This repository contains the training data for the paper Feedforward 3D Editing via Text-Steerable Image-to-3D.
Project Page:
https://glab-caltech.github.io/steer3d/
Code:
https://github.com/ziqi-ma/Steer3D
The metadata directory contains training metadata used for texture, removal, and addition. data/images contains the input images, identified by a unique uuid for each object. We release TRELLIS latents of all source and target assets in data/latents… See the full description on the dataset page:
https://huggingface.co/datasets/ziqima/Steer3D-Data.