😍 Implement "smart cut"
We cannot cut between keyframes, it's a technical limitation of modern video codecs. So ffmpeg will choose the nearest keyframe, which will not be the exact time you requested. Might be able to use [avcut](https://github.com/anyc/avcut) for h264 although it is experimental. See https://github.com/mifi/lossless-cut/issues/372#issuecomment-761844667 See also discussion in #13 **Update**: Have made an initial implementation of this. This experimental feature will re-encode the part of the video from the cutpoint until the next keyframe in order to attempt to make a 100% accurate cut while losslessly copying the rest of the segment. ## VideoRedo VideoRedo is a now-defunct software that had feature overlap with LosslessCut. - https://www.videohelp.com/software/VideoReDo - https://forum.videohelp.com/threads/407142-VideoReDo-owner-passed-away-product-ceased-seeking-alternatives#post2668738 ## [Smart Media Cutter](https://smartmediacutter.com/) [Recently built](https://www.reddit.com/r/videoredo/comments/1f71j0s/im_trying_to_build_a_modern_version_of_videoredo/) in order to replace VideoRedo They open sourced their smart cutting algorithm, could take inspiration from it: https://github.com/skeskinen/smartcut ## Remaining issues ### Inaccurate seeking Seeking is not always accurate, causing glitches and/or jumps in the stitch points. This is often because I haven't found a way to consistently seek to and cut on an exact frame. Sometimes it could be caused by the "audio sa
Posted by mifi on Feb 10, 2019, 2:20 PM
Tags
Related issues
- 2026 - Microsoft Office 365
- [Ubuntu] `ubuntu-latest` label will use Ubuntu 26.04 in November 2026
- Gamut mapping
- [Codex Desktop] Editing and resending a queued message triggers "App-server queued follow-up no longer exists"
- [Windows] ChatGPT 26.820.60940 fails with "invalid transport in mcp_servers.codex_app"; Beta 26.727.40816 works
- [BUG]: infinite loading of videos
Raise real-world problems, find people facing the same issue, and discuss solutions.4,000 issues0 comments
github (SaaS)
2026 - Microsoft Office 365
[Ubuntu] `ubuntu-latest` label will use Ubuntu 26.04 in November 2026
Gamut mapping
[Codex Desktop] Editing and resending a queued message triggers "App-server queued follow-up no longer exists"
[Windows] ChatGPT 26.820.60940 fails with "invalid transport in mcp_servers.codex_app"; Beta 26.727.40816 works
[BUG]: infinite loading of videos
Restore GPT-5.6 Sol’s 372k Codex context window, or provide an opt-in setting
Codex Desktop for macOS repeatedly triggers `syspolicyd` / `trustd` CPU and memory runaway
😍 Implement "smart cut"
We cannot cut between keyframes, it's a technical limitation of modern video codecs. So ffmpeg will choose the nearest keyframe, which will not be the exact time you requested. Might be able to use [avcut](https://github.com/anyc/avcut) for h264 although it is experimental. See https://github.com/mifi/lossless-cut/issues/372#issuecomment-761844667 See also discussion in #13 **Update**: Have made an initial implementation of this. This experimental feature will re-encode the part of the video from the cutpoint until the next keyframe in order to attempt to make a 100% accurate cut while losslessly copying the rest of the segment. ## VideoRedo VideoRedo is a now-defunct software that had feature overlap with LosslessCut. - https://www.videohelp.com/software/VideoReDo - https://forum.videohelp.com/threads/407142-VideoReDo-owner-passed-away-product-ceased-seeking-alternatives#post2668738 ## [Smart Media Cutter](https://smartmediacutter.com/) [Recently built](https://www.reddit.com/r/videoredo/comments/1f71j0s/im_trying_to_build_a_modern_version_of_videoredo/) in order to replace VideoRedo They open sourced their smart cutting algorithm, could take inspiration from it: https://github.com/skeskinen/smartcut ## Remaining issues ### Inaccurate seeking Seeking is not always accurate, causing glitches and/or jumps in the stitch points. This is often because I haven't found a way to consistently seek to and cut on an exact frame. Sometimes it could be caused by the "audio sa
Comments (0)
No comments yet. Start the discussion.