pointnerf icon indicating copy to clipboard operation
pointnerf copied to clipboard

The process of training my own data

Open LeeHW-THU opened this issue 2 years ago • 9 comments

Thank you for your work and contribution. I saw that you uploaded the code for COLMAP and it seems to only need the colmap_results/dense/fused.ply file. I'm having trouble making my own dataset, I have a batch of images now, how do I make my transforms_{}.json. the camera_angle_x and rotation in it I don't know where to get from, the transform_matrix seems to be available from COLMAP. Do you have a script to generate these files please? I'm new at this.

LeeHW-THU avatar Mar 18 '22 10:03 LeeHW-THU

I also have the question about how to covert llff format dataset to the dataset as in the nerf_synthetic or nerf_synthetic_colmap. Do you have any hints about this?

Heng14 avatar Mar 22 '22 19:03 Heng14

It seems that set dataset_name='llff_ft' in the config sh file will lead to xyz_world_all zeros shape. Does the config differ from what is in the dev_scripts/w_n360 ?

Heng14 avatar Mar 22 '22 20:03 Heng14

hi, the liff_ft is not well developed, so there might be some error. we don't have time to cover liff datasset in our submission.

Xharlie avatar Mar 22 '22 22:03 Xharlie

Thank you so much for your reply! Do you have the plan to cover llff_ft recently?

Heng14 avatar Mar 22 '22 22:03 Heng14

I think the issue may be related to the 'ranges' in the config. Could you please give me some hints on how to decide the ' ranges' in the config? Thank you so much!

Heng14 avatar Mar 22 '22 23:03 Heng14

I think the issue may be related to the 'ranges' in the config. Could you please give me some hints on how to decide the ' ranges' in the config? Thank you so much!

I think the range in the configuration file refers to x_min y_min z_min x_max y_max z_max

LeeHW-THU avatar Mar 23 '22 09:03 LeeHW-THU

hi Lee,i wonder if you solve this problem? for i also have trouble adding point cloud from external method, do you have any idea? thanks a lot.

Erisnoit avatar May 09 '22 04:05 Erisnoit

hi @Erisnoit . I did make a batch of data, but I got some Loss errors when I entered the data into PointNeRF. I gave up on continuing the experiment because of the poorly commented variables in the code.

LeeHW-THU avatar May 16 '22 20:05 LeeHW-THU

I think the issue may be related to the 'ranges' in the config. Could you please give me some hints on how to decide the ' ranges' in the config? Thank you so much!

I noticed the same thing and realized the same thing as what Lee mentioned x_min y_min z_min x_max y_max z_max.

But I have no idea about how to set these values for my own data. Could you provide some instructions? Thanks! @Xharlie

AlbertHuyb avatar Aug 30 '22 11:08 AlbertHuyb