a11enL

Expose option to control export of SH

1

When importing multiple PLYs and exporting them to one PLY, data is lost PLY A: 700M PLY B: 30M Exported single PLY by supersplat: 200M Exported single PLY by postshot:...

Is there any suggestions to train large scene_scale dataset with MCMC?

1

gsplat 1.4.0 python3 examples/simple_trainer.py mcmc --use_bilateral_grid --data_factor 1 --data_dir data/test123/ --result_dir exports/test123/ part of logs ----------------------------- ... [Parser] 120 images, taken by 120 cameras. Scene scale: 2009.6237303032315 Model initialized. Number...

How to avoid distortion of human faces when generating human model in ComfyUI?

1

How to avoid distortion of human faces when generating human model in ComfyUI? e.g. increase the face number (max_facenum) ?

gsplat 1.5.3 ubuntu 22.04 cuda 11.8 simple_trainer.py mcmc --eval_steps -1 --disable_viewer --data_factor 1 --save_ply --use_fused_bilagrid --with_ut --with_eval3d ... ../aten/src/ATen/native/cuda/MultinomialKernel.cu:112: binarySearchForMultinomial: block: [158,0,0], thread: [0,0,0] Assertion `cumdist[size - 1] > static_cast(0)`...

For VGGT inference, with batch size and overlapping frames, how to avoid attention dilution and camera head compression in global attention pass when batch process? tokens_per_frame = [t for t...