Guide

How to give a video to ChatGPT

You recorded something (a lecture, a tutorial, a meeting, your golf swing) and want ChatGPT to understand it. Uploading the video directly is hit-or-miss. This is the method that works every time.

Why uploading the video often isn’t enough

Depending on your plan, app and model, ChatGPT may accept a video file but only look at part of it, ignore the audio, or refuse large files. Even when it works, you can’t control which moments it pays attention to.

What ChatGPT is great at is reading text and looking at images. So the reliable approach is to turn your video into exactly that.

The method: screenshots + transcript

  1. Take screenshots of the important moments. One every few seconds for a tutorial, or only when the scene changes for a talk or a slideshow.
  2. Get the full transcript with timestamps, so ChatGPT knows what was said and when.
  3. Match them up. Tell ChatGPT which screenshot goes with which part of the transcript. This is the step most people skip, and it’s what makes the answers good.
  4. Upload the images and the text together, then ask your question.

Doing it in one step with Unreel

Doing this by hand means scrubbing through the video, screenshotting, and typing explanations. Unreel does all of it for you:

Then drag the files into ChatGPT and start with: “Read README_for_AI.md first, then answer my question.”

Tips for better answers

Try Unreel, free →Drop a video or paste a YouTube link. Get screenshots + the full transcript in one folder your AI can read.