AI Images for Social Media: Sizes, Styles and a Workflow That Works
A Kanakaw guide · about 9 min read
Match the Ratio to the Placement
The most common reason an AI image flops on social media has nothing to do with the image itself: it was generated in the wrong shape. A wide 16:9 landscape image looks great on YouTube and in a Facebook post, but the same image dropped into a Story gets cropped to a thin strip, and the subject is cut in half. Decide where the image will live before you generate it, and pick the aspect ratio for that placement.
| Aspect ratio | Shape | Best for |
|---|---|---|
| 16:9 | Wide landscape | YouTube thumbnails and covers, Facebook posts, blog headers |
| 9:16 | Tall portrait | Stories, Reels and TikTok — the full-screen vertical format |
| 1:1 | Square | Feed posts that need to look right everywhere, profile grids |
| 4:3 and 3:4 | Near-square and upright | Pinterest-style pins and product shots with room for a caption |
| 2:3 | Tall portrait | Portrait posts and pins where a person or product should fill the frame |
When one post has to work in several places, start with the square 1:1 version. It survives cropping better than any other shape, and it looks intentional in a feed grid. Then generate the tall 9:16 version separately for Stories and Reels — do not just crop the square, because the composition changes completely when the frame goes vertical. Kanakaw lets you pick the aspect ratio before you create, so the framing, the subject placement and the background are all composed for the shape you actually need.
One more sizing tip: generate at the highest quality you plan to use, and do not upscale a small image after the fact unless you need to. A clean, well-composed image at the right ratio beats a stretched or heavily cropped one every time, no matter how good the subject is.
Pick a Style That Stops the Scroll
People scroll past dozens of images a day, and they decide in about a second whether to stop. The style you choose is what buys that second. Three styles cover almost every social post, and each one works for a different reason.
Photorealistic is the default for a reason: it looks like a real photo, so it earns trust fast. Use it for product shots, food, travel, lifestyle and anything where the viewer needs to believe the thing is real. Ask for natural light, a real setting and a specific lens feel — "shot on a phone, golden hour, shallow depth of field" reads very differently from "studio product photo, soft white background". The first feels like a moment; the second feels like a catalogue. Pick the one that matches what you are selling.
Illustration is the better choice when you want a post to be recognised as yours before anyone reads the caption. A consistent illustrated style — flat vector, watercolour, comic, or a soft 3D look — becomes a visual signature across your feed. It also sidesteps the uncanny valley: a slightly imperfect illustrated face is charming, while a slightly imperfect photorealistic face is unsettling. Illustration is the safer style for characters, mascots and explainers.
Bold graphic is for announcements, quotes and offers: a strong colour field, one clear subject, lots of empty space. AI generators are good at the image part and bad at the words part, so keep any text out of the picture itself (more on that below) and add the headline in your design tool or in the caption. A graphic-style image with a single subject and a clean background will outperform a busy scene almost every time, because the eye has somewhere to land.
Whatever style you pick, keep it consistent for at least a month. A feed where every post looks different reads as random; a feed with one recognisable style reads as a brand. That consistency is worth more than any single viral image.
Write Prompts That Stay On Brand
A good prompt for social media is not a long prompt. It is a specific one. Start with the subject, then the setting, then the light, then the style, then the mood — in that order. "A ceramic coffee cup on a wooden cafe table, morning light through a window, photorealistic, warm and calm" will give you a usable image. "A nice coffee cup" will give you whatever the generator feels like, and you will burn credits re-rolling it.
Prompts on Kanakaw can be up to 3,000 characters long, but you rarely need more than two or three sentences. The extra room is useful for one thing: a short brand note you paste into every prompt. Keep a one-line description of your look — your colours, your lighting, your setting — and reuse it. "Warm terracotta and cream palette, soft natural light, minimal styling" pasted at the end of every prompt does more for brand consistency than any preset, because it travels with the image from the first draft to the finished post.
Write the prompt for the image, not for the post. The caption is where the message lives; the image is where the feeling lives. If you find yourself writing "with the text 'SALE' in the top left", stop — AI text in images often comes out garbled, with misspelled words or letters that melt into each other. Keep wording in the caption, not in the picture, and let the image do the visual work. The same rule applies to logos: generate the scene, then add your logo in your design tool where it will be crisp and exactly where you want it.
Finally, save the prompts that work. When an image comes out right, keep the prompt next to it. Your next ten posts get faster, your feed gets more consistent, and you stop paying to rediscover the same wording twice.
A Batch Workflow That Saves Credits
Making social images one at a time is the expensive way to do it. The cheaper way is to batch: sit down once, make a week's worth of images, and schedule them. Here is a workflow that keeps the cost down without making the results look cheap.
Start in Standard, at 6 credits per image. Standard is the fast, everyday tier, and it is the right place to test ideas: generate up to 4 variations of one prompt at once, pick the composition you like best, and throw the rest away. A morning of testing might cost you 30 or 40 credits and leave you with five or six strong compositions. Only then switch to Premium, at 41 credits per image, for the keepers — the two or three images that will actually go out this week. Premium is the members-only tier, and membership comes with any credit pack (30 days with a monthly pack, 12 months with an annual pack), so there is no separate fee to think about.
When you already have a photo you like, restyling it is often faster than generating from scratch. Image-to-image restyling costs 49 credits: upload your photo, describe the change, and the generator keeps the composition while changing the style, the season or the lighting. It is the right tool for turning a decent phone photo into a polished post, or for making a whole set of images from one shoot look like they belong together. Small fixes go through inpainting at 21 credits — clean up a background, remove an object, fix a detail — and the upscaler at 61 credits enlarges an image you want to print or post large without visible artifacts.
Before anything goes live, check it like a follower would. Open the image full screen and look at the hands and the faces first: extra fingers, a hand that bends the wrong way, eyes that do not quite line up — these are the tells that make people scroll past, and they are easy to miss at thumbnail size. If a face looks wrong, re-roll it in Standard rather than trying to rescue it; a fresh generation is cheaper than an hour of fixing. And check the image at the size it will actually be seen: a Story image viewed on a phone hides small flaws, but a feed image viewed on a desktop does not.
One last honest limit: AI images are a starting point, not a finished post. The workflow that works is generate, pick, check, then add your caption, your logo and your link in the tools you already use. Keep the AI part to what it is good at — the picture — and your feed will look better and cost less than generating everything from scratch every day.
Frequently Asked Questions
What size should an AI image be for Instagram?
It depends on the placement. Use 1:1 square for feed posts, 9:16 portrait for Stories and Reels, and 2:3 portrait for tall posts. Pick the aspect ratio before you generate, so the composition is made for that shape instead of cropped afterwards.
How much does it cost to make a social media image with Kanakaw?
A Standard image costs 6 credits and a Premium image costs 41 credits. Restyling one of your own photos costs 49 credits, inpainting 21 credits, and upscaling 61 credits. You can generate up to 4 variations of one prompt at once, and every generation shows its exact cost before you run it.
Can I use AI images for my business's social media?
Yes. Subject to the Terms of Service, you may use the images you generate for personal and commercial projects, including business social media accounts, ads and product pages.
Why does the text in my AI image look wrong?
AI image generators often render words with misspellings or letters that melt into each other. Keep wording in the caption, not in the picture, and add headlines or logos in your design tool where they will be crisp and exactly where you want them.
Do I need a subscription to make social media images?
No. Kanakaw runs on one-time credit packs — there is nothing to cancel. Membership comes with any credit pack (30 days with a monthly pack, 12 months with an annual pack) and unlocks Premium images, video generation and photo restyling. Sign-in is with Google only, and there is no free trial.
Try it: AI Image Generator · Pricing · All guides