[Feature Request] "Hands-off" screen recording

[ewancg]

The way ScreenToGif works right now is very intuitive and helpful for fine-tuning of a recorded video. However, I'm used to a more "hands-off" approach, which consists of the following:

• Potential command-line invocation with parameters such as output file, video format, whether to select the region or to use something else (fixed rect or last rect; you can already do most of this)
• If region select is chosen, prompt for the area to record. Once the mouse is released, begin recording
• Stop upon normal execution or invocation of the app again with a parameter such as --stop
• Once the video is finished, automatically save & encode the video.

This workflow is incredibly helpful with hotkeys; I'm able to achieve this in my normal setup with this script and another "shim" script I made to toggle the recording with hotkeys. It's very similar to how a screenshot would work, e.g. Windows' Snip & Sketch invocation with Win + Shift + S; the difference is, since it's a video, you'd press the hotkey twice. Once to start the recording, once to stop. The addition of this behavior might be a bit of an undertaking, but I think it would be worth it. Please let me know your thoughts!

[ewancg]

Here are some videos demonstrating the proposed behavior.

Video of desktop during process (mp4) Video resulting from process (mp4)

[RezhaBlue]

Wanted to ditto this ask. If we can script against an API or use the CLI we could automate this app and implement it into our QA Test Automation workflow in CI/CD. It would be used in place of or in tandem with technologies like ffmpeg for video recording. This would allow us to record the screen as a browser is being interacted with by automation for test purposes in the cloud. We would then be able to store the test run recording as a hopefully compact gif and link to that in our test report which is a generated HTML file. We would then be able to review and quickly figure out what went wrong if a test fails without having to manually reproduce the error or run the test multiple times to "see" what went wrong.