Ai Driven Film Emulation

Recorded Friday, October 10, 2025 In Part 1 of the Color Lovestream Film Emulation Special, Dado introduces a groundbreaking new approach to film emulation based on AI-driven image analysis and frequency-domain measurement. Presenting research behind the forthcoming white paper AI-Driven Film Emulation, he explains how Colourlab’s R&D team developed Filmlab Ai using super pixels, vector databases, and multimodal language models to quantify and recreate film’s spatial, chromatic, and temporal characteristics. The session dives deep into measuring modulation transfer, halation, tone mapping, grain behavior, and film’s cross-channel coherence — ultimately demonstrating how Filmlab Ai bridges the perceptual and scientific gap between digital capture and photochemical film. 00:00:00 Intro 00:02:24 Research Overview 00:03:44 Digital vs Film Comparison 00:0509 Current RGB Pixel-Based Analysis Omissions 00:08:22 Macro Level Comparison 00:10:06 Beyond Pixels: Higher Level Feature Extraction 00:10:51 Superpixel: Concept and Extraction 00:13:38 Ai Driven Film Emulation Design Methodology 00:13:48 Creation of Test Material 00:15:10 Sharpness / Optics Block 00:20:05 Tone Mapping Block 00:23:01 Grain / Texture Block 00:26:45 Chroma-Luma Cross-Correlation 00:27:29 ΔE00 Skin-Cluster Metrics 00:29:33 Vector Layout (canonical 64D) 00:30:39 The Dimension of Time 00:31:21 3DFFT Spatio-Temporal Frequency Block 00:33:58 Derived Temporal Features (4D) 00:34:30 Compact 24D Vector 00:35:06 Results & Analysis 00:38:21 Advantages Over Traditional Methods 00:41:32 Limitations 00:45:06 Future Work 00:46:56 Summary 00:48:15 Q&A 00:52:31 Closing Join Color Lovestream: https://lu.ma/colourlab Explore Filmlab Ai: https://colourlab.ai/filmlabai