Recalling Inceptionism: Infinite zooming trippy videos with VGG16
giffmana · x · 2026-08-30
The author recalls early Inceptionism techniques: fixing an ImageNet model's output class and backpropagating to optimize the input image. They then describe a university experiment using VGG16 that extended this into an infinite zooming trippy video by iteratively taking the center crop and re-optimizing. The project was streamed on Twitch, switching classes based on chat input, which helped the lab visualize the scope of ImageNet categories.
Related event: DeepMind Members Recall 2015 Inceptionism Experiments(2 posts)→
More from Research
- Non-technical explanation of Gaussian Splatting technology — CSProfKGD · 2026-08-31
- Paper: Deliberate Practice for Robot Skills under a Budget — tomssilver · 2026-08-31
- Interview: We Were Wrong About Our Minds—And AI — MacrinePhD · 2026-08-31
- ECCV 2026: Odoriko Enables Shape-Aware Unified Multimodal Motion Generation — mittu1204 · 2026-08-31
- ECCV Paper: Spectral Alignment Reduces Exposure Bias in Diffusion Models — mittu1204 · 2026-08-31
- ECCV Paper: Step-by-Step Video-to-Audio Synthesis via Negative Guidance — mittu1204 · 2026-08-31