RFD 1016 details: the inventory table

Every model below is a deep learning model. Each one is packaged as its own model image, per RFD 1036. RFD 1033 lists the geometric algorithms.

Model id Task
trellis2_image_to_textured_mesh Image to 3D
trellis2_image_mesh_painting Image mesh painting
pixal3d_image_to_textured_mesh Image to 3D (PBR)
p3sam_mesh_segmentation Mesh segmentation
krea2_turbo_text_to_image Text to image
qwen_q4_k_m_image_edit Image editing
seethrough_layer_decomposition Image to layers
kimodo_text_to_motion Text to motion
skintokens_auto_rig Auto rig (full)
voxhammer_text_mesh_editing Text mesh editing
voxhammer_image_mesh_editing Image mesh editing
weftspun_image_to_world Image to world
lingbot_map_environment_scan Environment scan
worldmirror2_reconstruct Photos to splat
triposplat_image_to_splat Image to splat

Client and external models

The avatar from photo task runs on AvatarSDK, which is an external cloud service. The client stores the credentials in VITEAVATARSDK* variables. The WebXR expression tracking runs in the browser. Neither one is part of the model catalog.