RELIGHT: PIII 1L gastric cancer 🇨🇳, now published in JCO. "Positive", but put it next to CheckMate and it suddenly loses its luster. Retlirafusp is a PD-L1/TGF-B bispecific. The study randomized 731 patients with HER2-negative gastric/GEJ adenocarcinoma to retlirafusp + CAPOX vs placebo + CAPOX. OS 16.8 vs 10.4 mo, HR 0.53 PFS 7.6 vs 5.5 mo, HR 0.52 Looking at the Chinese cohort in CheckMate-649, CPS ≥5 cohort (nivo + chemo vs chemo) results are strikingly similar: OS 15.5 vs 9.6 mo, HR 0.54 PFS 8.5 vs 4.3 mo, HR 0.52 Given a plain PD-1 inhibitor produced essentially the same hazard ratios here it is hard to see what TGF-β is adding. Chemo alone hasn't been standard for years, and the authors acknowledge the protocol was locked before PD-1 + chemo became SoC in China. RELIGHT shows retlirafusp beats chemo; we already knew that. Why am I posting about it? Well, it is another striking example that China has woken up to clinical trials and is running laps around the rest of the world in terms of speed, cost and efficiency. Anyone who has watched a trial sit in contracting for months knows reform is overdue (someone please teach me how to efficiently move legal forward). @OncoAlert @TheGutOncLab @Onco_Nexus

🚀fal just launched H3 Max Relight, an AI video relighting tool powered by MiniMax H3 Max. Upload an existing video, then use Lighting Studio to create a new lighting setup. H3 Max relights the entire video while preserving the original subject, motion, camera movement, and audio. Turn a flat daylight shot into golden hour, neon, or custom brand lighting — without reshooting. At 768p, it costs $0.08 per second. A 5-second video costs $0.40.

LTX-2.5 🌞🌙 IC-LoRA Day-to-Night Relighting It re-renders a daytime video as the same shot at night while preserving composition, framing, camera movement, and subject motion frame-for-frame. 👇

A team from Zhejiang University and vivo has introduced FlashRelight, an AI system that can relight portrait videos. It's more than just applying a filter. The goal is to change the lighting while preserving a person's identity, expression, and motion. FlashRelight takes a two-stage approach: First, a delighting model suppresses the original lighting and shadows. Then, a relighting model generates new dynamic lighting across the video. You can control the lighting in three ways: • Envmap sequences: Precise frame-level lighting control. • Text prompts: Describe color, direction, movement, and flicker. • Reference videos: Transfer lighting effects from another video. The team built a hybrid dataset with 34k+ paired videos: • ~24K synthesized from static OLAT lighting data • 10k+ generated by Wan2.2 This gives the model varied lighting and motion without costly dynamic light-stage capture. On envmap-based relighting, FlashRelight outperformed all baselines across all five evaluation metric For short videos, ads, or virtual-human content, this could make it easier to adjust a portrait’s lighting in post-production.

What a phenomenal flight for Starship!! Payload deployment, orbital relight, baby soft landing of the biggest spacecraft to ever return from space. The culmination of decades of intense work. Welcome to the Starship era 🚀

Thrilled to announce I'll never be going into a podcast studio again. The new Beeble SwitchX 2 model is insanely good at swapping the background and relighting any video 🤯 I've seen it work on much more complex multi-character scenes, will be testing those soon!