PLAYTEX AI

Make room for your next idea.

Opening your workspace…

Image to Three.js browser workspace

Your coding agent builds in its authorized project while this page receives the progress, editable source, review evidence, and sandboxed interactive preview. PLAYTEX AI does not launch a second agent or upload the reference image.

Create a browser session Agent protocol Read the complete guide

Your agent builds; the browser displays

Create a private 24-hour session, paste its generated prompt into Codex, Claude Code, Cursor, OpenCode, or another capable coding agent, and attach the reference there. The agent publishes meaningful stages and generated artifacts back to the open page.

A clear privacy boundary

The reference stays with the user's agent. Generated status, source, preview, sculpt spec, evidence, and artifact metadata are intentionally relayed to the private browser session. No PLAYTEX AI account or API key is required, and the session expires automatically.

A staged reconstruction instead of a one-shot mesh

  1. Validate whether the image is suitable and admit only readable references.
  2. Write a quality contract and inventory identity-defining details.
  3. Create and strict-validate the component, topology, material, pivot, and attachment spec.
  4. Build in locked passes from blockout through structure, form, materials, interaction, and optimization.
  5. Compare against the reference, review multiple useful angles, and self-correct within a bounded loop.
  6. Return the TypeScript factory, sculpt spec, review evidence, and honest fidelity and limitation notes.

Canonical agent outputs

  • createModel.ts with a factory returning THREE.Group
  • object-sculpt-spec.json with hierarchy, materials, attachments, pivots, sockets, and review state
  • Reference-versus-render comparison and multi-angle review evidence
  • Suitability, fidelity, assumptions, limitations, and reviewed-angle notes

GLB or GLTF export can be added downstream when the user requests it, but the core contract remains editable procedural source.

Agent workspace guide

What does the Image to Three.js workspace actually do?

It creates a private browser session that your own coding agent can use to publish procedural Three.js code, a sculpt specification, review evidence, artifacts, and a live preview. PLAYTEX AI displays the build; it does not launch a second agent.

Updated by PLAYTEX AI Editorial Team.

What changed

  • Added an answer-first explanation of the visitor-owned agent and private relay model.
  • Documented session lifetime, polling interval, output types, and the local-reference boundary.
  • Added a task table for setup, connection, review, and artifact export.

session lifetime

24h

Access expires automatically and expired relay data is purged.

update interval

2s

The browser checks the private relay for new generated output.

output views

4

Code, specification, evidence, and artifact tabs.

Decision table

Use the workspace according to who is doing the build

Decision guide for PLAYTEX AI Image to Three.js Workspace
SituationChooseWhy
You already use a coding agent with HTTP access. Create a browser session Copy the private prompt so that agent can publish progress and output here.
You want PLAYTEX AI to start an autonomous agent. Use your existing agent instead The page is a relay and review surface, not an agent launcher.
The reference image must stay on your machine. Keep it in the agent workspace Only generated output is sent through the relay.
You need reusable output. Download the active tab Export code or structured review artifacts from the browser session.

Scope and limitations

Output quality depends on the connected coding agent, its tools, and the reference. The browser isolates previews, but generated code still requires project-level review before production use.

How this page was reviewed

The review covered the current session API contract, visible privacy notes, polling behavior, output tabs, download paths, and agent protocol links.

Read the PLAYTEX AI editorial policy.