You can use a gallery image or a recent render in the editing tools. To change only one area, mask it in Paint, describe the replacement, and generate.
Can you use a model that is not listed in Models Setup?
Yes. The listed families have one-click setups; other models can be loaded manually.
Learn More about ComfyUI-RedNodeStudio
- Full guide: Details on the workspace tabs, nodes, settings, model files, and performance.
- Video tutorial: A visual tour of the models and the steps for making, fixing, and saving a picture.
- Changelog: Version-by-version changes.
- ComfyUI: Background on the application that runs the workspace.
