-
How to specify the audio for reasoning?
-
执行训练脚本报错
`python main.py data/obama/ --workspace trial_obama/ -O --iters 100000`
这是我的环境
![image](https://github.com/Fictionarry/ER-NeRF/assets/63634217/fefe386a-6cd6-49c7-9efb-889e34ae0ff8)
…
-
### Need Effect Rework
- [ ] Brother Cancer & Tumor
- [ ] Tumor: act as tiny meatboys dealing very little damage
- [ ] Cyanide
- [ ] Parasites
- [x] Rename to _Infestation 3_
- [ ] Re…
-
I used ave audio encoder to replace ernerf's audio encoder, and unfortunately I got a worse result.
-
本项目中loss_u和loss_static_uncertainty使用了与er-nerf不同的权重,请教下具体原因是什么?
-
Hi, thanks for your project.
When I download the pretrained checkpoints that you provide on the Obama video clip and test by
> python main.py data/obama/ --workspace trial_obama/ -O --test --ckp…
-
```
--------------------------------------------------------------------------------------------------------
| epoch | train_loss | val_loss | train_acc | val_acc | ema_val_acc | total_…
-
1. 我在用hubert来重新训练May数据,得到的人物嘴巴抖动很快,使用Hubert来训练需要更改其他的设置吗?
2. 为什么hubert的self.audio_in_dim = 27,在er-nerf中是1024?
if 'esperanto' in self.opt.asr_model:
self.audio_in_dim = 44
elif …
-
-
你好,我想问一下,对于预处理中的颈部的填充是如何填的呢,我借鉴于ER-NeRF的代码来进行修改
我根据论文的思想,取出对应图片的颈部的RGB颜色,然后填充,但是好像效果不一样,就请问一下,大概如何实现的
![image](https://github.com/ZiqiaoPeng/SyncTalk/assets/61195303/92a87aab-d0a3-406a-8830-21011c…