Moving and resizing the person
Six one-click actions in the editor's Avatar tab reposition the main subject in a finished thumbnail and reconstruct the background they were covering:
| Action | What it does |
|---|---|
| Move to Left | Repositions the subject into the left third, inpainting behind them |
| Move to Center | Centres the subject |
| Move to Right | Repositions into the right third |
| Make Bigger | Enlarges the subject toward a close-up |
| Make Smaller | Shrinks them so more background shows |
| Face Camera | Rotates them to face the viewer directly |
Each is one edit and one request. Identity, background style and any text stay as they were.
Pro and up
The Avatar tab — position, swap, pop-out and remove — unlocks on Pro. See plans and limits.
Why this beats regenerating
The most common flaw in a generated thumbnail is not the subject or the style — it is placement. The person is dead centre and there is nowhere for the headline to go, or they are too small to recognise at sidebar size. Regenerating rolls the dice on the entire composition to fix one thing. These actions fix only that thing.
Where the subject should actually be
- Off-centre beats centre. Subject in one third, text in the other. A centred subject with text on both sides fights itself.
- Bigger than feels right. Design at 1280×720, consumed at 168–360. If the face looks slightly too large on your monitor, it is about right in a feed.
- Face the camera for reaction content. Direct eye contact is a stronger hook than profile — that is what Face Camera is for.
- Smaller for scene-led videos. Travel, real estate and build videos are about the place. Make Smaller lets the location carry the frame.
- Leave the safe zones clear. The bottom-right corner is covered by the duration badge on YouTube. Nothing important goes there.
Actions vs. pop-out
| Avatar actions | Pop out & move | |
|---|---|---|
| Precision | Preset thirds and sizes | Drag to any pixel |
| Reliability | Consistent on any image | Depends on how cleanly the subject separates |
| Result | A regenerated frame | A background plus a movable layer |
| Use when | "They need to be over there" | "They need to be exactly here" |
Start with the actions. They are one click and they work on frames where a cutout would struggle. Reach for pop-out when you need a specific position the presets cannot reach.
If the move looks wrong
- Background reconstruction is smeared. The area behind the subject was detailed. Undo, and try a smaller move — or use a scene preset afterwards to give the reconstructed area something to be.
- The face changed identity slightly. Each generative edit reinterprets. Swap your avatar back in after the move rather than before it.
- Text is now behind the subject. Move the text layer — layers are free to reposition.
- Enlarged and now cropped badly. Undo and use Make Bigger once rather than twice; two enlargements in a row usually cut the head.