How to do Environment Scattering in Houdini with Machine Learning?

Sdílet
Vložit
  • čas přidán 5. 06. 2024
  • Master 3D Forest Scattering with Machine Learning & Real Data
    The usage of real forest data for scatter environments can revolutionize the way environment artist generate their scatters, offering unprecedented realism and depth in digital environments. By leveraging real forest data, we can now populate vast digital landscapes with 3D scatterings that mirror the complexity and diversity of natural forests. Machine learning accelerates this process, allowing for the rapid generation of these environments with stunning accuracy and detail.
    This method can quickly generate diverse forest ecosystems that accurately reflect real-world complexity and variability. Once the model is trained, environments can be created much faster than traditional procedural methods, potentially reducing production times. And it's making the process more accessible to artists without extensive technical backgrounds in procedural generation.
    Dive into the world of creating hyper-realistic 3D forests using machine learning. This tutorial guides you through the process of using real forest data to train a model that brings your digital environments to life. Perfect for visual effects artists, game developers, and machine learning enthusiasts, I cover everything from data preparation to model training and integration with Houdini.
    0:00 - 0:59 Introduction
    0:59 - 2:32 Real Forest Data Retrieval
    2:32 - 5:47 Fit the machine learning model
    5:47 - 7:25 Heighfield generation
    7:25 - 11:34 Determine what tree type to scatter in each location
    11:34 - 13:20 Reviewing model tree scattering results
    13:20 - 14:11 Comparing with a random classifier
    What You'll Learn:
    Basics of machine learning for VFX
    Preparing and using real forest data
    Training your machine learning model
    Generating realistic 3D forests in houdini,popular software
    Why Watch?
    Elevate your VFX skills
    Add depth and realism to game environments
    Practical, step-by-step instructions
    -
    Thanks for watching!
    Join my newsletter oriolmanzano.com/newsletter
    Code and houdini file can be found in
    github.com/mrnsapple/DailyTip...
  • Krátké a kreslené filmy

Komentáře • 13

  • @mostafasamir9472
    @mostafasamir9472 Před 23 dny

    Amazing

  • @Cheezburges
    @Cheezburges Před 2 měsíci

    Lovely work, cant wait to learn more!

  • @nopah
    @nopah Před 2 měsíci

    Great video. I didn't expect another upload 😅😊

  • @xrlearn
    @xrlearn Před 2 měsíci

    Love this. More content like this. I wish this was detailed and longer. I still wasn't able to understand how you were training the model and bringing it to Houdini. do you have any resources for Houdini + AI materials where you are using real trained data to learn about?

    • @oriolmanzanoduran
      @oriolmanzanoduran  Před 2 měsíci +2

      Thank you so much for your feedback and enthusiasm! Understanding the intricacies of training machine learning models and integrating them with Houdini can indeed be complex. To help you understand what is happening take a look at the houdini file and code, I've shared it free in github.com/mrnsapple/DailyTips/tree/main/projects/TreeScattering/src. And to learn more about Houdini + Ai subscribe to my newsletter oriolmanzano.com/newsletter, I publish every week about AI and machine learning in visual effects.

  • @vincentthomas8492
    @vincentthomas8492 Před 2 měsíci

    Now, you have this, could the model predic the type of Soil in Houdini according to the neighbor tree around?

    • @oriolmanzanoduran
      @oriolmanzanoduran  Před 2 měsíci

      Good question! what I am performing here is a supervised learning where the target variable is the tree type("Cover_Type") based in other parameters such as the terrain elevation and slope or the presence of water. So the trained algorithm is not going to give you soil information, just what type of tree has more probability to appear in the location.
      To make the algorithm tell you as well what soil to use, you can just simply include the soils types ("Soil_Type*") together with the tree_type("Cover_Type") in the dataframe of target variables to predict.
      Hope this helps and answers the question!!!!

  • @kairos6761
    @kairos6761 Před 2 měsíci

    tu hiciste el curso de houdini en megacursos ?
    Lo recomiendas para alguien nuevo en los VFX?

    • @oriolmanzanoduran
      @oriolmanzanoduran  Před 2 měsíci

      Para empezar lo mejor es aprender de cursos que estén al día con las últimas versiones de Houdini y que enseñen a hacer efectos complejos que puedas poner en tu reel. No recomiendo que compres ese curso porque la versión de houdini es muy antigua y los efectos muy básicos. En www.sidefx.com/tutorials/ puedes encontrar muy buenos tutoriales de todos los niveles

    • @kairos6761
      @kairos6761 Před 2 měsíci

      @@oriolmanzanoduran gracias por la pagina y por la sinceridad, habrá alguna pagina que enseñe pero en español ?

    • @oriolmanzanoduran
      @oriolmanzanoduran  Před 2 měsíci

      ​@@kairos6761 Yo no conozco, pero si veo alguno te hago saber! Siempre puedes ver un video en inglés con subtítulos traducidos!

    • @kairos6761
      @kairos6761 Před 2 měsíci

      Muchas gracias
      @@oriolmanzanoduran