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-vrm-pose-editor

ComfyUI Extension: comfyui-vrm-pose-editor

Repo Name

comfyui-vrm-pose-editor

Author
ketle-man (Account age: 1078 days)
Nodes
View all nodes(1)
Latest Updated
2026-11-03
Github Stars
0.02K

How to Install comfyui-vrm-pose-editor

Install this extension via the ComfyUI Manager by searching for comfyui-vrm-pose-editor
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter comfyui-vrm-pose-editor 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-vrm-pose-editor Description

comfyui-vrm-pose-editor adds an interactive 3D character posing node to ComfyUI. Adjust poses, cameras, lighting, and backgrounds, animate with keyframes, and capture PNGs for image workflows.

comfyui-vrm-pose-editor Introduction

comfyui-vrm-pose-editor adds an interactive 3D Pose Editor node to ComfyUI. You can load a 3D character into your browser, adjust its pose, camera, lighting, and background, then capture the result as a PNG for your ComfyUI workflow.

It is useful when you want more control over a character reference than a written prompt or a fixed pose image can provide. You can turn a limb, change the viewing angle, adjust the light, and see the result immediately. For moving scenes, you can arrange keyframes on a timeline and render an animation.

The editor supports local VRM, GLB, and GLTF character files. Some tools—such as VRM animation playback, automatic blinking, and image-to-pose transfer—specifically require a compatible VRM character.

3D Pose Editor in a ComfyUI workflow

How comfyui-vrm-pose-editor Works

Think of the editor as a small virtual photo studio:

  1. Place a character in the scene. Load a VRM, GLB, or GLTF file. The editor scales and centers it automatically.
  2. Pose the character. Drag the blue points at its joints to rotate bones. On compatible VRM characters, you can also adjust expressions using Shape Keys—sliders for features such as mouth or eye shapes.
  3. Set up the shot. Move the camera, arrange lights, and choose a background or environment.
  4. Capture or animate. Use 📸 Capture to send a PNG of the current view to the node output. To make a sequence, record poses and other scene settings as keyframes: snapshots of what should happen at particular frames.

A keyframe timeline works like a storyboard with precise timing. For example, you could save one pose at frame 0 and another at frame 30. The editor previews the transition between them. Separate tracks let you change the camera, lights, gaze, or wind without having to change the character’s pose at the same time.

You can also start from existing motion instead of posing everything by hand. A .vrma animation can be played as a clip or brought into the timeline as editable pose keyframes. With ComfyUI’s native SAM 3D Body nodes and model available, the Image tab can estimate a person’s pose from a picture, or turn movement in a video into pose keyframes.

comfyui-vrm-pose-editor Features

Pose and model controls

Click VRM on the node, or Load MODEL in the editor, to choose a local .vrm, .glb, or .gltf file.

  • VRM characters use humanoid bones recognized by @pixiv/three-vrm. VRM-specific features depend on the expressions, LookAt data, and spring bones present in your particular character.
  • GLB and GLTF files can be posed when they contain a skeleton, but features that rely on a VRM humanoid rig may not work.

Drag a blue joint control point to rotate a bone. Left/right dragging rotates it on one axis, up/down on another, and Alt + up/down on a third. Use RP to reset the pose, or ↔ to mirror it left-to-right. The Point Size slider changes the size of the joint handles if they are difficult to see or select.

Under Pose → K (Shape Keys), compatible model expressions appear as sliders from 0 to 1. They update the preview immediately and are saved with pose keyframes.

Pose controls and Shape Keys

Camera and scene composition

Use the mouse to frame your character:

ActionResult
Left dragRotate the view
Right drag or Ctrl + left dragPan the view
Scroll wheelZoom
Alt + right dragRoll the camera
Click an X/Y/Z gizmo axisSnap to that viewing direction

If scroll-wheel zoom does not work on your computer, enable Ctrl+Right drag zoom under Light → S (Settings) and use that gesture instead. RC resets the camera, while OT / PR switches between orthographic and perspective views.

The FOV slider changes how wide a perspective camera sees. Near controls how close an object can get to the camera before it is clipped; setting it too high can make nearby parts of the model disappear.

Under Pose → C (Camera), you can add, select, rename, recolor, and delete cameras. Each camera remembers its own viewpoint. Inactive cameras appear as draggable icons in the scene, making it easier to position another shot without abandoning your current view. 🖥 Monitor provides a free scouting view that does not follow camera keyframes; it is for arranging the scene, not for recording a camera animation.

Camera controls

Lighting, environment, and appearance

Click Light to open the combined Light & Pose Editor on its lighting tab. Its L sub-tab lets you add and adjust lights, including sun/directional, point, spot, rectangular area, and ambient lights. Depending on the light type, you can change properties such as color, intensity, position, target, spread, distance, and shadows. You can drag a light’s yellow helper in the preview to reposition it.

The E (Environment) sub-tab contains ground, background wall, shadow-quality, and wind controls. On the node, BG loads a background image, 🎨 selects a background color, and ✕ clears both. CC toggles color correction using sRGB and ACES Filmic. The S (Settings) sub-tab also includes an anti-aliasing enhancement option.

The Light and Pose tabs share the actual 3D preview and a timeline, so you can adjust the scene while watching the result in the same window.

Lighting controls

Environment settings

Gaze, spring bones, wind, and blinking

These controls add character and movement when the loaded VRM supports them:

  • 👁 LookAt: Makes the character’s eyes or head follow a target. Drag the cyan target marker, or switch the target to the active camera. The marker does not appear in captured images. A model without LookAt data will not respond.
  • 🎐 Spring bone physics: Enables movement defined for features such as hair or skirts. Switching it off freezes the current sway.
  • 🌬 Wind: Adds a drifting breeze to spring bones. Adjust its strength, direction, and gustiness under Light → E. Wind has no effect while spring bone physics is off.
  • 🧭 Wind source marker: Drag the orange cone to aim the wind in the 3D view, including upward or downward. While it is active, the wind-direction slider is disabled. The cone is not captured.
  • 😑 Auto Blink: Available at the top of the Shape Keys list when a VRM has a blink expression. Turn it on and set the average time between blinks.

Poses, animations, and the library

Use ⬇️ to download the current pose as JSON, ⬇️VP to download a VRoid Studio .vroidpose file for a VRM, 💾 to save a pose in poses/, or 📂 to load a .json or .vroidpose pose.

The 📚 Pose Library displays saved poses and .vrma animations as thumbnails. You can organize files in subfolders, filter by folder, mark favorites, and preview an animation before deciding whether to load it as a clip or as timeline keyframes. Thumbnails use the current or default VRM when one is available.

Pose Library

For animation, open the keyframe timeline at the bottom of the Light & Pose Editor. Choose a track, move to a frame, and press ✚ Add/Update to record its current settings. Tracks include:

TrackWhat you can animate
PoseBone rotations and Shape Keys
CameraEach camera’s movement and viewpoint
Cam SwitchCuts between cameras
LightLighting setups
WindBreeze settings and marker position
EyesLookAt state, target choice, and marker position
BlinkAuto Blink state and interval
LipTimed mouth movement from text or selected audio

You can move or delete keyframes, preview playback, save the timeline as a project, and use ⏱ Stretch to make a selected range faster or slower. A .vrma file can be played directly with VRMA, or sampled into Pose-track keyframes with VRMA (KEY). The keyframe importer also accepts supported .bvh motion-capture files.

Export distinction: The editor’s .vrma export writes bone animation, but not the preview’s camera, lighting, wind, gaze, blinking, lip movement, or Shape Key animation. To preserve those visible effects in a shareable result, render WebM, MP4, or GIF from Pose → K → Output. MP4 conversion requires ffmpeg.

Keyframe timeline

Image-to-pose and video-to-motion

Under the Image tab, you can estimate a pose from a picture of a person and apply it to a loaded VRM with humanoid bones. Load or drop an image, then press ▶ Run SAM3D. You can refine the result with options for mirrored images, hands and fingers, facing direction, grounding, head pitch, and wrist twist. Revert returns to the pose from before the first application.

If you load an MP4, WebM, or MOV video instead, ▶ Run SAM3D (Video) estimates movement across the selected time range and opens a motion-import dialog. You can then bring the result into the Pose track as keyframes. This mode is intended for one person per video; longer clips take more processing time and memory.

These tools require ComfyUI 0.38 or later, its native SAM 3D Body nodes, and a separately obtained SAM 3D Body model in models/detection/. The extension does not bundle that model or its weights, and the model has its own license. Foot rotation is not estimated from a still image.

Image-to-pose tab

Lip Sync and audio

On a compatible VRM, Lip Sync moves mouth expressions using Japanese kana or English text. ▶ Text previews mouth movement without audio. Timing is estimated from characters, so it is approximate; writing Japanese in kana gives more accurate timing than kanji.

For audio-based playback, you can select a generated audio file and let its loudness drive the mouth opening. Optional speech generation uses a locally running Lemonade or VOICEVOX engine; text-only Lip Sync does not require either one. The editor provides separate mouth-level controls for text and audio, and the Lip timeline track can record timed text or selected audio.

If you use VOICEVOX-generated audio, read its official terms (https://voicevox.hiroshiba.jp/term/) and the terms for the particular voice you choose. The editor does not add required credits to your published work.

Capturing your result

Press 📸 Capture to send the current view as a PNG to the node output. For repeated captures, enable ⏱ and set the node’s timer_interval between 1 and 3600 seconds; the default is 5 seconds. You can pause a loaded .vrma animation on a frame and capture or save that frame as a still pose.

comfyui-vrm-pose-editor Models

The extension does not bundle a character model. The file you choose determines which controls are available:

File or modelBest use
VRM (.vrm)Full character posing and VRM-dependent tools such as humanoid animation, compatible expressions, LookAt, spring bones, and image-to-pose transfer
GLB / GLTF (.glb, .gltf)Posing a skeleton-based 3D model when VRM-specific tools are not needed
VRM Animation (.vrma)Playing an existing animation on a loaded VRM, or importing it as editable pose keyframes
BVH (.bvh)Importing supported motion-capture animation as VRM pose keyframes

For an automatically loaded character, choose a file from model/ under Light → S → Default Model. Auto selects the first file by name, a specific filename selects that model, and None starts without one. You can also load a model manually whenever you need it.

The SAM 3D Body model used for image and video pose estimation is separate from the character model and is not supplied by this extension.

Default Model setting

Troubleshooting comfyui-vrm-pose-editor

ProblemWhat to check
The editor opens without a character.No model is bundled. Load a local file with VRM or Load MODEL, or check your Default Model choice.
A GLB or GLTF file loads, but VRM animation or image-to-pose does not work.Those tools require a loaded VRM with humanoid bones. Use a compatible VRM for them.
LookAt does nothing.Check whether the VRM includes LookAt data, then enable 👁 and move the cyan marker or choose the camera target.
Wind does not move hair or clothing.Enable 🎐 as well as 🌬. Wind only affects spring bones defined by the VRM.
The scroll wheel does not zoom.Open Light → S and enable Ctrl+Right drag zoom, then zoom using that gesture.
The camera jumps while you edit keyframes.Turn on 🖥 Monitor for a free view that ignores timeline camera switches during scouting and editing. Turn it off to preview the animated cameras.
Image-to-pose is unavailable or cannot run.Check that you have a compatible VRM, ComfyUI 0.38 or later, native SAM 3D Body nodes, and a SAM 3D Body model in models/detection/.
A .vrma export is missing lighting, camera cuts, or expressions.That export contains bone animation, not those preview-only effects. Render WebM, MP4, or GIF to preserve the visible scene.
MP4 export cannot complete.MP4 conversion requires ffmpeg. WebM or GIF are alternative outputs.

Frequently asked questions

Can I capture a single frame from an animation?
Yes. Load a .vrma, pause or scrub to the frame you want, then use 📸 Capture or the pose save controls.

Can I make an imported animation editable?
Yes. Choose VRMA (KEY) to sample a .vrma into Pose-track keyframes instead of loading it only as a playback clip.

Do I need a speech engine to animate the mouth?
No. Text-only Lip Sync works without Lemonade or VOICEVOX. Those engines are optional for generating audio.

Why does my character look different after loading another model?
Available bones, expressions, LookAt behavior, and spring-bone movement depend on the model itself. Check which features your new file provides.

Learn More about comfyui-vrm-pose-editor

comfyui-vrm-pose-editor 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-vrm-pose-editor detailed guide | ComfyUI