Trellis 2

Integrate

Model id trellis-2

# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.ru/v1/run/trellis-2?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

const API = "https://api.doitong.ru/v1";
const headers = {
  "Authorization": "Bearer " + process.env.DOITONG_API_KEY,
  "Content-Type": "application/json",
};

// Run the model; wait up to 20 s for the finished result in the same call
let res = await fetch(API + "/run/trellis-2?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  }),
});
let request = await res.json();
if (!res.ok) throw new Error(request.error.code + ": " + request.error.message);

// Not finished yet: long-poll until it is
while (request.status === "queued" || request.status === "running") {
  res = await fetch(API + "/requests/" + request.id + "?wait=20", { headers });
  request = await res.json();
}
if (request.status !== "succeeded") throw new Error(JSON.stringify(request.error));
console.log(request.outputs.map((o) => o.url), request.cost.credits + " API credits ($" + request.cost.usd + ")");

import os
import requests

API = "https://api.doitong.ru/v1"
headers = {"Authorization": "Bearer " + os.environ["DOITONG_API_KEY"]}

payload = {
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
}

# Run the model; wait up to 20 s for the finished result in the same call
r = requests.post(API + "/run/trellis-2", params={"wait": 20}, json=payload, headers=headers)
request = r.json()
if not r.ok:
    raise RuntimeError(request["error"]["code"] + ": " + request["error"]["message"])

# Not finished yet: long-poll until it is
while request["status"] in ("queued", "running"):
    request = requests.get(API + "/requests/" + request["id"], params={"wait": 20}, headers=headers).json()
if request["status"] != "succeeded":
    raise RuntimeError(request["error"])
print([o["url"] for o in request["outputs"]], request["cost"]["credits"], "API credits", "$" + request["cost"]["usd"])

# 1. Connect once (Claude Code; Cursor, VS Code, Claude Desktop: https://doitong.com/en-US/developers/docs/mcp)
claude mcp add --transport http doitong-models https://api.doitong.ru/v1/mcp \
  --header "Authorization: Bearer $DOITONG_API_KEY"

# 2. Your agent calls the "run" tool with:
{
  "model_id": "trellis-2",
  "input": {
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  }
}

Call the Trellis 2 model (trellis-2) through the Doitong API. Docs for agents: https://doitong.com/en-US/developers/models/trellis-2.md

POST https://api.doitong.ru/v1/run/trellis-2?wait=20
Authorization: Bearer $DOITONG_API_KEY
Content-Type: application/json

{
  "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
}

If the response status is "queued" or "running", poll GET https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20 until it is "succeeded"; the files are in outputs[].url. POST https://api.doitong.ru/v1/quote/trellis-2 with the same body returns the exact price before the run.

This request: $0.30 300 API credits · ≈ ₽31.64 Run it on this page

Get API key

Microsoft Image to 3D #image-to-3d

Turns a single image into a textured GLB 3D mesh with tunable structure, shape, texture and UV settings

OpenAPI Markdown

Microsoft · 3D

Your first result goes here

Build a request below and run it with your API key. Outputs are copied to our storage, kept forever or 7 days for free.

Build a request

Controls are generated from the model schema. The code and the price update as you change them.

image •required image_url

Drop a file, or

URL pointing to the source image for 3D conversion.

resolution default 1024

Resolution of the output; increasing it adds detail but takes longer.

Advanced texture_size · ss_guidance_strength · ss_guidance_rescale · ss_guidance_interval_start · ss_guidance_interval_end · ss_sampling_steps · ss_rescale_t · shape_slat_guidance_strength · shape_slat_guidance_rescale · shape_slat_guidance_interval_start · shape_slat_guidance_interval_end · shape_slat_sampling_steps · shape_slat_rescale_t · tex_slat_guidance_strength · tex_slat_guidance_rescale · tex_slat_guidance_interval_start · tex_slat_guidance_interval_end · tex_slat_sampling_steps · tex_slat_rescale_t · decimation_target · remesh_band · remesh_project · uv_unwrap_angle_threshold_deg · uv_unwrap_refine_iterations · uv_unwrap_global_iterations · uv_unwrap_smooth_strength · seed · remesh

texture_size default 2048

Mesh texture resolution for baking. Finer surface detail comes with larger files when this resolution is increased.

ss_guidance_strength default 7.5

Initial 3D structure fidelity control: increasing it favors closer agreement with the input image at the risk of additional noise.

ss_guidance_rescale default 0.7

Stage 1 guidance-artifact control, trading stronger guidance effects at low settings for a more stable output at high settings.

ss_guidance_interval_start default 0.6

Starting boundary of the classifier-free guidance window in stage 1 (sparse structure), expressed as a denoising fraction. Lower settings begin guidance sooner; only the interval from interval_start to interval_end uses guidance, with unguided processing elsewhere.

ss_guidance_interval_end default 1

Guidance cutoff for stage 1 (sparse structure): choose a value at least as large as ss_guidance_interval_start. The value identifies the fraction of denoising reached when classifier-free guidance stops.

ss_sampling_steps default 12

Initial-structure denoising count: potential quality gains from additional steps come with slower processing.

ss_rescale_t default 5

Set transition sharpness through the structure-generation noise schedule: increasing this value makes transitions sharper.

shape_slat_guidance_strength default 7.5

Detailed-geometry fidelity control: increasing adherence to the input image yields additional detail, with a risk of added noise.

shape_slat_guidance_rescale default 0.5

Increase this setting when geometry looks noisy: it reduces artifacts caused by strong guidance during the shape stage.

shape_slat_guidance_interval_start default 0.6

Classifier-free guidance activation point for shape refinement in stage 2. Expressed as a fraction of denoising progress, it sets interval_start; guidance operates only within the window bounded by interval_start and interval_end.

shape_slat_guidance_interval_end default 1

Guidance cutoff for stage 2 (shape refinement): choose a value at least as large as shape_slat_guidance_interval_start. The value identifies the fraction of denoising reached when classifier-free guidance stops.

shape_slat_sampling_steps default 12

Shape-refinement denoising count: potentially smoother geometry comes at the expense of speed when additional steps are used.

shape_slat_rescale_t default 3

Set geometric detail sharpness through the shape-refinement noise schedule; increasing this value yields sharper details.

tex_slat_guidance_strength default 1

Texture color fidelity control: closer adherence to colors in the input image at higher settings increases vividness but can cause oversaturation.

tex_slat_guidance_rescale default 0

Increase this setting to address noisy textures or color banding: it reduces artifacts caused by strong guidance during the texture stage.

tex_slat_guidance_interval_start default 0.6

Classifier-free guidance activation point for texture generation in stage 3. Expressed as a fraction of denoising progress, it sets interval_start; guidance operates only within the window bounded by interval_start and interval_end.

tex_slat_guidance_interval_end default 0.9

Guidance cutoff for stage 3 (texture): choose a value at least as large as tex_slat_guidance_interval_start. The value identifies the fraction of denoising reached when classifier-free guidance stops.

tex_slat_sampling_steps default 12

Number of denoising steps used to generate textures. Increasing the step count takes longer but may yield cleaner textures.

tex_slat_rescale_t default 3

Set texture detail sharpness through the texture-generation noise schedule; increasing this value yields sharper details.

decimation_target default 500000

Mesh vertex-count target, with 500k recommended for general use and 20k–50k for web/mobile. Reducing the target trades mesh detail for smaller files.

remesh_band default 1

Remeshing movement allowance: extra smoothing may sacrifice fine detail when this value is increased. The setting limits vertex displacement relative to the original surface.

remesh_project default 0

Projection strength for returning remeshed vertices to the original surface: 1 preserves detail through full projection, while 0 leaves projection off for a smoother result.

uv_unwrap_angle_threshold_deg default 90

UV chart angle limit in degrees, restricting how much neighboring face normals may differ within one chart. Raising the limit favors fewer, larger charts at the risk of stretching; reducing it favors more charts and less distortion.

uv_unwrap_refine_iterations default 0

Count of local refinement passes used to compute UV charts.

uv_unwrap_global_iterations default 1

UV chart balance can be improved by increasing this count of global clustering passes used to compute the charts.

uv_unwrap_smooth_strength default 1

Adjust how strongly UV chart edges are smoothed; less visible seams at higher settings come with a risk of UV distortion.

seed

Seed used for reproducible random generation.

POST /run/trellis-2?wait=20 quote → $0.30

# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.ru/v1/run/trellis-2?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

const API = "https://api.doitong.ru/v1";
const headers = {
  "Authorization": "Bearer " + process.env.DOITONG_API_KEY,
  "Content-Type": "application/json",
};

// Run the model; wait up to 20 s for the finished result in the same call
let res = await fetch(API + "/run/trellis-2?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  }),
});
let request = await res.json();
if (!res.ok) throw new Error(request.error.code + ": " + request.error.message);

// Not finished yet: long-poll until it is
while (request.status === "queued" || request.status === "running") {
  res = await fetch(API + "/requests/" + request.id + "?wait=20", { headers });
  request = await res.json();
}
if (request.status !== "succeeded") throw new Error(JSON.stringify(request.error));
console.log(request.outputs.map((o) => o.url), request.cost.credits + " API credits ($" + request.cost.usd + ")");

import os
import requests

API = "https://api.doitong.ru/v1"
headers = {"Authorization": "Bearer " + os.environ["DOITONG_API_KEY"]}

payload = {
    "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg",
}

# Run the model; wait up to 20 s for the finished result in the same call
r = requests.post(API + "/run/trellis-2", params={"wait": 20}, json=payload, headers=headers)
request = r.json()
if not r.ok:
    raise RuntimeError(request["error"]["code"] + ": " + request["error"]["message"])

# Not finished yet: long-poll until it is
while request["status"] in ("queued", "running"):
    request = requests.get(API + "/requests/" + request["id"], params={"wait": 20}, headers=headers).json()
if request["status"] != "succeeded":
    raise RuntimeError(request["error"])
print([o["url"] for o in request["outputs"]], request["cost"]["credits"], "API credits", "$" + request["cost"]["usd"])

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "run",
    "arguments": {
      "model_id": "trellis-2",
      "input": {
        "image": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
      },
      "wait_seconds": 20
    }
  }
}

Run this request

API key

Your key stays in this browser: it is sent only to the Doitong API and saved (localStorage) only if you tick Remember. Remember in this browser Get an API key

About Trellis 2

Builds a 3D asset from one image (`image_url`) and returns a GLB file. `resolution` is 512, 1024 or 1536, and `texture_size` is 1024, 2048 or 4096. Each of three stages (structure, shape, texture) has its own guidance and sampling-step controls. `decimation_target` sets the mesh vertex count (5,000 to 2,000,000), and remeshing plus UV-unwrapping options are exposed.

Parameters

Send either the canonical name or the native name, not both. Canonical names

NameType DefaultAllowed Description

seedoptional

Typeinteger | null

Seed used for reproducible random generation.

resolutionoptional

Typeenum

Default1024

51210241536

Resolution of the output; increasing it adds detail but takes longer.

ss_guidance_strengthoptional

Typenumber

Default7.5

0 – 10

Initial 3D structure fidelity control: increasing it favors closer agreement with the input image at the risk of additional noise.

ss_guidance_rescaleoptional

Typenumber

Default0.7

0 – 1

Stage 1 guidance-artifact control, trading stronger guidance effects at low settings for a more stable output at high settings.

ss_guidance_interval_startoptional

Typenumber

Default0.6

0 – 1

ss_guidance_interval_endoptional

Typenumber

Default1

0 – 1

ss_sampling_stepsoptional

Typeinteger

Default12

1 – 50

Initial-structure denoising count: potential quality gains from additional steps come with slower processing.

ss_rescale_toptional

Typenumber

Default5

1 – 6

Set transition sharpness through the structure-generation noise schedule: increasing this value makes transitions sharper.

shape_slat_guidance_strengthoptional

Typenumber

Default7.5

0 – 10

Detailed-geometry fidelity control: increasing adherence to the input image yields additional detail, with a risk of added noise.

shape_slat_guidance_rescaleoptional

Typenumber

Default0.5

0 – 1

Increase this setting when geometry looks noisy: it reduces artifacts caused by strong guidance during the shape stage.

shape_slat_guidance_interval_startoptional

Typenumber

Default0.6

0 – 1

shape_slat_guidance_interval_endoptional

Typenumber

Default1

0 – 1

shape_slat_sampling_stepsoptional

Typeinteger

Default12

1 – 50

Shape-refinement denoising count: potentially smoother geometry comes at the expense of speed when additional steps are used.

shape_slat_rescale_toptional

Typenumber

Default3

1 – 6

Set geometric detail sharpness through the shape-refinement noise schedule; increasing this value yields sharper details.

tex_slat_guidance_strengthoptional

Typenumber

Default1

0 – 10

Texture color fidelity control: closer adherence to colors in the input image at higher settings increases vividness but can cause oversaturation.

tex_slat_guidance_rescaleoptional

Typenumber

Default0

0 – 1

Increase this setting to address noisy textures or color banding: it reduces artifacts caused by strong guidance during the texture stage.

tex_slat_guidance_interval_startoptional

Typenumber

Default0.6

0 – 1

tex_slat_guidance_interval_endoptional

Typenumber

Default0.9

0 – 1

tex_slat_sampling_stepsoptional

Typeinteger

Default12

1 – 50

Number of denoising steps used to generate textures. Increasing the step count takes longer but may yield cleaner textures.

tex_slat_rescale_toptional

Typenumber

Default3

1 – 6

Set texture detail sharpness through the texture-generation noise schedule; increasing this value yields sharper details.

decimation_targetoptional

Typeinteger

Default500000

5,000 – 2,000,000

Mesh vertex-count target, with 500k recommended for general use and 20k–50k for web/mobile. Reducing the target trades mesh detail for smaller files.

texture_sizeoptional

Typeenum

Default2048

102420484096

Mesh texture resolution for baking. Finer surface detail comes with larger files when this resolution is increased.

remeshoptional

Typeboolean

Defaulttrue

Topology rebuilding that creates cleaner mesh triangles, usually improving suitability for later uses such as animation or 3D printing, at the expense of slower processing.

remesh_bandoptional

Typenumber

Default1

0 – 4

Remeshing movement allowance: extra smoothing may sacrifice fine detail when this value is increased. The setting limits vertex displacement relative to the original surface.

remesh_projectoptional

Typenumber

Default0

0 – 1

Projection strength for returning remeshed vertices to the original surface: 1 preserves detail through full projection, while 0 leaves projection off for a smoother result.

uv_unwrap_angle_threshold_degoptional

Typenumber

Default90

0 – 180

uv_unwrap_refine_iterationsoptional

Typeinteger

Default0

0 – 10

Count of local refinement passes used to compute UV charts.

uv_unwrap_global_iterationsoptional

Typeinteger

Default1

1 – 10

UV chart balance can be improved by increasing this count of global clustering passes used to compute the charts.

uv_unwrap_smooth_strengthoptional

Typenumber

Default1

0 – 10

Adjust how strongly UV chart edges are smoothed; less visible seams at higher settings come with a risk of UV distortion.

image_urlrequired image

Typeurl · image

URL pointing to the source image for 3D conversion.

required Canonical name, also accepted

File parameters take a public https:// URL, a data: URI up to 10 MB, or the file_url of an upload. Files

Output

GET /requests/req_8k2m4q7x1z9c3v5b6n0p2r4t

{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "trellis-2",
  "status": "succeeded",
  "created_at": "2026-10-20T12:00:00Z",
  "started_at": "2026-10-20T12:00:02Z",
  "completed_at": "2026-10-20T12:00:09Z",
  "queue_position": null,
  "cost": {
    "credits": 300,
    "usd": "0.30",
    "charged": 300,
    "refunded": 0
  },
  "input": {
    "image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
  },
  "output": {
    "model_glb": {
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_name": "0-output.jpeg",
      "file_size": 1843200
    }
  },
  "outputs": [
    {
      "type": "file",
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_size": 1843200,
      "stored": true
    }
  ],
  "error": null,
  "metrics": {
    "inference_seconds": 6.8
  },
  "urls": {
    "self": "https://api.doitong.ru/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t",
    "cancel": "https://api.doitong.ru/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t/cancel"
  }
}

Your files, your choiceEvery outputs[].url is a copy on our storage: kept forever (paid per GB-month) or 7 days for free, by your account setting or ?storage=.

Same shape on every modeltype, url, content_type, width, height and duration, for images, video and audio alike. output is the model’s own result object.

Fields of output

model_glb: object {url, content_type, file_name, file_size}3D model generated in GLB format.

More models: Image to 3D

For AI agents: this page as Markdown is at https://doitong.com/en-US/developers/models/trellis-2.md. All models and guides: https://doitong.com/developers/llms.txt

Your first result goes here

Build a request

Run this request

About Trellis 2

Parameters

Output

Other Trellis 2 variants

Trellis 2 · Retexture Variants: 2

More models: Image to 3D

Trellis Variants: 2

Trellis · Multi Variants: 2

TRELLIS.2 LoRA Inference

Hi3D Image to 3D Variants: 6