Fresh off the heels of their Viewer 2.0 update, Reflct has taken another step forward in supporting the 3DGS creator community. Today, they’re announcing the release of Sharp Frames Python, an open-source version of their popular frame extraction tool designed to make sharp frame selection even faster, more flexible, and accessible across platforms.
If you’ve worked with Radiance Field methods like Gaussian Splatting, you know that sharp input frames can make or break a reconstruction. Sharp Frames has been a go to for streamlining that part of the pipeline, but until now, it was tied to a Windows application. The new Python version expands your options.
Sharp Frames Python is built with the same core extraction and selection methods as the Sharp Frames app, but now it’s faster, more portable, and more customizable. It runs on Windows, macOS, and Linux, finally giving Mac and Linux users an easy way to incorporate sharp frame extraction into their workflows.
The sharpness calculation step has been significantly optimized, making it even faster than the current Sharp Frames app (those improvements are also slated to be backported to the app soon). Users can run the tool in interactive mode simply by executing python sharp_frames.py, stepping through inputs and options without needing to memorize command-line arguments. Batch processing, best-n selection, and outlier-removal methods are all included.
Other highlights include real-time progress bars across major steps (dependencies, extraction, scoring, selection, and saving), robust parallel processing using multiple CPU cores, and the ability to safely cancel operations mid-process. The tool is memory efficient and handles large sets of frames with ease, making it ideal for high-resolution video work or large image directories.
When it finishes, Sharp Frames Python saves your selected frames neatly into an output directory and generates a metadata file (selected_metadata.json) that details the input, parameters used, and sharpness scores of the selected frames—perfect for keeping track of your dataset quality.
Released under an MIT license, this open-source project empowers more creators and also invites contributions and improvements from the broader community.
Download and explore the repo.
Learn more about the full Sharp Frames app: https://sharp-frames.reflct.app
Join the Reflct Discord to access the beta Windows version: https://discord.gg/rfYNxSw3yx
