Orwlit

Results 3 issues of Orwlit

First off, huge thanks for all the hard work on this repo! It's been super helpful, and I'm really grateful for what you've put together here. I'm diving into recording...

First off, huge thanks for all the hard work on this repo! It's been super helpful, and I'm really grateful for what you've put together here. I'm diving into recording...

**感谢大家维护这个仓库!** # 问题现象: 我用Realsense D435i录制了一个室内数据集,但mesh出的效果稀烂,因为第一帧建图得到的结果不好,因此后续位姿估计和建图直接丢失,loss都是nan。 **这是第一帧mesh出的场景:** ![2a2c6d39572775f20f145bef1bc35c5](https://github.com/electech6/NeRF-Based-SLAM-Incredible-Insights/assets/86476301/c27dccd5-a071-4dd7-8e75-dfd433d0ab86) **这是它本应该有的样子:** ![微信图片_20231121010820](https://github.com/electech6/NeRF-Based-SLAM-Incredible-Insights/assets/86476301/0492f527-e67d-405a-bd22-ccfbcbc03a74) # 目前我的情况是: 1. 代码中我的`yaml`文件是从`replica.yaml`里改过来的,`dataset.py`中读数据的类也是从`RealsenseDataset`类中改过来的 2. 目前我确信进入代码的深度图没有问题,即每个像素的值代表真是距离(单位:米)。 3. 录制的是16位深度图,像素值和真实深度的缩放比例是0.001,尺寸是`640 * 480`。 **yaml文件中所有参数如下,除了cam以外基本没改:** ```yaml dataset: 'my_realsense' data: downsample: 1 sc_factor: 1 translation: 0...