Infer

mode

string

enum

NIM inference mode. If image is selected an input image is required. If text if selected an input prompt. If mode is not specified the mode would be determined based on the image and prompt inputs

Allowed:

prompt

string

length ≤ 77

What you wish to see in the output 3D model. A strong, descriptive prompt that clearly defines elements, colors, and subjects will lead to better results.

image

string | null

An image input to generate the 3D model from. Preview API NIM supports only a predefined set of images. The image should be in form of data:image/png;example_id,{example_id} with example_id in a range [0,3].

no_texture

boolean

Defaults to false

If True skips the texture baking stage at the end of model generation pipeline

output_format

string

enum

Defaults to glb

The output format to use. If stl format is selected no_texture parameter would be ignored

Allowed:

samples

integer

1 to 1

Defaults to 1

Number of 3D Objects to generate. Only samples=1 is supported

seed

integer

≥ 0

Defaults to 0

The seed which governs generation. Changing the seed with other inputs fixed results in different outputs. (Omit this option or use 0 for a random seed)

slat_cfg_scale

≤ 10

How strictly the diffusion process adheres to the input text or image in the structured latent diffusion

ss_cfg_scale

≤ 10

How strictly the diffusion process adheres to the input text or image in the sparse structure diffusion

slat_sampling_steps

10 to 50

Number of structured latent diffusion steps to run.

ss_sampling_steps

10 to 50

Number of sparse structure diffusion steps to run.

200Successful Response

422Validation Error