Video to Shorts, whatever the source

A webinar does not convert like an interview, and a screen recording does not convert like either of them. This page is the source by source guide: which reframing mode each type needs, which moments are worth pulling, and where each one usually goes wrong.

First Short free. No account, no software, nothing to upload.

Interviews and panels Webinars and lectures Screen recordings Stream VODs

Start here

The conversion depends entirely on what you are converting

Most converters offer one button and one behaviour: crop the middle, add captions, done. That works for a single person filmed centrally and fails everywhere else, which is why so many converted Shorts look subtly broken without the person who made them being able to say why.

There are really only three conversion modes, and choosing the right one for your source is ninety percent of the quality. Everything below is about making that choice correctly.

Mode 1

Track the subject

Find the person and put the vertical window on them, following as they move or as the conversation switches speakers. The right answer for anything where a human is the content.

Mode 2

Fit the whole frame

Keep the entire widescreen picture inside the vertical canvas and fill the space above and below. Nothing is cropped away. The right answer whenever detail at the edges matters more than scale.

Mode 3

Leave it alone

If the source is already vertical or square, no reframing happens and every pixel survives. Rarer than you would think, and the easiest case when it turns up.

The lookup table

Which mode your source needs

Find your source type, use the mode in the second column, and avoid the mistake in the last one.

Source	Mode	Pull these moments	Common mistake
Two person interview	Track the subject	Disagreements, counterintuitive claims	Centre crop, which frames the empty table
Panel with three or more	Track, or fit if everyone matters	One person's complete answer	Clipping a crosstalk moment nobody can follow
Webinar with slides	Track the speaker	Speaker driven explanations only	Clipping a slide, which nobody watches
Screen recording or tutorial	Fit the whole frame	One completed action, start to finish	Cropping, which cuts off half the interface
Conference talk	Track the speaker	The single strongest argument	Starting before the point begins
Stream VOD	Track, unless gameplay matters	Reactions your chat already flagged	Keeping the overlay clutter in frame
Lecture or course	Track the lecturer	One self contained idea	Clipping something that needs the prior module
Product demo	Fit, then crop for the reaction	The moment the thing actually works	Clipping the setup instead of the result

Source by source

The four hard cases, in detail

Webinars, where half the frame is a slide

The classic webinar layout puts slides across most of the frame and the presenter in a small inset. Both halves want opposite treatments, and no automatic mode handles both at once.

The workable answer is editorial rather than technical: only clip the stretches where the presenter is carrying the moment, and crop to them. Slide heavy passages are not Short material in the first place, because a vertical video of a bullet list is something people scroll past instantly. If the slide genuinely is the point, use fit mode and accept a smaller picture.

Screen recordings, where cropping is always wrong

Cropping a screen recording is destructive by definition. The important detail in a tutorial is spread across the full width, often in small text, and the vertical window keeps less than half of it. Use fit mode without exception here.

The other half of the job is choosing the window. A screen recording clip should show one completed action from beginning to end, because a partial action is worse than useless. That usually means a slightly longer clip than you would cut for a talking head, and that is fine.

Stream VODs, where the overlay fights you

Stream footage arrives with alerts, chat boxes, webcam frames and subscriber tickers arranged for a widescreen canvas. Crop that to vertical and you inherit a random slice of interface furniture.

Pick moments where the camera is the story, crop to the camera, and let the gameplay go. If the gameplay genuinely is the moment, fit mode keeps the whole scene at the cost of scale. Either way, the overlay is the thing to watch for in the preview before you render.

Panels, where three people share one frame

Tracking works when one person is clearly speaking for a stretch. It struggles during crosstalk, which is also when panels are most entertaining and least clippable.

The practical rule is to clip complete answers rather than exchanges. A panel moment that needs three voices to make sense needs too much context for a feed of strangers.

Settings

What to export, whatever the source

The output specification does not change with the input. This is the target for everything above.

Resolution	1080 by 1920
Aspect ratio	9 to 16
Container and codecs	MP4, H.264, AAC
Frame rate	Match the source, never convert
Length	Under 3 minutes, usually far under
Captions	Burned in, kept out of the bottom fifth

One export covers every platform. Shorts, TikTok, Reels and Facebook Reels all use the same frame, so a correct file posts everywhere untouched. The full specification with safe zones is in the size and specs reference.

Questions

Converting from an unusual source

Which source types convert well to Shorts, and which do not?

Anything speech driven converts well: interviews, webinars, lectures, panels, product demos, stream VODs. What converts badly is footage where the value is cumulative rather than momentary, and anything where the important detail is spread across the full width of the frame, such as spreadsheets, code editors and dense slides. Those need fit mode rather than a crop.

Do I need a YouTube link, or can I use another source?

A link is how you get footage to us without uploading it, so anything publicly reachable works. If your recording lives on a hard drive, the fastest route is to publish it unlisted first and then convert from that link, which also gives you a permanent home for the long version.

How do I convert a screen recording without destroying it?

Use fit mode, not a crop. Cropping a screen recording to vertical throws away the sides where most of the interface lives. Fit mode places the whole widescreen frame inside the vertical canvas and fills the space above and below, so nothing is lost. The picture is smaller, and for screen content that is the correct trade.

What about a webinar with slides on one side and a speaker on the other?

This is the hardest layout, because the two halves of the frame want opposite treatments. The usual answer is to pick moments where the speaker is doing the work and crop to them, and to leave the slide heavy stretches out of your Shorts entirely. A Short of a slide is rarely worth posting.

Can I convert a video that has no speech at all?

Yes, but you lose captions and the AI moment suggestions, both of which read the transcript. You are back to choosing the window yourself on the timeline, which is fine and often faster for visual content anyway.

What if my source is already vertical?

Then no reframing happens at all and the conversion is just a trim and a caption pass. Vertical sources are the easiest case and they keep every pixel.

Keep reading

Convert whatever you have

Paste the link, pick the mode your source needs, and check the preview before you render.

Open the converter