Makes creating a Nerf, simple and easy, based on Instant-ngp

• Requires Python To Be Installed 3.8 And Above Along With The Latest Cuda Driver

• NVIDIA GPU REQUIRED TO RUN AND USE INSTANT-NGP

• CMAKE GIT MSVC CUDA-TOOLKIT CONDA MUST BE INSTALLED IF YOU WOULD LIKE TO BUILD FROM SCRATCH

Things to know

• any paths with a space will cause errors which include:

  • the location where you cloned/downloaded this repo

  • any paths that you input in the config.txt - config.txt will appear after running the install bat

• any slashes at the end of any inputted path inside config.txt will cause errors

• any Y/N questions can only be responded in capitals, Y or N

• if you don't have anaconda or miniconda installed and come across " - use conda (Y/N): " simply reply N

• dir/path/folder may be used interchangeably in this repo

• any numerical questions can only be responded to in numbers or it can create unwanted errors

• when running the install_rtx_3000_and_4000.bat or any Install bats you agree to the Licence that comes with this repo

• build instant ngp yourself using included bats if any problems occur refer to Compilation for Windows

How to build yourself

1. run initbuild.bat

2. (OPTIONAL) If you want dlss support download vulkan sdk then run the installer and when it comes to configuring simply press the select all button then continue until installed

3. run cmakebuild.bat make sure you run it from the Installs folder and using an activated conda environtment, if it fails, go inside instant-ngp under the Installs folder and delete the build folder and try again, if that doesn't work refer to Compile errors

How to use - Video to Images To Nerf

1. download this repo/clone it in any folder of your choice - make sure the path that you choose has no spaces

2. run the install_rtx_3000_and_4000.bat or build your self

3. open the config.txt file and paste the full path to any empty folder right after the "=", it should look something like this "ProjectDir=C:\project" or alternativly drag and drop the project folder on to DragAndDropProjectDir.bat
   note: the project folder is where any saved data will go including images, videos that you wish to convert into nerfs and generated colmap data

4. place any video that you want to convert into a nerf inside the project folder, then run the videotoimages.bat in QuickRuns
   it will ask " - Name of video inside Where ever you decided the project path to be: " and respond with the video file name - for example, VIDEO.MP4
   it will then ask, " - Frames per second ( 5 recommended ): " - high fps will produce better nerf models but will negatively impact performance

5. run imagestonerf.bat in QuickRuns the speed of this process may depend on what hardware you are running

6. (OPTIONAL) running openproject.bat in QuickRuns will quickly open instant-ngp

Command line shortcuts

Install bats

Command	Use
-h	Prints help page - not implemented
--conda env-name	Activates the conda environment you chose, passing 0 will not activate conda but will avoid it asking if you would like to use conda
--colmapforcuda	Installs the cuda version of colmap for supported devices

videotoimages.bat

Command	Use
-h	Prints help page - not implemented
--video vid-name	The target video to convert to images inside the project dir
--fps fps	How many frames you will extract for every second of the video

imagestonerf.bat

Command	Use
-h	Prints help page - not implemented
--conda env-name	Activates the conda environment you chose, passing 0 will not activate conda but will avoid it asking if you would like to use conda
--colmaprun	Runs colmap in order to convert images into a nerf
--highdetail	Changes camera position accuracy to the highest level