ObjV1

README.md

@@ -12,7 +12,8 @@ vast 3D data corpus of [Objaverse-1.0](https://objaverse.allenai.org/objaverse-1
 3. (To be uploaded) Rendered 4D images by our team to save you GPU time.
 
 ## 4D Dataset ID
-介绍下下总体统计情况，id的链接,text caption  @hanwen
+We curated a total of 54K dynamic 3D assets from Objaverse-1.0 and Objaverse-xl. We provide the selected ids from Objaverse-1.0 in `rendering/src/ObjV1_curated.txt`. More about the dataset and curation scrips coming soon!
+For text-to-4D generation, the captions are obtained from the work [Cap3D](https://huggingface.co/datasets/tiange/Cap3D). 
 
 ## 4D Dataset Rendering Script
 1. Clone the repository and enter the rendering directory:
@@ -34,7 +35,7 @@ rm blender-3.2.2-linux-x64.tar.xz
 
 ```bash
 pip install objaverse
-python download.py --id_path curated.txt
+python download.py --id_path src/sample.txt
 ```
 
 Right-click the objaverse package in  `download.py` and open the `objaverse/__init__.py` file. Change `BASE_PATH='./obj_data'` at line 14 or another path you prefer. Otherwise, it will download to the default system path.
@@ -107,7 +108,7 @@ This project is based on numerous outstanding research efforts and open-source c
 - https://github.com/graphdeco-inria/diff-gaussian-rasterization
 - https://objaverse.allenai.org/objaverse-1.0/
 - https://github.com/allenai/objaverse-xl
-
+- https://github.com/crockwell/Cap3D
 
 ## If our 4D dataset and rendering script help your work, please consider citing our paper and starring the repository ⭐
 ```

rendering/download.py

@@ -1,17 +1,20 @@
+import argparse
 import objaverse
+import multiprocessing
 
-import argparse
 parser = argparse.ArgumentParser()
-parser.add_argument("--id_path", type=str, default="test.txt")
+parser.add_argument("--id_path", type=str, default="src/sample.txt")
 args = parser.parse_args()
-multiprocessing_cpu_count=32
+multiprocessing_cpu_count=multiprocessing.cpu_count()
 
 with open(args.id_path, 'r', encoding='utf-8') as file:
     lines = file.readlines()
 object_uids = [line.strip() for line in lines]
-#print(object_uids)
-print('file number:',len(object_uids))
+# print(object_uids)
+print('file number:', len(object_uids))
 
+objaverse.BASE_PATH = '/home/projects/Diffusion4D/rendering/' 
+objaverse._VERSIONED_PATH = 'objaverse_v1/'
 objaverse.load_objects(
     uids=object_uids,
     download_processes=multiprocessing_cpu_count