jzyu

> I have known the problem and how to fix/bypass it, just rerun, or adjust the parameters. (yes, adjust the parameters to avoid) I do not focus on to modify...

> Hello, I've spent many hours to get this great work running on my PC (R5 2600, 16GB ram, rtx 3070). > > When using cuda 12.1 with Visual Studio...

> Thanks! I found it! ! I would also like to ask if I want to modify the content in "submodules\diff-gaussian-rasterization\cuda_rasterizer\forward.cu", where should I go to modify it (because I...

你好，我也在看复现精度的问题，你说的width32中的batch size指的是batch_size_per_gpu吗？源码里面yml文件中不是写的8吗？32怎么来的。主要是我用width64的那个yml文件，把gpu数量以及num_worker_per_gpu和batch_size_per_gpu都改成了3，因为我的服务器最多只能3卡训练，迭代40万次后发现psnr跟论文中的不一样。不知道是什么原因，难道必须要8卡训练才能达到论文中的效果吗

> Hello, I have a question for you, why I don't find the loss function used in train.py? Thank you very much! Me to. Have you find

The config file I use is tx_blurfactory_full.txt

The command is：python3 run_nerf.py --config configs/blurfactory/tx_blurfactory_full.txt

> You may want to check on this link ([#6 (comment)](https://github.com/limacv/Deblur-NeRF/issues/6#issue-1301430857)), which is caused by me mistakenly leaving one config on when doing the evaluation. But I'm not pretty sure...

哦哦好的，非常感谢！我还有一个问题就是训练数据集中的poses_bounds.npy文件是由sharp image计算出来的还是blur image计算的呢？ ------------------ 原始邮件 ------------------ 发件人: "limacv/Deblur-NeRF" ***@***.***>; 发送时间: 2023年4月13日(星期四) 晚上9:51 ***@***.***>; 抄送: " ***@***.******@***.***>; 主题: Re: [limacv/Deblur-NeRF] 急切请求真实物体运动和相机运动的清晰数据集 (Issue #22) 对于真实场景中的blur，因为没办法保证在同一视角下同时拍到模糊和清晰对图像，所以我们只提供了在novel view下的清晰图像。这些清晰图像已经包含在数据集当中啦。 — Reply to this email directly, view it on GitHub, or...

> @ksupasate Sure! This is the zero123 link (https://huggingface.co/bennyguo/zero123-xl-diffusers/tree/main) for your convenience. Oh！thank you very much！