Improved COLMAP process

README.md

@@ -19,11 +19,40 @@ NOTE: can no longer use `sh`
 
 See `opt/render_imgs.py`
 
+(in opt/)
 `python render_imgs.py <CHECKPOINT.npz> <data_dir>`
 
-## Automatic hypertuning
+## Parallel task executor
 
+Including evaluation, ablations, and hypertuning (based on the task_manager one from PlenOctrees)
 See `opt/autotune.py`. Configs in `opt/tasks/*.json`
 
 Automatic eval:
 `python autotune.py -g '<space delimited GPU ids>' tasks/eval.json`. Configs in `opt/tasks/*.json`
+
+## Using a custom image set
+
+First make sure you have colmap installed. Then
+
+(in opt/)
+`bash scripts/proc_colmap.sh <img_dir>`
+
+Where `<img_dir>` should be a directory directly containing png/jpg images from a 
+normal perspective camera.
+For custom datasets we adopt a data format similar to that in NSVF
+<https://github.com/facebookresearch/NSVF>
+
+You should be able to use this dataset directly afterwards. The format will be auto-detected.
+
+To view the data use:
+`python scripts/view_data.py <img_dir>`
+
+This should launch a server at localhost:8889
+
+## Random tip: how to make pip install faster
+
+You may notice that this CUDA extension takes forever to install.
+A suggestion is using ninja. On Ubuntu,
+install it with `sudo apt install ninja-build`.
+Then set the environment variable `MAX_JOBS` to the number of CPUS to use in parallel (e.g. 12) in your shell startup script.
+This will enable parallel compilation and significantly improve iteration speed.