3D Amodal Reconstruction with Amodal3R

Step 1 - Generate Visibility and Occlusion Mask.

  • Please click "Load Example Image" when using the provided example images (bottom).
  • Please wait for a few seconds after uploading the image. Segment Anything is getting ready.
  • Click to add the point prompts to indicate the target object (multiple points supported) and occluders (one point for an occluder for better usability).
  • "Add mask", current mask will be saved if the input needs to be added sequentially.
  • The scale of target object can be adjusted for better reconstruction, we suggest 0.4 to 0.7 for most cases.

Please wait a few seconds after uploading the image.

Point Prompt Type
Dropdown
0.3 1

Step 2 - 3D Amodal Reconstruction. (Thanks to TRELLIS for the 3D rendering component!)

  • Different random seeds can be tried in "Generation Settings", if you think the results are not ideal.
  • The boundary of the segmentation may not be accurate, so here we provide the option to erode the visible area (try 0, 3 or 5).
  • If the reconstructed 3D asset is satisfactory, interactive GLB file can be extracted (may look dull due to the absence of light source) and downloaded.
0 2147483647
0 5

Stage 1: Sparse Structure Generation

0 10
1 50

Stage 2: Structured Latent Generation

0 10
1 50
0.9 0.98
512 2048
Examples