Here is the whole game, in one test. Shrink your latest frame to the size it will actually be seen — a thumbnail in a fast feed — and hold it at arm’s length. If a follower can tell whose page it came from before reading a single caption, you pass. If it could belong to anyone, you have a pretty picture and a branding problem.
Why small is the real size
Creators judge their work at full screen, lit beautifully, one frame at a time. Audiences meet it at forty pixels wide, between two strangers’ frames, for about half a second. Recognition at that size is not about detail — detail is the first thing the feed throws away. It is about silhouette and signature: a hair shape that reads as a blob of the right outline, a colour pairing that belongs to one character only, one detail that survives cropping.
The five anchors that survive shrinking
A character built for the sticker test carries five anchors, and each is chosen to still work when tiny. Age, stated and consistent, because a face that ages up and down between frames breaks recognition quietly. Hair, described as silhouette first — length, outline, one style default — colour second. Build, in two or three words. One face detail bold enough to hint at thumbnail size: strong brows, a beauty mark, striking eyes. And a signature item — the jacket, the clip, the colour she always wears somewhere. Write these into a character card once, paste the card at the top of every prompt, and only ever change the scene lines below it.
Running the test honestly
Do it once a week, on the week’s hero frame, against last week’s. Put both at thumbnail size side by side. Same girl? Same world? If either answer wobbles, the fix is almost never the model — it is a scene line that contradicted an anchor, or an anchor too vague to hold. Tighten the card, not your standards. The characters who pass this test week after week are the ones feeds learn to stop for, because recognition is the one thing an algorithm cannot give you and a scrolling thumb cannot ignore.