More control modality

[LaFeuilleMorte]

Hi, Thanks for open-sourcing this great work! Will it add more input modality in the future. I believe image or text to 3D models are very important in 3D generation. Thanks for your reply

[buaacyw]

Thanks! I mostly use Rodin for image or text to 3D. So I can't integrate it into my repo.
For other opensource image or text to 3D methods, their surface quality is too low compared to Rodin.

[claudiomartella]

the papers shows that MeshAnythingV2 supports image to mesh. Is that not the case?

[buaacyw]

the papers shows that MeshAnythingV2 supports image to mesh. Is that not the case?

This is detailed in MeshAnythingV1. We achieve image to mesh by combining with 3D assets production pipeline.

We split any modality to mesh into any modality to shape and then shape to mesh.

Any modality to shape is intensively researched in recent years. We use Rodin to achieve text/image to shape.

[YUBO427]

the papers shows that MeshAnythingV2 supports image to mesh. Is that not the case?

This is detailed in MeshAnythingV1. We achieve image to mesh by combining with 3D assets production pipeline.

We split any modality to mesh into any modality to shape and then shape to mesh.

Any modality to shape is intensively researched in recent years. We use Rodin to achieve text/image to shape.

are you from buaa?