# TRELLIS.2 LoRA Inference API

> Trellis 2 da maletg a 3D che po applitgar adaptaders LoRA persunalisads per structura, geometria e textur

Generescha in GLB da `image_url` e po applitgar LoRAs safetensors via `sparse_structure_lora_url`, `geometry_lora_url` e `texture_lora_url`. `resolution` è 512 u 1024 e sto correspunder a la resoluziun d'entrenament dals adaptaders. Disponibels èn er controlls per guidance e pass per staziun, `decimation_target` e `texture_size`.

- Model id: `trellis-2-lora`
- Made by: Microsoft
- Category: Image to 3D
- Output: 3d
- Price: from $0.25 / request (cheapest run $0.25 (250 API credits))
- Run: `POST https://api.doitong.ru/v1/run/trellis-2-lora`
- Quote: `POST https://api.doitong.ru/v1/quote/trellis-2-lora`
- Docs page: https://doitong.com/rm/developers/models/trellis-2-lora

## Pricing

Price: **from $0.25 / request** (the lowest tier; the quote prices your exact input). Billed once per request.

Price rules (the first rule that matches the input wins; a parameter you leave out counts as its default):

- When `resolution` = 1024: $0.30 (300 API credits) / request

The full example input below costs $0.25 = 250 API credits (1 request × 250 API credits).

Prices are the upstream price in USD, paid from the API balance in API credits (1000 API credits = 1 USD). `POST https://api.doitong.ru/v1/quote/trellis-2-lora` with the same JSON body returns the exact price for that input; the quote is exactly what a run charges. Invalid input is never charged and failed or canceled requests are refunded automatically.

## Input parameters

Send either the canonical name or the native field name (not both). Canonical names are the same across models.

| Parameter | Canonical name | Type | Required | Default | Allowed values / range | Description |
|---|---|---|---|---|---|---|
| `image_url` | `image` | string (image URL) | yes |  |  | URL dal maletg da funtauna per la conversiun en in model 3D. |
| `sparse_structure_lora_url` |  | string \| null | no |  |  | Checkpoint LoRA opziunal en format safetensors, trenà spezialmain per il denoiser da la structura sparsa. |
| `geometry_lora_url` |  | string \| null | no |  |  | Checkpoint LoRA opziunal en format safetensors, trenà spezialmain per il denoiser da la geometria. |
| `texture_lora_url` |  | string \| null | no |  |  | Checkpoint LoRA opziunal en format safetensors, trenà spezialmain per il denoiser da la textura. |
| `seed` |  | integer \| null | no |  |  | Semenza casuala che renda la generaziun reproducibla. |
| `resolution` |  | enum (integer) | no | `512` | `512`, `1024` | Resoluziun da generaziun, cun 1024 che tscherna la pipeline cascade. Tscherna la resoluziun utilisada per trenar ils adapters LoRA furnids; ina correspundenza è necessaria. |
| `ss_guidance_strength` |  | number | no | `7.5` | 0 – 10 | Forza da la guidance senza classificader durant la generaziun da la structura sparsa. |
| `ss_guidance_rescale` |  | number | no | `0.7` | 0 – 1 | Rescaling da la guidance durant la generaziun da la structura sparsa. Valurs pli grondas rendan l'output pli stabil. |
| `ss_guidance_interval_start` |  | number | no | `0.6` | 0 – 1 | Fracziun dal progress da reducziun dal ramur nua che la guidance da la structura sparsa cumenza. |
| `ss_guidance_interval_end` |  | number | no | `1` | 0 – 1 | Utilisescha ina fracziun da reducziun dal ramur almain eguala a `ss_guidance_interval_start` per inditgar cura che la guidance da la structura sparsa sa ferma. |
| `ss_sampling_steps` |  | integer | no | `12` | 1 – 50 | Dumber da pass da reducziun dal ramur durant la generaziun da la structura sparsa. |
| `ss_rescale_t` |  | number | no | `5` | 1 – 6 | Nettezza dal plan da ramur durant la generaziun da la structura sparsa. |
| `shape_slat_guidance_strength` |  | number | no | `7.5` | 0 – 10 | Forza da la guidance senza classificader durant la generaziun da la furma. |
| `shape_slat_guidance_rescale` |  | number | no | `0.5` | 0 – 1 | Rescaling applitgà a la guidance durant la generaziun da la furma; augmenta ella sche la geometria para ramurusa. |
| `shape_slat_guidance_interval_start` |  | number | no | `0.6` | 0 – 1 | Fracziun da reducziun dal ramur nua che la guidance da la fasa da la furma cumenza. |
| `shape_slat_guidance_interval_end` |  | number | no | `1` | 0 – 1 | Punct da fermada da la guidance da la fasa da la furma, exprimì sco fracziun da reducziun dal ramur betg pli pitschna che shape_slat_guidance_interval_start. |
| `shape_slat_sampling_steps` |  | integer | no | `12` | 1 – 50 | Dumber da pass da reducziun dal ramur durant la generaziun da la furma. |
| `shape_slat_rescale_t` |  | number | no | `3` | 1 – 6 | Nettezza dal plan da ramur durant la generaziun da la furma. |
| `tex_slat_guidance_strength` |  | number | no | `1` | 0 – 10 | Forza da la guidance senza classificader durant la generaziun da la textura. |
| `tex_slat_guidance_rescale` |  | number | no | `0` | 0 – 1 | Rescaling applitgà a la guidance durant la generaziun da la textura; augmenta ella sche las texturas paran ramurusas. |
| `tex_slat_guidance_interval_start` |  | number | no | `0.6` | 0 – 1 | Fracziun da reducziun dal ramur nua che la guidance da la fasa da la textura cumenza. |
| `tex_slat_guidance_interval_end` |  | number | no | `0.9` | 0 – 1 | Punct da fermada da la guidance da la fasa da la textura, exprimì sco fracziun da reducziun dal ramur betg pli pitschna che tex_slat_guidance_interval_start. |
| `tex_slat_sampling_steps` |  | integer | no | `12` | 1 – 50 | Dumber da pass da reducziun dal ramur durant la generaziun da la textura. |
| `tex_slat_rescale_t` |  | number | no | `3` | 1 – 6 | Nettezza dal plan da ramur durant la generaziun da la textura. |
| `decimation_target` |  | integer | no | `500000` | 5,000 – 2,000,000 | Dumber giavischà da vertizels en l'export GLB. |
| `texture_size` |  | enum (integer) | no | `2048` | `1024`, `2048`, `4096` | Resoluziun da las texturas incrustadas en l'export GLB. |
| `enable_safety_checker` |  | boolean | no | `true` |  | La controlla po vegnir schluccada cun definir false, premess che l'appelant saja autorisà. |

File parameters (`image`) take a public `https://` URL, a `data:` URI up to 10 MB, or the `file_url` of an upload from `POST /files`.

## Example input

Required parameters only:

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

A fuller example:

```json
{
  "image_url": "https://content.doitong.com/images/showcase/v3/gen_fox_cartoon.jpg"
}
```

## Run it

Set `DOITONG_API_KEY` to a key from https://doitong.com/developers/dashboard#keys. `?wait=20` returns the finished request in the same call when it is ready within 20 seconds; otherwise poll `GET /requests/{id}` (it also accepts `?wait=`), or pass `webhook_url`.

### curl

```bash
# 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-lora?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "image_url": "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"
```

### JavaScript

```javascript
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-lora?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "image_url": "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 + ")");
```

### Python

```python
import os
import requests

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

payload = {
    "image_url": "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-lora", 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"])
```

### MCP

The MCP server at `https://api.doitong.ru/v1/mcp` exposes the tools `search_models`, `get_model`, `quote`, `run`, `get_request`, `cancel_request`, `list_requests`, `create_upload` and `get_account`. A `run` call for this model:

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

## Response

Every call returns a request object. `status` is one of `queued`, `running`, `succeeded`, `failed`, `canceled`. `outputs` lists the generated files with URLs on our storage (kept forever, or 7 days with `?storage=temporary`; `storage` says which); `output` is the model's own result object with the same URLs.

```json
{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "trellis-2-lora",
  "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": 250,
    "usd": "0.25",
    "charged": 250,
    "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
    },
    "seed": 42,
    "resolution": 1,
    "sparse_structure_lora_url": "string",
    "geometry_lora_url": "string",
    "texture_lora_url": "string",
    "decimation_target": 1,
    "texture_size": 1
  },
  "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"
  }
}
```

### Output fields (`output`)

| Field | Type | Description |
|---|---|---|
| `model_glb` | object {url, content_type, file_name, file_size} | Datoteca GLB che cuntegna il model 3D generà. |
| `seed` | integer | Semenza applitgada durant la generaziun. |
| `resolution` | integer | Resoluziun a la quala la generaziun è vegnida exequida. |
| `sparse_structure_lora_url` | string \| null | Checkpoint LoRA da la structura sparsa utilisà durant la generaziun, sch'in tal è vegnì applitgà. |
| `geometry_lora_url` | string \| null | Checkpoint LoRA da la geometria utilisà durant la generaziun, sch'in tal è vegnì applitgà. |
| `texture_lora_url` | string \| null | Checkpoint LoRA da la textura utilisà durant la generaziun, sch'in tal è vegnì applitgà. |
| `decimation_target` | integer | Finamira dal dumber da vertizels applitgada a l'export GLB. |
| `texture_size` | integer | Resoluziun da las texturas incrustadas en l'export GLB. |

## Related models

- [Trellis](https://doitong.com/developers/models/trellis.md) (`trellis`): $0.02 / request
- [Trellis 2](https://doitong.com/developers/models/trellis-2.md) (`trellis-2`): from $0.25 / request
- [Trellis 2 · Retexture](https://doitong.com/developers/models/trellis-2/retexture.md) (`trellis-2/retexture`): from $0.20 / request
- [Trellis · Multi](https://doitong.com/developers/models/trellis/multi.md) (`trellis/multi`): $0.02 / request
- [Hi3D Image to 3D](https://doitong.com/developers/models/hitem3d/hi3d/image-to-3d.md) (`hitem3d/hi3d/image-to-3d`): $0.50 / run
- [Hi3D Image to 3D · v3.0](https://doitong.com/developers/models/hitem3d/hi3d/v3.0/image-to-3d.md) (`hitem3d/hi3d/v3.0/image-to-3d`): from $1.80 / run
- [Hi3D Multiview to 3D · Multi View to 3D](https://doitong.com/developers/models/hitem3d/hi3d/multi-view-to-3d.md) (`hitem3d/hi3d/multi-view-to-3d`): $0.50 / run
- [Hi3D Multiview to 3D · v3.0 · Multi View to 3D](https://doitong.com/developers/models/hitem3d/hi3d/v3.0/multi-view-to-3d.md) (`hitem3d/hi3d/v3.0/multi-view-to-3d`): from $1.80 / run

## Links

- HTML page: https://doitong.com/developers/models/trellis-2-lora
- All models and guides for AI agents: https://doitong.com/developers/llms.txt
- OpenAPI 3.1 with this model's typed request body: https://doitong.com/developers/openapi.json?model=trellis-2-lora
- Errors: https://doitong.com/developers/docs/errors
- Model details as JSON: https://api.doitong.ru/v1/models/trellis-2-lora