A two-hour lecture recording might contain sixty slides. Capturing them by hand means playing the video, pausing at each transition, taking a screenshot, cropping out the player interface, and saving with a sensible name. Sixty times. Realistically that is an hour of work to extract material that took the lecturer ninety minutes to deliver.

Batch extraction turns the same job into one setup step and a couple of minutes of waiting. This guide covers picking the right interval, why PNG beats JPEG for this specific content, and the quirks of the three platforms most recordings come from.

The workflow

1Download the recording from your course platform. You need the file locally - streaming will not work.
2Open the extractor and select it.
3Set a frame count that samples roughly every 20 to 30 seconds. The table below does the arithmetic.
4Choose PNG. For slide content it is usually both sharper and smaller than JPEG.
5Extract, download the ZIP, and delete the duplicate slides - they sit next to each other when sorted by filename.

Try FrameRipper - free, no upload

Extract frames from any video directly in your browser. No sign-up, no file size limits.

Open FrameRipper

Choosing an interval

Slide-based lectures change on a slow, irregular cadence. Most lecturers spend somewhere between one and three minutes per slide, with occasional long pauses on a diagram and occasional bursts through a series of quick points.

Because the cadence is irregular, you cannot sample perfectly. What you can do is sample densely enough that even the fastest transitions get caught, and accept some duplicates as the cost. Every 20 to 30 seconds does this reliably.

The asymmetry matters here: deleting six copies of the same slide takes five seconds, while discovering you missed slide 34 entirely means running the job again. Always err denser.

Lecture length	Every 20s	Every 30s	Every 60s
50 minutes	150 frames	100 frames	50 frames
90 minutes	270 frames	180 frames	90 frames
2 hours	360 frames	240 frames	120 frames
3 hours	540 frames	360 frames	180 frames

Use PNG, not JPEG

This is the one situation where the usual format advice inverts, and the reason is worth understanding because it applies to all screen content.

JPEG was designed for photographs. Its compression works by discarding detail the eye is bad at noticing in natural images - subtle colour variation, fine texture. It handles smooth gradients and organic shapes well.

Slides are the opposite kind of image: large flat regions of solid colour with hard-edged black text on top. JPEG handles hard edges badly, producing a halo of noise around every letter known as ringing. At normal quality settings this makes small text noticeably harder to read, which defeats the point of capturing the slide.

PNG compression works by finding repeated pixel values, which is exactly what a slide is full of. A 1080p slide with a white background often lands under 100 KB as PNG - smaller than the JPEG equivalent and with perfectly crisp text. Better and smaller simultaneously, which does not happen often.

Platform-specific notes

Zoom

Cloud recordings download as H.264 MP4 and extract without any trouble. The thing to watch is layout: if the recording captured speaker view or gallery view, the slide occupies only part of the frame and you will want to crop afterwards.

If you control the recording settings, choosing to record the shared screen only produces a full-frame slide with no cropping needed.

Microsoft Teams

Teams recordings live in OneDrive or SharePoint depending on how the meeting was set up. You need to download rather than stream - look for the download option in the file view rather than the playback view. The files are H.264 MP4 and work fine.

Panopto

Panopto records multiple streams simultaneously, typically a webcam feed and a screen capture. When downloading, make sure you take the screen capture stream rather than the presenter video, or you will end up with three hundred frames of someone talking.

Panopto downloads are usually MP4 and decode normally. Some institutions disable downloading entirely, in which case there is no workaround that respects the platform’s terms.

Check your institution’s rules

Worth stating plainly rather than burying: lecture recordings are usually covered by a policy, and the policies vary.

Most universities permit downloading and personal use for study, and prohibit sharing or redistribution. Some course materials incorporate third-party licensed content - published figures, film clips, images from journals - where copying is restricted more tightly than the lecture itself. A minority of institutions prohibit downloading altogether.

Extracting slides for your own revision is the normal case and is almost always fine. Uploading them to a note-sharing site is a different act and frequently is not. If you are unsure, the policy is usually findable in your library or IT services documentation.

Turning frames into usable notes

A folder of three hundred PNGs is raw material, not notes. A few things make the gap smaller.

Sort by filename first and delete duplicate runs - because frames were captured in order, identical slides cluster together and you can select whole ranges at once. This usually cuts a 250-frame extraction down to the 60 or so distinct slides.

Frame numbers also map back to timestamps, which is useful. If you sampled every 30 seconds, frame 42 sits at roughly 21 minutes into the recording. When a slide needs the lecturer’s explanation to make sense, that gives you a direct pointer back to the moment rather than a hunt through the timeline.

From there, most note apps accept image imports directly. Obsidian, Notion, OneNote and Apple Notes will all take a batch drop, and because the files are named in sequence they land in the right order.