Skip to content

Free, open-source 3D sculpting for Linux, macOS and Windows: dense meshes, remeshing, painting, built-in AI (image/text to 3D), Send to Unreal, game-ready LODs, Gaussian splats and OpenUSD.

7.4 KB
Raw Blame

AI services: privacy and model licences

Every AI feature in SculptForge is optional and off until you use it. SculptForge has no telemetry, no account and no servers of its own. When you use an AI feature, SculptForge talks directly to the service you picked; nothing goes anywhere else.

Your art is never uploaded or used for training. AI features run only when you use them and can be turned off. Your sculpts stay on your computer; only what you hand an AI feature (a brush description, or the picture or text you give Generate 3D) goes to the service you picked.

Turning AI off

Preferences > AI > Enable AI features (also asked in the first-run welcome). Off: Brush Lab and Generate 3D are hidden, their commands (brush_ai, image_to_3d, text_to_3d, comfyui_check) are not registered in the app, sculptforge-cli, Python or the MCP tool list, and no AI request is made. The setting is "enabled" in ai.json. SF_AI=off in the environment (or --no-ai) turns it off and locks it, for studios; SF_AI=on turns it on whatever was saved. A studio's policy file locks it either way (see IT deployment).

What is sent where

Feature Service (your choice) What is sent Key
Brush Lab Free (default): Pollinations.ai's hosted open models, falling back to a running local Ollama. Or Anthropic (Claude), or any OpenAI-compatible endpoint such as a local Ollama / LM Studio Your brush description and the current brush settings (text only) None for Free and local models; your API key otherwise
Generate 3D (image) Hugging Face Spaces (free), fal.ai, Meshy, your own ComfyUI, or your own command The image you chose, cropped and resized to at most 1024 px Optional HF token; your fal.ai / Meshy key; none for ComfyUI
Generate 3D (text) Hugging Face Spaces (free), fal.ai, Meshy, your own ComfyUI Your text description As above
Send to Unreal Your own Unreal Editor (local network) The exported FBX None

Keys are stored in your operating system's keychain (macOS Keychain, Windows Credential Manager, or the Secret Service on Linux), never in a settings file, or read from environment variables (ANTHROPIC_API_KEY, OPENAI_API_KEY, OPENROUTER_API_KEY, HF_TOKEN, FAL_KEY, MESHY_API_KEY). Keys saved by older versions in ai.json move into the keychain at the first start.

Preferences > AI > Provider has presets for Anthropic, OpenAI, Ollama, LM Studio, vLLM, llama.cpp server, a LiteLLM proxy (a studio's own gateway, which can front Amazon Bedrock, Google Vertex AI and Azure OpenAI) and OpenRouter. Studios can limit which providers and addresses artists may use, require their single sign-on, and keep an audit log of every AI action: see IT deployment.

The free Brush Lab model

The Free provider sends your brush request to Pollinations.ai's anonymous, OpenAI-compatible text endpoint (currently GPT-OSS 20B, an Apache-2.0 open model). There's no sign-up and no key, and it only ever receives text: your description and the brush values. The anonymous tier allows about one request every few seconds, which is plenty for a chat. If the service can't be reached and Ollama is running on your machine, SculptForge uses your first local model instead. For fully private use, pick OpenAI-compatible and point it at Ollama or LM Studio: see Run AI on your own computer for step-by-step setup on Windows, macOS and Linux.

The free image-to-3D and text-to-3D models

The free option runs each model author's own public demo on Hugging Face's shared ZeroGPU hardware. SculptForge isn't affiliated with Hugging Face or any of the model authors, and doesn't host, bundle or redistribute any model. GPU time is rationed per user per day; a free Hugging Face token raises the allowance.

Each model has its own licence, and you're responsible for following it:

Model Author / provider Licence Key terms (summary, not legal advice)
TRELLIS.2 Microsoft (Space by Microsoft) MIT Permissive.
Hunyuan3D 2.1, Hunyuan3D 2, Hunyuan3D 2 mini Turbo Tencent (Spaces by Tencent) Tencent Hunyuan Community License Not licensed for use in the European Union, the United Kingdom or South Korea. Products over 1 million monthly active users need a separate licence from Tencent. Tencent claims no rights in your outputs. Outputs may not be used to improve other AI models.
Stable Fast 3D Stability AI (Space by Stability AI) Stability AI Community License Free, including commercial use, for individuals and organisations under US $1M annual revenue; above that, an enterprise licence is required. You own your outputs. "Powered by Stability AI".

These summaries are for convenience only. Read the licences linked above before using outputs commercially, and check where you are. If you're in the EU, UK or South Korea, choose TRELLIS.2, Stable Fast 3D or a paid service instead of Hunyuan3D.

Your own ComfyUI

With ComfyUI - your server, the image or description goes only to the ComfyUI address you enter: your own computer, or a GPU server at your studio. It runs there with whatever models that server has, and nothing is sent to anyone else. The picture is saved in ComfyUI's input folder and the result in its output folder, as ComfyUI does with any job. The licences of the models in your workflow apply as usual; the Hunyuan3D terms above apply to the bundled Hunyuan3D workflows too. See Generate 3D with your own ComfyUI.

Paid services

fal.ai, Meshy and Hyper3D are paid services with their own terms of service and pricing; generations are billed to your account with them. Model availability and names can change on their side.

Hyper3D Rodin 2.5 can run two ways:

  • fal.ai > Rodin 2.5: with your fal.ai key. The image or description goes to fal.ai, which runs Rodin.

  • Hyper3D Rodin (direct): with your own Hyper3D key, from the API dashboard at hyper3d.ai (a Creator plan or higher; Basic API access covers the Extreme Low and Low tiers, the others need Full access). The image or description goes to Hyper3D (api.hyper3d.com) and nowhere else; each generation is billed to your Hyper3D account, and Hyper3D keeps results as its data retention policy says.

    What SculptForge does to keep this narrow:

    • It is off until you enter your own key. The key is kept in the system keychain, never in ai.json, never in a log, and there is no environment variable, shared or trial key.
    • The API address is fixed in the app: https only, certificates checked, no redirects followed. The key is sent only there. The model file is accepted only over https from Hyper3D's own domains (hyper3d.com, hyper3d.ai, deemos.com), up to 200 MB, and only when it is a well-formed GLB; it is saved in a private temporary folder, read by the usual glTF importer and deleted after import.
    • It waits politely: the first status check after 5 s, then slower, up to every 30 s, for 20 minutes at most.
    • With Enable AI features off nothing is sent. Studios: the provider id is hyper3d in ai.providers (and https://api.hyper3d.com/ in ai.endpoints when that list is used); each generation writes one audit log line (tier, image or text, prompt length; never the key or the image).