Docs

Running a model

MicrosoftImage to 3D

Microsoft TRELLIS.2

High-fidelity image-to-3D generative model with compact structured latents

POSTauray-serverless/microsoft/trellis-2-4b
openapi.jsonllms.txt

One of Auray's catalogue models: the same key, the same credits, the same job lifecycle as every other address.

Its request shape is published below. Whether it runs right now is GET /v1/catalogue's word, with no key needed: its addresses say connected for an address that runs, and a submit to any other is refused before anything is charged — described answers 501 model_not_connected, paused 503 model_disabled.

Category image-to-3d · Scope 3d:write (or external:write on an older key) · Price Credits per call (counted as settings.shapeSlat.steps+settings.sparseStructure.steps+settings.texSlat.steps@36*settings.decimationTarget@500000*settings.textureSize@2048): 2.669 for settings.resolution=1536; 3 for settings.resolution=512; settings.resolution=1024. Each band has its own least and most a request is counted at. Quoted when you submit, at the most the request can cost; the charge settles to what the job cost and never above the quote. The submit's answer carries the quote as credits_quoted, and credits_charged says what has been taken.

#Input

FieldTypeRequiredDefaultDescription
inputsobjectyes—
inputs.imagestringyes—Image input (an http(s) URL; upload a local file first and send its read URL). A URL.
inputs.meshFilestringno—3D model to re-texture (an http(s) URL; upload a local file first and send its read URL). Skips geometry generation and applies new textures using the reference image. A URL.
seedintegerno—Random seed for reproducible generation. When not provided, a random seed is generated in the unsigned 32-bit range. Documented range: 0 to 9,223,372,036,854,776,000.
settingsobjectno—
settings.resolutionintegerno1024Voxel grid resolution for 3D generation. Higher values capture finer geometric detail. One of 512, 1024, 1536.
settings.alphaModestringno"OPAQUE"Material alpha mode for the output mesh. One of "OPAQUE", "BLEND", "MASK".
settings.decimationTargetintegerno500000Target polygon count for mesh simplification. Lower values produce lighter models at the cost of geometric detail. Documented range: 5,000 to 1,000,000.
settings.dracoCompressionobjectno—
settings.dracoCompression.enabledbooleanno—Compress the output GLB geometry with Draco.
settings.dracoCompression.levelintegerno7Encoding effort for Draco compression. Higher values reduce file size further but encode more slowly. Documented range: 0 to 10.
settings.dracoCompression.quantizationNormalintegerno14Quantization bits for vertex normals. Higher values produce smoother shading at the cost of file size. Documented range: 8 to 30.
settings.dracoCompression.quantizationPositionintegerno16Quantization bits for vertex positions. Higher values preserve more positional detail at the cost of file size. Documented range: 8 to 30.
settings.dracoCompression.quantizationTexCoordintegerno14Quantization bits for texture coordinates. Higher values keep textures sharper at UV seams at the cost of file size. Documented range: 8 to 30.
settings.meshClusterobjectno—
settings.meshCluster.globalIterationsintegerno1Global iterations for chart clustering. Documented range: 0 to 64.
settings.meshCluster.refineIterationsintegerno0Local refinement iterations for chart clustering. Documented range: 0 to 64.
settings.meshCluster.smoothStrengthintegerno1Smoothing strength applied during chart clustering. Documented range: 0 to 10.
settings.meshCluster.thresholdConeHalfAngleRadnumberno1.5707963267948966Cone half-angle threshold in radians used during UV chart clustering. Documented range: 0 to 3.142.
settings.remeshbooleannotrueWhether to remesh the output for cleaner topology.
settings.remeshBandnumberno1Narrow-band width in voxels used for dual-contouring remeshing. Documented range: up to 16.
settings.remeshProjectnumberno0Projection factor for snapping remeshed vertices back to the original surface. Documented range: up to 1.
settings.shapeSlatobjectno—
settings.shapeSlat.stepsintegerno12Number of denoising steps. More steps improve quality but increase generation time. Documented range: 1 to 50.
settings.shapeSlat.guidanceIntervalEndnumberno1End of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.shapeSlat.guidanceIntervalStartnumberno0.6Start of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.shapeSlat.guidanceRescalenumberno0.5Dampens oversaturation from high guidance values. Lower values apply more correction. Documented range: 0 to 1.
settings.shapeSlat.guidanceStrengthnumberno7.5How closely the output follows the input image. Higher values enforce stricter adherence to the visible geometry. Documented range: 1 to 10.
settings.shapeSlat.rescaleTnumberno3Timestep rescaling factor for the sampling schedule. Controls how guidance strength is distributed across steps. Documented range: 1 to 6.
settings.sparseStructureobjectno—
settings.sparseStructure.stepsintegerno12Number of denoising steps. More steps improve quality but increase generation time. Documented range: 1 to 50.
settings.sparseStructure.guidanceIntervalEndnumberno1End of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.sparseStructure.guidanceIntervalStartnumberno0.6Start of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.sparseStructure.guidanceRescalenumberno0.7Dampens oversaturation from high guidance values. Lower values apply more correction. Documented range: 0 to 1.
settings.sparseStructure.guidanceStrengthnumberno7.5How closely the output follows the input image. Higher values enforce stricter adherence to the visible geometry. Documented range: 1 to 10.
settings.sparseStructure.rescaleTnumberno5Timestep rescaling factor for the sampling schedule. Controls how guidance strength is distributed across steps. Documented range: 1 to 6.
settings.texSlatobjectno—
settings.texSlat.stepsintegerno12Number of denoising steps. More steps improve quality but increase generation time. Documented range: 1 to 50.
settings.texSlat.guidanceIntervalEndnumberno0.9End of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.texSlat.guidanceIntervalStartnumberno0.6Start of the sampling schedule fraction over which guidance is applied. Documented range: 0 to 1.
settings.texSlat.guidanceRescalenumberno0Dampens oversaturation from high guidance values. Lower values apply more correction. Documented range: 0 to 1.
settings.texSlat.guidanceStrengthnumberno1How closely the output follows the input image. Higher values enforce stricter adherence to the visible geometry. Documented range: 1 to 10.
settings.texSlat.rescaleTnumberno3Timestep rescaling factor for the sampling schedule. Controls how guidance strength is distributed across steps. Documented range: 1 to 6.
settings.textureFormatstringno"WEBP"Texture format for the output mesh. One of "WEBP", "PNG".
settings.textureSizeintegerno2048Resolution of the output texture map in pixels. Higher values produce finer surface detail. Documented range: 1,024 to 4,096.
outputTypestringno"URL"3D output type. One of "URL".
outputFormatstringno"GLB"Specifies the file format of the generated output. The available values depend on the task type and the specific model's capabilities. - `JPG`: Best for photorealistic images with smaller file sizes (no transparency). - `PNG`: Lossless compression, supports high quality and transparency (alpha channel). - `WEBP`: Modern format providing superior compression and transparency support. - `MP4`: Widely supported video container (H.264), recommended for general use. - `WEBM`: Optimized for web delivery. - `MOV`: QuickTime format, common in professional workflows (Apple ecosystem). - `GIF`: Animated image format (no audio), suitable for short loops or previews. - `MP3`: Compressed audio, smaller file size. - `WAV`: Uncompressed, high-quality audio. - `FLAC`: Lossless compression. - `OGG`: Open-source compressed audio format (Vorbis codec). - `SVG`: Scalable Vector Graphics. - `TIFF`: High-quality output supporting layers. - `GLB`: Binary glTF format for 3D models, including geometry, textures, and PBR materials. > [!NOTE] > **Transparency**: If you are using features like background removal or LayerDiffuse that require transparency, you must select a format that supports an alpha channel (e.g., `PNG`, `WEBP`, `TIFF`). `JPG` does not support transparency. One of "GLB".
idempotency_keystringno—Your own name for this request. Send the same one twice and the second call is answered with the first request rather than charged again. Idempotency-Key: as a header is accepted as an alias — that is the spelling a retrying proxy replays, and the one to use when running a model on auray.run, where a request that outlives its connection is answered 504 and recovered by replay. Omit it and the platform mints one. 1 to 64 characters.

#Rules

Across fields, as the model's documentation states them. Each is judged before anything is charged — warned with a Deprecation header until enforcement starts — and listed as JSON Schema under x-auray-rules in the openapi.json.

  • When inputs.meshFile is sent, do not send a request where settings.shapeSlat is sent, or where settings.sparseStructure is sent, or where settings.decimationTarget is sent, or where settings.remesh is sent, or where settings.remeshBand is sent, or where settings.remeshProject is sent, or where settings.meshCluster is sent.
  • Unless settings.dracoCompression.enabled is true, do not send settings.dracoCompression.level. Unless settings.dracoCompression.enabled is true, do not send settings.dracoCompression.quantizationNormal. Unless settings.dracoCompression.enabled is true, do not send settings.dracoCompression.quantizationPosition. Unless settings.dracoCompression.enabled is true, do not send settings.dracoCompression.quantizationTexCoord.

#Example

curl -X POST https://queue.auray.run/auray-serverless/microsoft/trellis-2-4b \
  -H "Authorization: Bearer $AURAY_API_KEY" \
  -H "content-type: application/json" \
  -d '{
  "inputs": {
    "image": "https://v3b.auray.media/examples/samples/still-life.jpg"
  }
}'

The same body on https://auray.run runs synchronously and answers the result — unless it usually takes longer than a run may hold its connection, which is refused 400 model_not_synchronous before anything is charged. Options — ?webhook=, ?wait=, priority and timeout headers — are on Submit a request.

#Machine contracts

  • openapi.json — this page as an OpenAPI 3.1 document, public.
  • llms.txt — the same, written for an agent.

Start typing to search every page of the documentation.

Up arrow Down arrow to moveEnter to openEscape to close