Demo
MachineLearning in Houdini
Sora, published by OpenAI, seems like a game-changer for the movie industry. The demo video generated by Sora showcases incredibly high-resolution graphics and an abundance of details, which would take a VFX team months to achieve the same results. However, AI video generators is not new to the VFX industry. Even though Sora is currently the best video generator in AI, many AI technologies have already been integrated into the field but have never replaced VFX artists. The reason is AI video generators similar like Sora only generate a whole video, not able to make change specific part, which means generating a totally different video each time. Additionally, AI video generators are not yet fully controllable. The details are not always precise or ideal, and a tons of visual effects are very abstract, making them difficult to describe in language. Even though AI video generater is such badly, trainning and using AI is still really promising to VFX industry. AI in the hands of VFX artists will optimize their work and improve efficiency. AI can also help solve some persistent issues in the VFX industry.
​
BTW, the Houdini 20 version has added a new ONNX node that can connect AI model at any point.
Tutorial Couse From ENTAGMA
ML101
What is ONNX?
Open Neural Network Exchange (ONNX) is an open-source framework that establishes standards for representing machine learning models and software tools.
Example:
Phone brands pair their products with each AI models on a one-to-one basis.
Apple
Samsung
LG
Model 1
Model 2
Model 3
Model 4
Complexity: 12
3 * 4 (N*M)
​
Phone brands pair their products with AI models using the ONNX protocol, ensuring both the brands and models adhere to ONNX standards.
Apple
Samsung
LG
ONNX
Model 1
Model 2
Model 3
Model 4
Complexity: 7
3 + 4 (N+M)
​
Not only for pairing in phone, this technology also helps support AI across all software platform, reducing labor and time costs.