Skip to content

Texture loader, dds to png converter, upscaler, pbr generator, and replacements writer for rtx remix. Enhancing he's work

Notifications You must be signed in to change notification settings

v-progs/OctoTex

 
 

Repository files navigation

logo

OctoTex

This tool is heavily experimental!

Description

This tool allows you to load captured textures from rtx remix, convert them to png, upscale with ESRGAN or RealESRGAN, generate octahedral normals, roughness, metalness maps, and write it back to the existing rtx remix mod, or create a new one ( recomended ).

AI PBR Models

You can download old PBR ai models from Alex's Gdrive and from mine Gdrive

Put the old models to the

  1. OctoTex/ai/PBR/checkpoints/disp
  2. OctoTex/ai/PBR/checkpoints/norm Folders!

And New Roughness Model on Gdrive

Put the new model to the

  1. OctoTex/ai/PBR/checkpoints/rough Folder!

New Steps

  1. python webui.py // That's it!

Outdated Steps

  1. python load.py ( then select from what folder you want to load textures ) // Textures will be in the textures/processing/diffuse folder
  2. python upscale.py ( you will need an Nvidia GPU, pytorch with cuda support installed ) // Textures will be in the textures/processing/diffuse folder
  3. python pbr.py ( this will generate all the pbr textures to the their folders ) // Textures will be in the textures/processing/normals ** roughness ** metallness folders
  4. python write.py ( this will write all the changes back, it will promt to what mod you want to write it, IT'S RECOMENDED TO CREATE A NEW MOD! ) And that's it!

TO DO

  1. Train models
  2. Make upscaler

Good luck! :)

By Alex:

This project is possible because my boss gave me an RTX gpu, and allowed me to work on this project in office some time, so, i'll be glad if you check his website, thanks! https://fst.kz/

About

Texture loader, dds to png converter, upscaler, pbr generator, and replacements writer for rtx remix. Enhancing he's work

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • JavaScript 85.3%
  • Python 13.5%
  • Other 1.2%