The team at Reflct has rolled out one of the biggest updates yet to Sharp Frames, the company’s free frame selection tool used throughout the radiance field and 3D reconstruction community. The update focuses squarely on speed, memory efficiency, and usability.
They have switched to Mediabunny for frame extraction, delivering roughly 4–5× faster performance when parsing through videos. But the gains don’t stop there. Sharpness calculation itself now runs 2–3× faster, drastically reducing processing times for users working with lengthy or high-resolution captures.
Perhaps the most noticeable change, however, is how much better Sharp Frames now handles massive datasets. Earlier versions struggled under the weight of large image directories. Those memory issues are now fully solved, enabling smooth operation even when thousands of frames are loaded simultaneously.
The visualization layer has been refactored as well. The old Recharts based system has been replaced with a Canvas powered chart, offering far greater memory efficiency and responsiveness when plotting data from large sets.
And for those navigating sprawling projects, there’s a mini map that makes it easy to scroll, zoom, and locate frames within extensive timelines. It had been roughly eight months since the last proper update, though it appears the wait was worth it! Check out Sharp Frames here.
New to Gaussian Splatting? Start here
