Face size in a YouTube thumbnail is the question no two guides answer with the same number, and the generator you tried answered by cutting the top of the head off. The honest answer is a size per job: a head about half the frame's height for a talking video, a third when an object shares the frame, none when the object is the story. One test settles all three. As of September 2026, YouTube's thumbnail impressions and watch time page says impressions are counted if the thumbnail is shown for more than 1 second and at least 50% of the thumbnail is visible on the screen. Half a thumbnail, one second: a face that reads on those terms is big enough.
Below: the three sizes, the test, the crops that read as mistakes, what changes on Shorts and TV, and how to set the size in WThumb. Face size is one of the twelve checks in YouTube thumbnail best practices for 2026.
Three sizes, three jobs
This section gives three face sizes and why the guides disagree: they answer different jobs. A talking-head guide wants a big face; a review guide wants room for the product. YouTube gives no percentage at all. As of September 2026, YouTube's Thumbnail & title tips page says content targeting casual viewers might focus on actions and emotions that are more universally relatable, like a shocked face. Emotion, yes. A number, no.
Head and shoulders, the head about half the frame's height
For a talking video the face is the subject, so it gets the room. Our own guidance, not YouTube's: head and shoulders, with the head about half the frame's height. At that size the expression reads on a phone and the words still have a side to sit on. I WAS WRONG beside a face at that scale is a thumbnail; beside a face the size of a coin it is a title card.
Waist up, when hands or an object matter
For a review, a cooking video or a repair, the thing in your hands is half the story, so the crop loosens to the waist and the head drops to about a third of the frame's height. You still read as a person with a reaction; the pan or the drill reads as the reason to click. IT BROKE beside a cracked phone says more than the face alone. The same crop fits the fish held up in the fishing thumbnail ideas and the plate in the cooking thumbnail ideas.
No face, for the object-led thumbnail
For a screen recording or a drone flight, the object is the face. A clean, large object with DAY 30 over it does the job; a tiny face in a corner adds clutter, not trust. Pick a size by asking what the viewer is clicking for: you, you and the thing, or the thing.
The one-second test
This section is the test that decides whether a face is big enough, and it comes from how YouTube counts a showing. As of September 2026, YouTube's thumbnail impressions and watch time page says impressions are counted if the thumbnail is shown for more than 1 second and at least 50% of the thumbnail is visible on the screen, and the same page counts them on computers, TVs, game consoles, Android, iPhone, and iPad. That describes the viewer: someone on a phone who saw half your image for about a second.
So the test is this. Open the finished thumbnail on your phone. Cover half of it with your hand, glance for one second, look away. Could you tell it was a person, and what the person felt? If not, the face is too small. There is no checker for this in WThumb; your phone and your hand are the instrument, and the viewer's too.
Two rules follow. The face has to read at half size, because half is what the count allows. And the face should never be smaller than the biggest word: when WORTH IT? is taller than your head, the words are the subject, and the face may not belong at all. Emotion is what a face carries that words cannot, and a face too small to show it is doing nothing. To run the same check on a desktop window and a TV as well as a phone, see how to preview a YouTube thumbnail before upload.
The photo is not the crop
This section separates two sizes that get confused: the face in the photo you shoot, and the face in the finished thumbnail. Our post on how to take a photo for a YouTube thumbnail says your head should fill a good third of the picture. That is advice for the input. It leaves air above the hair and the whole chest below, so a design can crop tighter later without running out of head.
The finished thumbnail crops tighter than the photo. A head-and-shoulders thumbnail with the head at half the frame's height is a photo at a third, cropped in. A waist-up thumbnail is the same photo used almost whole. A tight selfie, where the hairline already touches the top edge, serves neither: a design that wants the face larger has nowhere to go but through the forehead, and that is the cropped head you saw.
So shoot looser than you want the thumbnail to be. Stand back, or hold the phone at arm's length, and get the chest in. If a tight photo is all you have, say so in the brief: my face large but the whole head in frame gives the design a reason to leave air rather than crop.
Where the face sits, and which crops read as mistakes
This section covers placement and cropping, because the right size in the wrong place still fails the test. As of September 2026, YouTube's Thumbnail & title tips page says you can apply the rule of thirds to create interesting images. For a face that means the eyes in the upper third, not the center, and the face on one side with the words on the other. A face on the right with I QUIT on the left gives each half one job. Our post on why the composition matters more than the face goes deeper.
Crops that read as style
Some crops read as choices. A crop through the forehead, eyes and mouth large and the top of the head gone, is a close-up: the reaction-channel look, for when the expression is the whole promise. A crop at the shoulders with air above the hair is the standard portrait. Both keep the eyes and the mouth, which the one-second glance needs.
Crops that read as mistakes
A crop at the chin leaves a face with no jaw and looks pasted on. A crop at the ears squeezes the head into a column. A crop exactly at the hairline is the worst: not close enough to be a close-up, not loose enough to be a portrait, so the eye reads it as a slip. If a generator returns any of these, the fix is a phrase, not a slider: whole head in frame, in the brief or in a revision.
A brief to paste
- Words. I QUIT on one line.
- Instructions. My face large on the right, eyes in the upper third, whole head in frame with a little air above the hair, the words on the left.
The reference decides the size
This section explains the main lever on face size in a reference-based generator: the reference itself. In WThumb the reference is a brief, not a template. The idea carries over and the design is new: layout and typography are new on every run. But the scale of the subject is part of the idea. A reference where the creator's face fills the frame says a huge face is the point, and a huge face tends to come back. A reference with a small person beside a large car says the car is the point, and the face tends to come back small. Neither is copying; both are the idea carrying over.
So pick the reference for the scale you want, not only for the niche. If you want head and shoulders at half the frame's height, paste a link to a thumbnail built that way. Our guide to how to pick a reference thumbnail for your niche covers the search; add scale to its checklist.
The instruction line is the second lever. A reference with a small face plus my face large, right side, whole head in frame usually brings back a larger face. If it still comes back small, the levers were pulling against each other; swap the reference before rewriting the line.
The Shorts crop, the TV and the end screen
This section covers the three surfaces that change the face's size most.
The Shorts crop
As of September 2026, YouTube's Add custom thumbnails page says for videos, try to upload a thumbnail with an aspect ratio of 16:9, and for Shorts, 9:16. The tall frame has no side to put the words on, so the face moves higher and gets larger, the words drop below it, and a head-and-shoulders crop becomes a head crop. That is where a clipped hairline is most likely. Our post on landscape to Shorts covers the rest; for faces, tick both formats in one run and look at the 9:16 result on its own.
The TV
As of September 2026, YouTube's 5 tips for YouTube success on TV post, published in November 2024, says to think of your thumbnails as movie posters, that on TV screens they are supersized, front and center, and asks for high-resolution images (1280x720 or higher) that are clean, balanced and easy to read. A head at half the frame's height is the size of a dinner plate there, so the photo has to be sharp.
The end screen
The smallest surface is inside another video. As of September 2026, YouTube's Add end screens to videos page says end screens can be added to the last 5 to 20 seconds of a video and that a video element can feature a specific video. What the viewer sees in that element is your thumbnail at a fraction of the player's size. YouTube's thumbnail impressions and watch time page, read the same month, does not count that showing at all: content within the video player, for example in cards or end screens, is excluded. Fewer, bigger elements pass all three surfaces. A big face, three words and one object read on a phone, a TV and a corner tile.
How to do this in WThumb
This section is the run from start to finish, with face size set by the two levers above. Every result is a Full HD+ JPEG, sharper than 1080p.
- Paste the reference. A YouTube link (WThumb fetches only that video's public thumbnail) or a PNG, JPEG or WebP, with the face at the scale you want: the idea carries over, the design is new.
- Type the words and one instruction line. Up to three lines, so I QUIT on one, plus one plain line: my face large, right side, whole head in frame, the words on the left.
- Add your photo, shot from the chest up. With a photo you are placed in the design. Skip it and the thumbnail is faceless: the person in the reference is removed and the space filled in the same style. Nobody else's face is ever put in.
- Tick 16:9, 9:16 or both, and one or two designs. One brief, two crops; two designs gives you two takes to compare on your phone.
- Wait about thirty seconds. Then run the one-second test on each result.
- Revise it if the head is clipped. Press the pencil on the finished thumbnail and write bigger face, or do not crop my head. Each revision costs one credit (1 credit = 1 thumbnail), and it works later too, from the reopened conversation.
A waist-up photo gives the design room
A tight selfie sets a ceiling: a larger face can only mean a cropped one. A photo from the chest up can become head and shoulders, waist up or a close-up. Shoot loose, brief tight. On a new account the first two runs cost nothing, because the account starts with two free credits, no card needed; spend them on a loose photo and a tight one.
Shoot from the chest up. Let the design decide how close.
