Awesome

Overview

A viewer and some tools to work with guassian splatting reconstructions. Currently will open .ply files and gaussian splatting workspaces from the original guassian-splatting implementation. Intended primarily for testing taichi-splatting

Example data

Some example scenes can be found from the official gaussian-splatting page. Under Pre-trained models.

Installing

External dependencies

Create an environment (for example conda with mambaforge) with the following dependencies:

python >= 3.10
pytorch - from either conda Follow instructions https://pytorch.org/.
taichi-nightly pip install --upgrade -i https://pypi.taichi.graphics/simple/ taichi-nightly

Install

One of:

pip install splat-viewer
Clone down with git clone and install with pip install ./splat-viewer

splat-viewer

A gaussian splatting viewer. An example of some visualizations produced by this viewer can be seen:

Arguments

usage: splat-viewer [-h] [--model MODEL] [--device DEVICE] [--debug] model_path

positional arguments:
  model_path       workspace folder containing cameras.json, input.ply and point_cloud folder with .ply models

options:
  -h, --help       show this help message and exit
  --model MODEL    load model from point_clouds folder, default is latest iteration
  --device DEVICE  torch device to use
  --debug          enable taichi kernels in debug mode

Keyboard Controls

Switch View mode

1: normal rendering
2: render gaussian centers as points
3: render depth map

Show/hide

0 : cropped foreground
9 : initial points
8 : camera markers

Misc

prntsc: save high-res snapshot into workspace directory
shift return: toggle fullscreen

Camera

'[' : Prev camera
']' : Next camera
'=' : zoom in
'-' : zoom out
w/s a/d q/e : forward/backward left/right up/down
keypad plus/minus: navigate faster/slower

Animation

space: add current viewpoint to animaiton sequence
control-space: save current animation sequence to workspace folder
return: animate current sequence
shift plus/minus: animation speed faster/slower

splat-benchmark

A benchmark to test forward and backward passes of differentiable renderers. Example splat-benchmark models/garden --sh_degree 1 --image_size 1920

Arguments

usage: splat-benchmark [-h] [--device DEVICE] [--model MODEL] [--profile] [--debug] [-n N] [--tile_size TILE_SIZE] [--backward] [--sh_degree SH_DEGREE] [--no_sort] [--depth]
                       [--image_size RESIZE_IMAGE] [--taichi]
                       model_path

positional arguments:
  model_path            workspace folder containing cameras.json, input.ply and point_cloud folder with .ply models

options:
  -h, --help            show this help message and exit
  --device DEVICE       torch device to use
  --model MODEL         model iteration to load from point_clouds folder
  --profile             enable profiling
  --debug               enable taichi kernels in debug mode
  -n N                  number of iterations to render
  --tile_size TILE_SIZE
                        tile size for rasterizer
  --backward            benchmark backward pass
  --sh_degree SH_DEGREE
                        modify spherical harmonics degree
  --no_sort             disable sorting by scale (sorting makes tilemapping faster)
  --depth               render depth maps
  --image_size RESIZE_IMAGE
                        resize longest edge of camera image sizes
  --taichi              use taichi renderer