GaussianCube

The `example_data/shapenet/shapenet_train.txt` contains the names of all 3D objects used in training. We render the shapenet data using [this script](https://github.com/nv-tlabs/GET3D/tree/master/render_shapenet_data) and save the renderings in `example_data/shapenet/shapenet_rendering_512`.

Hi, we have updated the inference code, now it will automatically save the gaussiancubes for further mesh conversion. Please try it again. Thanks.

As mentioned in our paper, we train the unconditional generation with 1000K steps using 16 V100 GPUs.

Thanks for your interests in our work. For shapenet data rendering, we adopt the [rendering script](https://github.com/nv-tlabs/GET3D/tree/master/render_shapenet_data) from GET3D. For OmniObject3D, we adopt the official [rendering script](https://github.com/omniobject3d/OmniObject3D/tree/main).

Sure, we are going to release the implementation after ECCV conference. Thanks for your interests!

Hi all, I have upload the avatar training code. Basically, you just need to encode and save the DINO feature of each avatar, and in the dataset_avatar.py, we will take...

We use the 3D models of the synthetic face dataset and render the multi-view images for Gaussian cube fitting. However, this part of the data will not be publicly available...

Hi, we are currently training our text-to-3d model on larger datasets, and we will release the weight once ready, please stay tuned. Thanks.

Hi, it is normal to encounter this warning. The training will proceed normally when lg_loss_scale is greater than 0.

Hi @tiangexiang, Thanks for your questions and observations. 1. The start_idx and end_idx values in the script were simply set for demonstration purposes. You can certainly adjust these values to...