“Just recently we discussed the importance of Multimodal RAG approach on the example of systems that can retrieve image info. (You can read about it here
https://x.com/TheTuringPost/status/1878932305177453037
“FastProtect makes AI art protection instant by pre-training defense patterns instead of calculating them live. FastProtect enables real-time protection against AI mimicry by using pre-trained perturbations and adaptive inference, making image protection 200-3500x faster than
https://x.com/rohanpaul_ai/status/1878002104633319681
“A diffusion model that lights up real portraits with studio-level shadows, highlights, and identity fidelity. Paper: “SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Faces” → Proposes SynthLight, a diffusion model for portrait relighting
https://x.com/rohanpaul_ai/status/1880176475682615711
“GAN is back! A “The GAN Is Dead; Long Live the GAN!” paper from @Cornell and @BrownUniversity reignited interest in GANs (Generative Adversarial Network) with a simpler, improved approach that can beat diffusion models. The core innovations: • Relativistic GAN loss – A better
https://x.com/TheTuringPost/status/1879111514210402681
“GANs are so back?! Scientists from Brown and Cornell have published a paper with a ✨ modern architecture GAN ✨ that is 🗿 stable to train 🗿 and competitive with SOTA GANs and even diffusion models Paper and demo 👇
https://x.com/multimodalart/status/1877724335474987040
Decentralized Diffusion
https://decentralizeddiffusion.github.io/
“If you want to train your own diffusion model on real images without selling you house, I recommend Micro Diffusion by Sony Research that allows training your own diffusion model on a tight budget (< 2k$). Link in🧵” / X
https://x.com/hkproj/status/1879603337206919365
“When I first saw diffusion models, I was blown away by how naturally they scale during inference: you train them with fixed flops, but during test time, you can ramp it up by like 1,000x. This was way before it became a big deal with o1. But honestly, the scaling isn’t that” / X
https://x.com/sainingxie/status/1880106419573387528
“Visual editing through code generation helps AI focus on what matters in complex charts and tables. Like using a marker to highlight important parts, AI edits images to understand better. ReFocus enables LLMs to perform visual editing on structured images like tables and
https://x.com/rohanpaul_ai/status/1880178970634936707
