Nodes/ComfyUI/Tripo P1: Text to Model
ComfyUI Node Runs on cloud

Tripo P1: Text to Model

Describe a prop, get a clean low-poly mesh — no image needed

By Comfy-Org·Created 4 years ago·Updated about 18 hours ago· 130,663
Tripo P1: Text to Model
    • model_file
    • model task_id
    • GLB
    prompt
    output_mode
    negative_prompt
    image_seed42
    face_limit-1
    model_seed42
    auto_sizefalse
    export_uvtrue
    compress_geometryfalse

    Tripo P1: Text to Model is the text-input sibling of Tripo's game-ready P1 engine - the one built for low-poly assets with stable topology, not render-farm sculptures. If you're prototyping props, weapons, or environmental pieces that need to survive contact with a game engine, this is the version of "text to 3D" you actually want: fewer triangles, cleaner edge flow, and a mesh that rigs instead of fighting you.

    Like every node in this family it's a Partner Node - built into ComfyUI core, nothing to install, and the work happens on Tripo's cloud. You log in with a ComfyUI Account, pay from prepaid credits, and the node polls Tripo until your mesh is done. The price badge shows an estimate that scales with your output_mode and texture_quality choices, so what you see before you run is roughly what you'll pay.

    The inputs that matter

    • prompt - required, up to 1024 characters. This is the whole creative payload, so spend it: material, style, shape, vibe. P1 rewards specific descriptions.
    • negative_prompt - optional, up to 255 characters. Genuinely useful here; telling it "no extra limbs, no broken geometry, avoid weird overhangs" costs nothing and steers the mesh.
    • output_mode - the same dynamic combo as the P1 image node. Geometry only gives you an untextured mesh at the lowest credit cost. Textured adds color/PBR maps and surfaces a pbr toggle and texture_quality (standard/detailed).
    • image_seed - a seed for the internal image step P1 uses to drive texturing. There's also a model_seed in the advanced section (shared by all P1 nodes) for the geometry pass, so you have one per stage if you want them.

    In advanced: face_limit (target face count 48–20,000, -1 = let Tripo choose adaptively), auto_size (scale output toward real-world meters), export_uv (UV unwrap during generation - off for faster geometry-only runs), and compress_geometry (geometry-based compression; decompress before editing).

    What comes out

    The standard Tripo outputs: model_file (string), model task_id, and the GLB. Preview it with the built-in Preview 3D & Animation node or save via Save 3D Model - files also land in ComfyUI/output/. The task_id is the bridge onward: it feeds Tripo: Rig, Retarget, and Convert, so a single prompt can walk all the way to an animated, rigged, format-converted asset with no external tools.

    Where it shines, and where it doesn't

    P1's whole identity is "game-ready," so don't expect the cinematic dense detail of the mainline Tripo v3.x or Hunyuan3D outputs - that's the point. If you need hero-asset fidelity, the classic Tripo: Text to Model with detailed geometry quality is the better spend. If you need a shelf of props at a polycount an engine won't blink at, P1 is the one. Keep face_limit honest for your target and remember textured runs cost more than geometry-only ones - the price badge is your friend.

    Categorypartner/3d/Tripo

    Inputs (9)

    NameTypeDefaultDescription
    promptSTRINGUp to 1024 characters.
    output_modeCOMBO"Geometry only" returns an untextured mesh. "Textured" adds color/PBR maps.
    negative_promptoptSTRINGUp to 255 characters.
    image_seedoptINT42
    face_limitoptINT-1-1–20000Target face count, 48-20000. -1 lets Tripo pick adaptively.
    model_seedoptINT42
    auto_sizeoptBOOLEANfalseScale the output to approximate real-world meters.
    export_uvoptBOOLEANtrueUV unwrap during generation. Turn off for faster geometry-only runs.
    compress_geometryoptBOOLEANfalseApply geometry-based compression. Decompress before editing.

    Outputs (3)

    NameTypeDescription
    model_fileSTRING
    model task_idMODEL_TASK_ID
    GLBFILE_3D_GLB