Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > comfyui-lucid-vrp

ComfyUI Extension: comfyui-lucid-vrp

Repo Name

comfyui-lucid-vrp

Author
EBEBEEBE (Account age: 68 days)
Nodes
View all nodes(4)
Latest Updated
2026-11-05
Github Stars
0.04K

How to Install comfyui-lucid-vrp

Install this extension via the ComfyUI Manager by searching for comfyui-lucid-vrp
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter comfyui-lucid-vrp in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

comfyui-lucid-vrp Description

comfyui-lucid-vrp combines image loading, masks, layered canvas editing, and visual regional prompting in ComfyUI, with tools for Anima regional prompts and MOSS-Image layout guidance.

comfyui-lucid-vrp Introduction

LuCiD VRP — Visual Regional Painter

comfyui-lucid-vrp, also called LuCiD VRP (Visual Regional Painter), brings image preparation, layered editing, and visual regional prompting into ComfyUI.

Instead of describing everything in one long prompt and hoping the model places each element correctly, you can prepare a composition visually: arrange image layers, paint a rough sketch, draw regions, and give different areas their own descriptions.

The extension combines three main tools:

  • VRP Image Loaders: Load an existing image or create a blank canvas, edit masks, resize your composition, and optionally prepare the starting image for generation.
  • LuCiD Canvas: A lightweight built-in image editor with layers, painting, transforms, distortion, and reversible cropping.
  • Visual prompting nodes: Guide Anima with regional prompts or create MOSS-Image layout prompts by drawing boxes directly over your canvas. It is especially useful for image-to-image workflows, where a sketch, collage, mockup, or previous generation helps guide the next image.

What can you use it for?

You can use LuCiD VRP to:

  • Arrange characters and objects before generating a finished illustration.
  • Give different subjects separate descriptions, such as a red-coated character on the left and a blue-coated character on the right.
  • Build a rough composition from several image layers without switching to an external editor.
  • Reposition, reshape, paint, and crop elements while keeping edits adjustable.
  • Explore different compositions using Undo and Redo.
  • Create MOSS-Image layout prompts without manually writing JSON or calculating box coordinates. You do not need Anima or MOSS-Image to use the image loaders and Canvas editor. Their image and mask outputs can be used with other image-model workflows. Built-in latent encoding requires a compatible VAE.

Important: Regional prompts guide generation; they do not lock pixels or guarantee exact placement. Some attributes can still spread between regions.

See the workflow in action

[Demo showing layout creation, editing, and regional prompting](docs/images/demos/regional-sync.webp)

A demonstrated creative sequence is:

  1. Build a box-guided mockup with MOSS-Image.
  2. Bring that image into an Anima image-to-image workflow.
  3. Refine character placement and composition in LuCiD Canvas.
  4. Generate or regenerate the finished image.

How comfyui-lucid-vrp Works

Think of LuCiD VRP as a combination of a digital sketchbook and a composition map for your generation workflow.

1. Prepare the starting image

You begin with either an imported image or a blank canvas. In LuCiD Canvas, you can combine layers, paint adjustments, and move or reshape elements.

For example, you could place two character cutouts over a background, resize them to establish perspective, and paint a simple shadow underneath them. This mockup gives an image-to-image workflow a visual starting point.

2. Mark where different ideas belong

A region is an area associated with its own prompt. You can draw rectangular boxes or, for Anima workflows, use masks.

For a scene with two characters, your descriptions might be:

  • Overall scene: “Two travelers in a snowy forest, soft evening light.”
  • Left region: “A traveler wearing a red winter coat.”
  • Right region: “A traveler wearing a blue cloak.” The overall description establishes the shared scene. Regional descriptions help associate individual details with particular areas.

3. Choose the guidance method

LuCiD VRP supports two different approaches:

| Approach | How it works | What it means for you | |---|---|---| | Anima regional prompting | Adapts Anima’s attention processing so different image areas are guided by different text descriptions. | You can use boxes or masks, regional positive and negative prompts, influence controls, and prompt mixing. | | MOSS-Image layout prompting | Converts your visual layout into MOSS-Image’s native bounding-box prompt format. | You can describe and position elements visually rather than writing layout JSON yourself. |

In simple terms, Anima regional prompting helps direct which description an area listens to, while MOSS-Image layout prompting gives the model a structured plan of the scene.

These are not interchangeable systems: their supported controls and limitations differ.

4. Send the result into your generation workflow

The loaders can provide:

  • image: The prepared picture.
  • mask: A mask for use by compatible workflow nodes.
  • latent: A generation-ready starting representation when enabled.
  • canvas: Canvas information used by the regional-layout tools on the Canvas loader. A latent is the compressed representation that a diffusion model works with during generation. A VAE converts between ordinary images and that representation.

Your canvas edits and authored layouts update visually without running the workflow. Generating a new AI image still requires executing the generation workflow.

comfyui-lucid-vrp Features

Image loaders for image-to-image preparation

Choose the loader that matches your task:

| Node | When to use it | |---|---| | VRP Standalone Image Loader | You need an image and mask source for an existing workflow, without a regional-prompting canvas connection. | | VRP Canvas Image Loader | You also need canvas information for Anima regional prompting or MOSS-Image layout prompting. |

Both support starting from an image or a blank canvas and preparing your composition inside ComfyUI.

Resize your composition and mask together

You can select a resolution preset or enter custom dimensions, with an aspect-ratio preview to help you judge the shape of the workspace.

For example:

  • A wide canvas gives you more horizontal space for several subjects.
  • A portrait canvas suits a tall character composition.
  • Resizing the image and mask together helps keep their positions aligned. Working dimensions align to multiples of 32 by default.

[Resolution selection and resizing demonstration](docs/images/demos/loader-resize.webp)

Undo and Redo

Editing history lets you step backward and forward through saved image, paint, and mask states.

This is useful when you want to compare two arrangements or recover an earlier stage after making an unwanted change.

[Image-loader Undo and Redo demonstration](docs/images/demos/loader-undo-redo.webp)

Built-in latent preparation

For compatible workflows, the loaders can replace a separate VAE Encode step.

| Mode | Purpose | |---|---| | Off | Use the image and mask outputs without producing a starting latent. | | Empty | Start generation from an empty latent rather than encoding the canvas image. | | Encode | Encode the loaded or edited image using the connected VAE. | | Encode (Tiled) | Use tiled VAE encoding. | | Auto | Encode imported images and RGB-painted blanks; use Empty for an unpainted white blank. |

Choose the mode based on whether the picture should actually guide generation. A visible reference image is not the same as an encoded image-to-image starting point.

[Built-in latent preparation demonstration](docs/images/demos/loader-latent.webp)

LuCiD Canvas: layered editing inside ComfyUI

Open Edit canvas on a loader to work in LuCiD Canvas. Press Apply when you want to return the edited composition to the loader.

Layers and independent image objects

Import additional images and arrange them as separate elements. Keeping objects separate makes it easier to adjust one part of a composition without rebuilding everything else.

Attached paint layers help organize painted changes alongside the relevant image objects.

[Canvas layers and composition demonstration](docs/images/demos/canvas-layers.webp)

Painting, masking, and erasing

Use the editor to paint over your mockup, adjust masks, or erase unwanted areas. Brush controls and color picking help you make quick visual adjustments.

For example, you can paint a rough color block to suggest clothing color or modify a silhouette before sending the composition into generation.

Transforms, distortion, and reversible cropping

You can transform and distort elements to improve their placement and shape. Reversible cropping lets you change the visible portion without permanently discarding the original content.

These tools are useful when a character needs to be larger, an object needs a different shape, or a layer needs tighter framing.

[Canvas distortion demonstration](docs/images/demos/canvas-distort.webp)

Customizable editing controls

You can customize shortcuts and mouse interactions to suit your preferred editing habits.

[Canvas color picker and editing controls demonstration](docs/images/demos/canvas-picker.webp)

See the LuCiD Canvas manual and Canvas quick reference for detailed controls.

Visual region authoring

Draw, move, and resize boxes directly over the canvas. Layout previews update as you edit.

You can organize regions in two ways:

  • Internal regions: Keep boxes and prompts inside the main regional-prompting node for a compact workflow.
  • External regions: Use separate VRP Region nodes to keep individual regions organized and reusable within the graph. You can also combine external regions with boxes drawn inside the Anima core node.

[Visual regional prompting demonstration](docs/images/demos/regions.webp)

Anima-specific prompting controls

Anima regional prompting supports:

  • Rectangular boxes.
  • Masks painted with ComfyUI’s native Mask Editor.
  • Externally connected MASK inputs.
  • Regional positive prompts.
  • Regional negative prompts.
  • Adjustable regional influence.
  • Prompt mixing. Regional influence adjusts how strongly a region contributes to guidance; it is not a guarantee of precise placement. Prompt mixing provides additional control over how descriptions are combined. See Regions and prompts for the available behavior and options.

Advanced regional-routing settings offer further control and performance tradeoffs. See Advanced settings before changing these controls extensively.

MOSS-Image layout authoring and export

The **MOSS

  • Visual Regional Prompt** node turns your boxes and descriptions into native MOSS-Image layout prompts. You can:

  • Describe the overall scene with a global prompt.

  • Give each region its own positive description.

  • Create layouts without manually calculating coordinates.

  • Use built-in text encoding for sampler connections.

  • Export native layout JSON for separate encoding. MOSS-Image’s native format describes the overall scene and individual elements with bounding boxes. LuCiD VRP handles the visual layout authoring so you do not have to write that structure by hand.

MOSS-specific restrictions:

  • Incoming masks become enclosing rectangles.
  • Regional negative prompts are not supported.
  • Negative prompting applies globally.
  • Fractional region weights are not supported.

Flexible prompt connections

Built-in encoding makes it easier to connect regional nodes to a sampler.

For more customized workflows, VRP Prompt Combine collects prompts from external Region nodes for your own encoding setup or a refiner branch. MOSS layout JSON can also be encoded separately.

See the Node reference for exact inputs and outputs.

comfyui-lucid-vrp Models

LuCiD VRP is a workflow extension, not a standalone image-generation model. Its editing tools and its model-specific prompting tools have different compatibility requirements.

Supported model families and uses

| Model or workflow | Supported use | Requirements and guidance | |---|---|---| | Other image models | Image loading, Canvas editing, and image/mask preparation. | Image and mask outputs are model-agnostic. Latent compatibility depends on your workflow and VAE. | | Anima base model | Anima regional prompting and compatible image-to-image workflows. | Use compatible Anima text encoders and VAE. | | Anima base finetunes | Regional prompting with an Anima finetune. | Included in the author’s tested setups. | | Anima 2.9B finetunes | Regional prompting with these Anima variants. | Included in the author’s tested setups; use matching model components. | | MOSS-Image | Experimental native bounding-box layout prompting. | Requires native ComfyUI MOSS-Image support, MOSS weights, a compatible Qwen 3.5 text encoder, and a Flux.2 VAE. |

MOSS-Image-Base and MOSS-Image-SFT

The MOSS model family includes:

  • MOSS-Image-Base: The base text-to-image model.
  • MOSS-Image-SFT: A version fine-tuned from Base on curated data. The MOSS project shows an example where SFT produces clearer text grouping and a more organized whiteboard layout. This is an example of the difference between the models, not a guarantee for every prompt.

LuCiD VRP’s supplied MOSS layout workflow uses SFT, making it the documented example to follow.

ComfyUI-format MOSS model resources are available separately. Unofficial MOSS workflows and custom nodes have not been tested with this extension.

You need no model or VAE for image editing alone. A compatible VAE is needed when encoding an image into a latent.

comfyui-lucid-vrp Quick Start and Example Workflows

Find the extension’s nodes under LuCiD VRP in ComfyUI’s Add Node menu.

Prepare an image for an existing workflow

  1. Add VRP Standalone Image Loader, or use VRP Canvas Image Loader if you also need regional layout connections.
  2. Press Import, drag an image onto the loader, or choose Create blank.
  3. Open Edit canvas.
  4. Import layers, arrange objects, paint, and refine the composition.
  5. Press Apply.
  6. Connect the loader’s image and mask outputs where needed. For image/mask-only use, set latent mode to Off and leave latent disconnected.

Use the prepared picture for image-to-image generation

  1. Connect a VAE compatible with your generation model.
  2. Choose Auto, Encode, or Encode (Tiled).
  3. Connect the loader’s latent output to the sampler’s latent_image input.
  4. Set sampler denoise below 1.0 when you want the starting image to influence the result. Use denoise 1.0 for an empty latent or when the imported image is only a reference rather than an image-to-image source.

Create a MOSS-Image SFT layout

  1. Begin with a working native MOSS-Image SFT generation workflow.
  2. Add VRP Canvas Image Loader and MOSS - Visual Regional Prompt.
  3. Create a blank canvas in Empty mode.
  4. Draw boxes inside the MOSS node, or use external VRP Region nodes connected through the canvas and regions inputs.
  5. Enter an overall scene prompt and a positive prompt for each region.
  6. Connect a MOSS-compatible CLIP/text encoder to the MOSS node.
  7. Send its positive and negative outputs to your sampler.
  8. Use the Canvas loader’s latent output as the starting latent. [MOSS-Image SFT layout workflow](docs/images/workflows/moss-sft-layout.webp)

The image contains an embedded workflow. Drag it into ComfyUI, then select your model files.

For details, see the MOSS layout guide.

Create a minimal Anima regional workflow

  1. Begin with your usual working Anima model loaders and sampler.
  2. Add VRP Canvas Image Loader and the Anima VRP core node.
  3. Create a blank canvas in Empty mode.
  4. Connect the loader’s canvas and latent outputs to Core’s canvas and latent_image.
  5. Draw boxes inside Core.
  6. Select each region to edit its name, weight, and positive/negative prompts.
  7. Connect the required Anima model and compatible CLIP/text encoder.
  8. Send Core’s patched model, conditioning, and latent outputs to your sampler. [Minimal Anima regional workflow](docs/images/workflows/anima-minimal.webp)

Combine Anima regional prompting with image-to-image

  1. Import a sketch, mockup, or existing image into VRP Canvas Image Loader.
  2. Refine it with Edit canvas, then press Apply.
  3. Connect a compatible VAE.
  4. Choose Auto or Encode.
  5. Connect the loader’s canvas and latent outputs to Anima Core.
  6. Define your regions and prompts.
  7. Connect an Anima-compatible CLIP/text encoder.
  8. Use Core’s outputs in your sampler and set denoise below 1.0. [Anima image-to-image regional workflow](docs/images/workflows/anima-image-to-image.webp)

You can also use the example input image.

Organize regions in separate nodes

For an external region chain:

  1. Create one VRP Region node per region.
  2. Define its box or, for Anima, provide a painted or external mask.
  3. Carry both canvas and regions through the chain.
  4. Connect each preceding regions output to the next node’s previous_regions input.
  5. Connect the final canvas and regions outputs to the main regional node. [Anima external-region workflow](docs/images/workflows/anima-external-regions.webp)

The Anima workflow images also contain embedded workflows that you can drag into ComfyUI.

comfyui-lucid-vrp Compatibility and Limitations

ComfyUI interface support

Classic ComfyUI nodes are required. Nodes 2.0 is currently unsupported.

The author’s current test setup is:

  • ComfyUI 0.42.0

  • Frontend 1.57.1 The image loaders, Canvas editor, and Anima regional prompting were also tested with:

  • ComfyUI 0.32.0 and frontend 1.48.7

  • ComfyUI 0.37.0 and frontend 1.53.6 Those older setups do not include native MOSS-Image support.

Creative boundaries

  • Regions are approximate guides. They do not guarantee exact positions, prevent all attribute leakage, or preserve particular pixels.
  • MOSS-Image integration is experimental. Its layout controls are more limited than Anima’s regional controls.
  • Masked inpainting is not included. Mask outputs and regional masks do not by themselves create an inpainting workflow.
  • Regional LoRA and built-in ControlNet integration are not included. Use separate workflows or nodes where appropriate.
  • Anima attention patches may conflict with other extensions that modify the same attention operations.
  • Saved workflows are not complete image-editing documents. Keep referenced images, masks, and Canvas layer assets with your workflow. Features demonstrated by related projects—such as FLUX regional prompting, identity preservation, or regional LoRA—should not be assumed to be features of LuCiD VRP.

See Compatibility and limitations for more information.

Troubleshooting comfyui-lucid-vrp

Nodes or editing controls do not appear correctly

  1. Confirm that ComfyUI is using classic nodes, not Nodes 2.0.
  2. Restart ComfyUI and refresh your browser.
  3. Look under LuCiD VRP in the Add Node menu.
  4. Compare your setup with the documented tested versions.
  5. Consult the Troubleshooting guide and ComfyUI’s custom-node troubleshooting documentation.

Canvas changes do not reach the loader

  1. Reopen Edit canvas.
  2. Check that your intended edits are visible.
  3. Press Apply to return the composition to the loader.
  4. If using image-to-image generation, confirm that the loader is encoding the edited image rather than using Empty mode.

The generated image ignores my starting picture

  1. Check that latent mode is not Empty or Off.
  2. Connect the correct VAE.
  3. Choose Encode explicitly if you are unsure what Auto selected.
  4. Confirm that the encoded latent reaches the sampler’s latent_image input.
  5. Set denoise below 1.0 so the starting image can influence generation. An unpainted white blank uses Empty when latent mode is Auto.

Latent encoding fails or the sampler rejects the latent

  1. Confirm that the VAE matches the generation model.
  2. Check that the selected latent mode matches your intended workflow.
  3. Review image dimensions and alignment in Using Image Loaders.
  4. If using a special pipeline or custom sampler, use that pipeline’s required latent nodes when necessary. The loader’s standard Empty placeholder is adapted by compatible native samplers, but not every custom pipeline supports it.

External regions are missing

  1. Check that each Region node has a defined box or mask.
  2. Verify that regions connects to the next node’s previous_regions.
  3. Carry canvas forward as well as regions.
  4. Connect both final outputs to the main regional node.
  5. Compare your graph with the external-region example.

Subjects or colors appear in the wrong regions

  1. Check the boxes or masks against your intended composition.
  2. Review each regional prompt for conflicting descriptions.
  3. For Anima, review regional influence and prompt-mixing settings.
  4. If placement needs stronger visual guidance, build a mockup in Canvas and use image-to-image generation.
  5. Use a separate finishing or inpainting workflow for precise corrections. Some leakage remains possible even with a correctly connected workflow.

MOSS layout generation does not work

  1. Confirm that your ComfyUI version has native MOSS-Image support.
  2. Select compatible MOSS weights, a Qwen 3.5 text encoder, and a Flux.2 VAE.
  3. Start from the supplied SFT example.
  4. Check that layout dimensions match the intended generation size.
  5. Remove assumptions about Anima-only controls: MOSS does not support regional negatives or fractional region weights.
  6. See the MOSS layout guide for conversion limits.

Anima generation fails when combined with another extension

  1. Test the supplied minimal Anima workflow.
  2. Temporarily remove other nodes that modify model attention.
  3. Reintroduce those nodes one at a time to identify a conflict.
  4. Review Compatibility and limitations. Extensions that patch the same attention operations may not work together.

A saved workflow reopens with missing layers or history

  1. Check that the original images, masks, and layer assets are still available.
  2. Restore missing referenced assets.
  3. Keep those files alongside the workflow when archiving or moving a project. Saving the workflow alone does not package every image-editing asset.

Frequently asked questions

Can I use LuCiD Canvas without generating an image? Yes. Image editing without latent encoding requires no model or VAE.

Can I use it with models other than Anima and MOSS-Image? Yes, for image loading, Canvas editing, and image/mask preparation. The model-specific regional-prompting systems are limited to their supported families.

Do I need to run the workflow to move a box or edit the canvas? No. Canvas edits and authored layout previews update live. A new generated result still requires running the workflow.

Will a region preserve the exact pixels inside its boundary? No. Regional guidance is not pixel locking or masked inpainting.

Can I draw a detailed mask for MOSS-Image? Incoming masks are converted to enclosing rectangles, so their detailed outlines are not preserved as native layout regions.

Can I use MOSS first and Anima afterward? Yes. The documented demonstration uses a MOSS mockup as an image-to-image source for Anima, with Canvas editing between stages.

Learn More about comfyui-lucid-vrp

Practical guides

| Resource | How it helps | |---|---| | Node reference | Choose the right node and understand its connections and outputs. | | Using Image Loaders | Learn importing, resizing, history, latent modes, and alignment. | | LuCiD Canvas manual | Work through layers, painting, transforms, distortion, cropping, and masks. | | Canvas quick reference | Look up editor controls, shortcuts, preferences, and saved state. | | Regions and prompts | Understand internal regions, external chains, masks, and Prompt Mix. | | MOSS layout prompting | Learn scene prompts, box layouts, encoding, and conversion limits. | | Advanced settings | Explore regional-routing controls and performance tradeoffs. | | Configuration | Customize resolution presets, colors, layout, and starting regions. | | Troubleshooting | Check common problems and errors. | | Compatibility and limitations | Review supported setups and current boundaries. |

Model and background resources

comfyui-lucid-vrp License and Credits

The author’s original software and documentation, including LuCiD Canvas and the MOSS layout node, are licensed under the MIT License.

LuCiD VRP adapts Anima regional-conditioning logic from ComfyUI-Anima-Regional-Conditioning and uses MOSS-Image’s documented native layout format. It relies on ComfyUI and its frontend for host functionality.

Adapted third-party code retains its applicable copyright and license notices. See Third-party notices for details and development references.

Models, text encoders, and separately installed projects retain their own terms.

comfyui-lucid-vrp Related Nodes

RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

comfyui-lucid-vrp detailed guide | ComfyUI