Free Houdini Tutorial: Volume Style Transfer With ONNX

Поделиться
HTML-код
  • Опубликовано: 27 май 2024
  • Support us on www.patreon.com/entagma
    In an earlier episode we took a first look at the new ONNX inference node, a node that allows us to use trained neural nets inside Houdini. And today we want to use this node for something usefull - Volume Style Transfer! We want to apply a pretrained style to voxels instead of pixels and give our simulations a unique look and much more detail. And we're going to do all of this using just a tiny bit of machine learing and a lot of SOPs!
    Watch our previous ONNX episode here:
    • Free Houdini Tutorial:...
    Download the scenefile here:
    entagma.com/downloads/Free_ON...
    Chapters:
    00:00 Intro
    02:18 A quick and dirty Pyro setup
    04:57 Creating volume slices
    09:38 Writing voxel data to slices
    11:22 Deleting empty slices
    12:43 Style transfer
    16:48 Assembling a volume from slices
    19:00 Mixing with original volume
  • КиноКино

Комментарии • 4

  • @felixblacke
    @felixblacke 15 дней назад

    Very cool and unexpected integration. Would love to see more ways like this that these tools can be implemented beyond wholesale image creation/slapping a coat of paint on an image.

  • @vfxskull6208
    @vfxskull6208 15 дней назад +1

    this is fantastic!

  • @PawelGrzelak
    @PawelGrzelak 15 дней назад

    Interesting idea Christopher

  • @rustytoe178
    @rustytoe178 14 дней назад

    cool!