Dan Ring is an expert in empowering VFX artists with cutting-edge research. He began his 20-year career by creating tech to analyze surgery videos (so that no one else would have to), then moved to Foundry to develop CopyCat, the Nuke machine-learning tool that applies corrections to sequences based on a small number of frames.
Now, Dan has joined Chaos to help implement machine learning in its visualization products. In this conversation with Chris, Dan covers just about every aspect of modern CG, including generative AI, Gaussian Splats and NeRFs, the surprising VFX roles that AI might not replace, and the steps Chaos is taking to ensure users have access to helpful but unintrusive tools that are ethically sound.
Links:
0:04:05 Analyzing surgery videos
0:11:32 Joining Foundry and using machine learning in Nuke
0:16:15 How AI tools are trained
0:18:25 What is Copycat?
0:24:18 Why we need good-quality training data with HDR
0:27:03 How generative AI works
0:31:08 CLIP, image captioning, and diffusion
0:36:38 Ray tracing, rasterization, and AI
0:41:38 Light fields and machine learning
0:46:15 The differences between Gaussian Splats and NeRFs
0:50:42 Gaussian Splats and ray tracing
0:54:13 Chaos' Innovation Lab and its initiatives
0:58:23 Keeping up with the rapidly changing world of tech
1:01:58 Are we on an AI hype train?
1:07:05 Improving how AI works
1:12:48 Machine learning for Chaos users
1:16:53 AI won't replace this one VFX job
1:21:48 Will AI be able to direct people?
1:27:33 How Chaos is handling the ethics of datasets
1:33:35 Hallucinations and bias
1:38:35 How Dan is using AI at home
Chris is a CG industry veteran and Director of Chaos Labs. He can also be heard regularly as the host of the CG Garage podcast which attracts 20,000 weekly listeners. With a background in both VFX and Design, Chris has worked for Gensler, Digital Domain, Imageworks and Method Studios. His credits include Maleficent, Oblivion and Tron: Legacy.