tabletopSign up

Features

Your whole asset pipeline on one canvas.

Link nodes, pick the AI for each step, and check each result as it is made. Your team and your agents work on the same canvas. All of this works in tabletop today.

  • Theo
Generate Image
Prompt *
References
Images
Provider
OpenAI
Model
GPT Image 2
Donev30:24
Images
Frog knight concept art
  • Maya
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Image to model
Donev11:48
Models
Untextured clay model of the frog knight
Texture Model
Model *
Prompt
Model
Texture model
v3.5
Quality
Detailed
Donev22:10
Model
Textured model of the frog knight
Rig Model
Model *
Rigged model
Rig type
biped
Bone naming
unreal
Running0:38

Placing the bones

Plain Text
Text
Text

A small frog knight with a dented bronze helmet, a lily-pad shield, a reed spear and a red scarf.

  • Theo
Generate Image
Prompt *
References
Images
Provider
OpenAI
Model
GPT Image 2
Quality
High
Donev30:24
Images
Frog knight concept art
  • Maya
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Image to model
Donev11:48
Models
Untextured clay model of the frog knight
Texture Model
Model *
Prompt
Model
Texture model
v3.5
Quality
Detailed
Donev22:10
Model
Textured model of the frog knight
Rig Model
Model *
Rigged model
Rig type
biped
Bone naming
unreal
Running0:38

Placing the bones

RunStep

Nodes and providers

One node for each task. You pick the AI.

A node does one job, such as Generate Image. Pick the provider that runs it, and change it at any time. Run one prompt on three models and keep the best take.

Plain Text
Text
Text

A mushroom witch with a lantern.

Generate Image
Prompt *
References
Images
Provider
OpenAI
Model
GPT Image 2
Donev10:21
Images
A mushroom witch, made by OpenAI
Generate Image
Prompt *
References
Images
Provider
Gemini
Model
Nano Banana 2.1
Donev10:12
Images
A mushroom witch, made by Gemini
Generate Image
Prompt *
References
Images
Provider
Leonardo
Model
FLUX.2 Pro
Donev10:17
Images
A mushroom witch, made by Leonardo
Plain Text
Text
Text

A mushroom witch with a lantern.

Generate Image
Prompt *
References
Images
Provider
OpenAI
Model
GPT Image 2
Donev10:21
Images
A mushroom witch, made by OpenAI
Generate Image
Prompt *
References
Images
Provider
Gemini
Model
Nano Banana 2.1
Donev10:12
Images
A mushroom witch, made by Gemini
Generate Image
Prompt *
References
Images
Provider
Leonardo
Model
FLUX.2 Pro
Donev10:17
Images
A mushroom witch, made by Leonardo
  • Images, video, 3D and text

    OpenAI, Gemini, Leonardo, Runware, Tripo and Anthropic, side by side on one canvas.

  • Agents as providers

    Claude and Codex agents can make an image, a model or a file in a closed worker.

  • Tokens or your own key

    Pay for each node with tokens, or with a named API key of your org.

3D

From a concept to a model that moves.

Turn an image into a model, paint it, give it a skeleton and add motion. Retarget a preset clip, or let a Blender agent rig the model to a standard skeleton.

Sam4m

Can the frog jump too?

Maya2m

Yes. Retarget Animation takes any clip: run, jump or slash.

Reply
Rig Model
Model *
Rigged model
Rig type
biped
Bone naming
unreal
Donev11:05
Rigged model
The frog knight with its skeleton
Retarget Animation
Rigged model *
Animated model
Animation
run
Donev10:52
Preview
Value *
Value
The frog knight mid-leap in its run clip
TexturedWireframeNormals

Sam4m

Can the frog jump too?

Maya2m

Yes. Retarget Animation takes any clip: run, jump or slash.

Reply
Rig Model
Model *
Rigged model
Rig type
biped
Bone naming
unreal
Donev11:05
Rigged model
The frog knight with its skeleton
Retarget Animation
Rigged model *
Animated model
Animation
run
Donev10:52
Preview
Value *
Value
The frog knight mid-leap in its run clip
TexturedWireframeNormals
  • Model, texture, rig

    Nodes for image to model, textures, rigs, segments, retopology, UVs and engine formats.

  • Any body

    Rig bipeds, quadrupeds, birds, snakes and more, with Unreal, Unity, Mixamo or VRM bone names.

  • See it move

    The model preview turns, plays each clip, and shows wireframe and normals.

2D and sprites

Sprites with clean pixels, from one concept.

Make a clip from an image model or a video model. Then clean it, give all frames one palette and export it to your engine. The sprite nodes use no AI: they cost no tokens and give the same result each time.

Theo1h

Where are the frame steps?

Maya52m

In the group: start frame, motion and clip. Double-click it to open it.

Reply
Generate Image
Prompt *
References
Images
Provider
Gemini
Model
Nano Banana 2.1
Donev10:14
Images
Axolotl ranger concept art
Generate Sprite (Video)

Start Frame

Concept art *

Motion Prompt

Motion

Sprite Clip

Sprite
Done · Sprite Clipv23:12
Limit Colors
Sprites *
Palette images
Sprites
Palette
Palette
Auto
DoneCached0:01
Export Sprite
Sprite *
Files
Format
Godot (SpriteFrames)
Done0:01
Preview
Value *
Value
run

Theo1h

Where are the frame steps?

Maya52m

In the group: start frame, motion and clip. Double-click it to open it.

Reply
Generate Image
Prompt *
References
Images
Provider
Gemini
Model
Nano Banana 2.1
Donev10:14
Images
Axolotl ranger concept art
Generate Sprite (Video)

Start Frame

Concept art *

Motion Prompt

Motion

Sprite Clip

Sprite
Done · Sprite Clipv23:12
Limit Colors
Sprites *
Palette images
Sprites
Palette
Palette
Auto
DoneCached0:01
Export Sprite
Sprite *
Files
Format
Godot (SpriteFrames)
Done0:01
Preview
Value *
Value
run
  • Clean frames

    Key out the backdrop, hold the feet in place, find the loop, remove halos and redraw each pixel on its grid.

  • One palette

    Limit colors, recolor, outline and dither, and paint single pixels in the Pixel Editor.

  • Ready for your engine

    Export to Godot, Unity, Unreal (Paper2D) or JSON and PNG, with normal maps for 2D light.

Pixel art from any image or model.

Dither turns a painting into pixel art with one palette. Dither Textures does the same to the textures of a GLB model.

Generate Image
Prompt *
References
Images
Provider
OpenAI
Donev10:19
Images
Painted jungle temple ruins
Dither
Images *
Images
Color mode
Auto palette
Dither
Bayer 4×4
Done0:01
Images
The ruins as dithered pixel art
Generate Image
Prompt *
References
Images
Provider
OpenAI
Donev10:19
Images
Painted jungle temple ruins
Dither
Images *
Images
Color mode
Auto palette
Dither
Bayer 4×4
Done0:01
Images
The ruins as dithered pixel art

Hair that moves.

Generate Hair makes hair with spring bones. Attach Hair fits it to your rigged character, and the GLB export keeps the springs.

Generate Hair
Character image *
Hair
Variations
1
Spring bones
on
Donev13:30
Attach Hair
Character *
Hair *
Scene
Donev15:12
Scene
A nomad girl with spring bones in her hair
Generate Hair
Character image *
Hair
Variations
1
Spring bones
on
Donev13:30
Attach Hair
Character *
Hair *
Scene
Donev15:12
Scene
A nomad girl with spring bones in her hair

Play test

Play it before you ship it.

Link sprites or a model to Play Test and play them in a game world. Check the clips, the scale, the speed and the hair with the keyboard, the mouse or a gamepad.

  • Four presets

    Top-down, side-scroller, billboard and third-person.

  • Game feel

    Set the walk and run speed, the jump height, the gravity and the collider of each preset.

  • Spring physics

    Hair and other spring bones move as they do in your game.

Play Test
Sprites
Model
Game
The axolotl ranger runs through a pixel art sci-fi levelSide-scroller · click to play
Combine Sprites
Sprites *
Sprite
Done0:01
Play Test
Sprites
Model
Game
The axolotl ranger runs through a pixel art sci-fi levelSide-scroller · click to play

Human in the loop

You approve each step.

Run step by step and check each result before the next node starts. Each result is kept as a version, so you can go back at any time.

Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Text to model
Needs approvalv31:52
ApproveRe-run
Models
A copper airship with cloth sails
Texture Model
Model *
Prompt
Model
Quality
Detailed
Waiting for inputs

Versions

  • Airship version 3v3Needs approval · 1:52
  • Airship version 2v2Done · 1:47
  • Airship version 1v1Done · 2:03
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Text to model
Needs approvalv31:52
ApproveRe-run
Models
A copper airship with cloth sails
Texture Model
Model *
Prompt
Model
Quality
Detailed
Waiting for inputs
Export Model
Scene *
Model
Donev20:31

Versions

  • Airship version 3v3Needs approval · 1:52
  • Airship version 2v2Done · 1:47
  • Airship version 1v1Done · 2:03
  • Step mode

    Step pauses after each job. Approve the result, or run the node again.

  • Versions and locks

    Each result is a version. Lock a node to keep a good result when its inputs change.

  • Run only what changed

    A changed node shows Outdated. Select the nodes to run. A node with no change comes from the cache.

Multiplayer

Build it together, live.

Your team edits one canvas at the same time. See each cursor and selection, leave notes on the canvas, and watch runs progress with no reload.

Ren3m

Make the lighthouse taller and keep the warm windows.

Maya1m

On it. Running v3 now.

Reply
  • Maya
Generate Image
Prompt *
References
Images
Provider
Leonardo
Size
1376×768 (16:9)
Donev20:22
Images
A moonlit harbor town diorama

Ren3m

Make the lighthouse taller and keep the warm windows.

Maya1m

On it. Running v3 now.

Reply
Plain Text
Text
Text

A moonlit harbor town on a rock: wooden piers, a lighthouse, fishing boats, warm windows.

  • Maya
Generate Image
Prompt *
References
Images
Provider
Leonardo
Size
1376×768 (16:9)
Donev20:22
Images
A moonlit harbor town diorama
  • Sam
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Image to model
Queued0:03
  • Cursors and presence

    See who is on the canvas, where they point and what they select.

  • Your own undo

    Undo takes back your changes, not the changes of your team.

  • Notes

    Talk it over in a thread on the canvas, next to the nodes you mean.

Agents

Your agent drives. You watch the canvas.

Connect Claude Code, Codex or Cursor through MCP, or use the tabletop CLI. The agent builds the workflow, adds notes, gets the cost and starts runs. You see each edit as it happens.

> Make a crystal golem and rig it for Unreal.

tabletop workflows edit-graph
 Added Generate Model, AI Rig Model and a note.
tabletop runs quote
 Up to 96 tokens.
tabletop runs start
 AI Rig Model needs your approval.
  • MCP and CLI

    Each agent action is an MCP tool and a tabletop command. Sign in with OAuth.

  • Blender agents

    AI Rig Model, AI Animation and Attach Hair run an agent with Blender in a closed worker.

  • Your keys stay yours

    An agent never gets a provider key.

Claude Code6m

Step 2: rig the golem to the standard rig. Check the bones, then approve.

Mayanow

Bones look good. Approving.

Reply
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Text to model
Donev11:40
AI Rig Model
Scenes *
Scene
Provider
Agent: Claude
Model
Claude Opus 5.5
Needs approvalv16:40
ApproveRe-run
Scene
A crystal golem with its skeleton

Claude Code6m

Step 2: rig the golem to the standard rig. Check the bones, then approve.

Mayanow

Bones look good. Approving.

Reply
Generate Model
Prompt
Image
Models
Provider
Tripo
Type
Text to model
Donev11:40
AI Rig Model
Scenes *
Scene
Provider
Agent: Claude
Model
Claude Opus 5.5
Needs approvalv16:40
ApproveRe-run
Scene
A crystal golem with its skeleton

Templates and groups

Make a step once. Use it everywhere.

Turn any node into a template with new defaults and hidden settings. Write prompt templates with fields. Collapse many nodes into one group.

Creature Views
Creature Type *
Views *
Text
Creature Type
pangolin
Views
front, side, back
Generate Image
Prompt *
References
Images
Provider
OpenAI
Size
1536x1024
Donev10:26
Images
Front, side and back views of a crystal pangolin
Creature ViewsTemplate
Text
Text
A {{creatureType}} creature. Views: {{views}}
Creature Views
Creature Type *
Views *
Text
Creature Type
pangolin
Views
front, side, back
Generate Image
Prompt *
References
Images
Provider
OpenAI
Size
1536x1024
Donev10:26
Images
Front, side and back views of a crystal pangolin
  • Org templates

    A change to a template updates each node that uses it, in each workflow of your org.

  • Prompt fields

    Type {{creatureType}} in a prompt template, and each node that uses it gets a field to fill in.

  • Groups

    A group shows only the ports and previews that you pick. Double-click to open it.

AI Animation

Animate your rig from a video.

Make a video of the motion, or use a video of your own. An agent animates your rigged character to match it. Name the clips, then export them to your engine.

Generate Video
Prompt *
Start frame
Video
Provider
Leonardo
Model
Seedance 2.5
Donev12:41
Video
A reference video of a gray figure that walks in side view
0:02 / 0:08
AI Rig Model
Models
Scene
Provider
Agent: Claude
Model
Claude Opus 5.5
Donev16:40
AI Animation
Scenes
Videos
Scene
Provider
Agent: Claude
Instructions
One clip named walk
Donev18:12
Scene
The crystal golem in its walk clip
Export Model
Scene *
Model
Format
FBX (.fbx)
Donev10:18
Plain Text
Text
Text

The golem walks forward in a heavy loop, seen from the side, full body in frame.

Generate Video
Prompt *
Start frame
Video
Provider
Leonardo
Model
Seedance 2.5
Donev12:41
Video
A reference video of a gray figure that walks in side view
0:02 / 0:08
AI Rig Model
Models
Scene
Provider
Agent: Claude
Model
Claude Opus 5.5
Donev16:40
AI Animation
Scenes
Videos
Scene
Provider
Agent: Claude
Instructions
One clip named walk
Donev18:12
Scene
The crystal golem in its walk clip
Export Model
Scene *
Model
Format
FBX (.fbx)
Donev10:18
  • From any video

    Generate Video shows your character in the motion. Or link a video file that you have.

  • Clips you name

    Write the clips in Instructions, such as walk, idle and wave.

  • Ready for your engine

    Export Model writes an FBX file for Unreal, Unity or Godot.

And more

The rest of the table.

  • Workflows page

    The workflows of all your orgs in one place, each with a screenshot of its canvas.

  • Starter workflows

    Start from a ready workflow, then change it to fit your game.

  • Asset library

    Each result goes to your library. Load it into any workflow with one node.

  • A preview for each file

    Images, video, sprites and 3D models, with file info and sharp views of pixel art.

  • Safe runs

    A queue runs each node with limits and retries. A crash never starts a paid job two times.

  • Tokens or your own keys

    Pay with tokens, or with named provider keys of your org, sealed with AES-256-GCM.

  • Cost before you run

    The toolbar shows the most tokens that a run of the selected nodes can use.

  • Sign in your way

    Sign in with email, Google or GitHub. Your team shares one org.

FAQ

Questions about the features

How does a tabletop workflow work?+

A workflow is a canvas of nodes. Each node does one job, such as Generate Image, and links carry its result to the next node. Select the nodes you want and press Run: tabletop runs them in order and shows each result as it is made.

Can tabletop rig and animate 3D models?+

Yes. Turn an image into a model, texture it, and give it a skeleton with Rig Model or AI Rig Model. Then retarget a preset clip, such as run or jump. Or use AI Animation: an agent animates the rig to match a video of the motion.

Can tabletop make pixel art sprites?+

Yes. Make a clip with an image model or a video model. Then clean the frames, give them one palette and export them to Godot, Unity or Unreal (Paper2D). The sprite nodes use no AI, so they cost no tokens and give the same result each time.

Can I try my assets in a game before I export them?+

Yes. The Play Test node lets you play your sprites and models in game presets, such as a side-scroller, right on the canvas.

Can my team work on the same canvas?+

Yes. Your team edits one canvas at the same time. You see each cursor and selection. Undo takes back only your own changes, and runs update live with no reload.

Can Claude Code or Codex build workflows in tabletop?+

Yes. Connect Claude Code, Codex or Cursor through MCP, or use the tabletop CLI. The agent builds the workflow, adds notes, gets the cost and starts runs. Steps that need approval wait for you.

Can I use my own API keys?+

Yes. Add named provider keys to your org. tabletop seals each key with AES-256-GCM. A run that uses your key uses no tokens.

Pull up a chair.

Start on the free plan. Your first workflow is one prompt away.