Using AI images for social media content
Matching ratio and style to each platform, and turning one prompt into a week of posts.
SamImageGenerator ·
Social media is where a generator pays for itself fastest, because the images are numerous, short-lived and format-specific. The same idea has to exist as a square, a tall story and a wide banner, and each one wants a different composition rather than a different crop.
Generate the format, do not crop it
| Placement | Ratio | Composition note |
|---|---|---|
| Feed post | 1:1 | Subject centred, space for the caption overlay |
| Story or reel | 9:16 | Subject in the upper two thirds, text goes below |
| Link preview | 16:9 | Simple, readable at thumbnail size |
| Profile or avatar | 1:1 | One clear subject, no fine detail |
Cropping a wide image into a story throws away the sides you framed the shot around. Generating it again at 9:16 costs one credit and gives you a composition that was built for the tall frame, which is the whole point of choosing the ratio before generating.
Making a feed look like one account
Consistency comes from two habits: one style for everything, and a repeated sentence in every prompt describing the palette and light. "Muted sand and terracotta palette, soft daylight" pasted into each prompt is enough to make ten unrelated subjects read as one account.
Leave room for text. If a caption, a headline or a logo goes on top, ask for it: "plain area in the lower third", "simple background on the left". Generating a busy image and then covering a third of it is how good images become unreadable posts.
What not to post
Do not generate real people, real logos or anything that reads as documentary. Label AI imagery where the platform asks for it — several now require it — and keep product photos honest: the item in the picture has to be the item you ship. The rest of the routine, including the prompt patterns, is in the prompt guide.