Andrew Ryan Carpenter

Cool. If I am understanding you correctly, these bitstreams can be rendered by any rasterization method that is faithful the to the 3DGS rasterization implementation? Meaning SIBR or something similar...

Oh, very nice! I did not realize that. I'm excited to try this out :)

> I tried to plot it, looks a little better but it seems I am very zoomed out and I see the whole point cloud instead. I am not sure...

> While Colmap provides W2C coordinated, Gsplat anticipates W2C world camera. Perform the conversion from c2w to w2c and retest Sorry, I was not clear about the issue I'm facing...

Sure thing, thank you in advance for your help :) cfg.json: ``` {"disable_viewer": true, "ckpt": null, "data_dir": "../../dataset/indoor/milago_dark3", "data_factor": 1, "result_dir": "../../dataset/indoor/milago_dark3/gsplat/", "test_every": 8, "patch_size": null, "global_scale": 1.0, "port": 8080,...

Do you think this could likely be caused using a left-handed quaternion system where a right-handed system is expected?

Thanks @hariharan1412 and @abrahamezzeddine that context helps a lot!

Each image was 4MB and you had 4000 of them? How many splats were in your scene when it crashed?

I need to dig up the code again and take a look. I know we had this issue fixed. Is this a time sensitive issue for you?

Use something like this: ``` def lazy_call(f, *args, **kwargs): return lambda: f(*args, **kwargs) def cameraList_from_camInfos(cam_infos, resolution_scale, args, lazy_load): camera_list = [] # for id, c in enumerate(cam_infos): # camera_list.append(loadCam(args, id,...