Post images without leaving

Text-on-image is a reach lever on LinkedIn, and it is usually the point where writing stops and a second tool starts. It doesn't have to be: Zsper already holds the post, the kind of post it is, and the line worth enlarging — so it can make the image itself.

The claim is narrow and worth stating plainly: the tool that wrote the post designs the image. You review it; you don't make it.

1. Set your look once

Go to Brand → Look. You can paste your website and let Zsper read its palette, or set six colours by hand:

• Canvas — the image's background
• Type — headlines and primary text
• Accent — the one colour that carries emphasis
• Secondary type — eyebrows, attribution, your handle
• Panel — cards and split panels on the canvas
• On accent — text inside a filled accent block

Then pick a typeface pairing, upload a logo, and add your photo if you want a small circular mark on your images. Everything shows against a real post image as you change it, so you're judging a headline rather than a swatch.

Two things are enforced rather than suggested. A colour pair that would be unreadable in the feed is nudged to the nearest readable version, and you're told exactly what moved. And text is never cut off — if a line can't fit, Zsper says which line and why instead of quietly clipping it.

2. Open the Visual step

Finish a LinkedIn post, then switch the preview to Visual. The image is already made: the layout came from the kind of post you wrote, the words came from your post, the colours came from your look.

It renders inside the feed mock, so you see what a reader sees — image and text together.

3. Change what you want

Click any element on the image to select it:

• Edit the text inline
• Nudge it with the arrow keys
• Resize the type a step at a time
• Re-align it, or recolour it to another of your brand colours
• Hide anything that isn't the point of the layout
• Reset any element back to where Zsper put it

Nothing you can reach is off-brand or broken. If an edit would push text off the safe area, into another element, or below the readable size, Zsper refuses that one edit and tells you which element — it never silently damages the image.

4. Swap the layout

Browse layouts shows every template that works on that shape, each one drawn with your copy in your brand rather than a stock thumbnail. Recommended ones come first. A layout that can't hold this post's copy on this shape says so rather than offering you a broken preview.

5. Download

Download gives you a PNG at LinkedIn's spec — 2× by default for a crisp feed render, or 1× for the exact platform size. Images are saved under Assets → Images, so you can re-download one later without rebuilding it.

What this deliberately doesn't do

• No freeform canvas. No layers, no dragging things anywhere, no arbitrary colours. That's not a missing feature — the blank canvas is the thing being removed. Every state you can reach is a valid, on-brand design.
• No AI-generated artwork. An image model can't hold your palette and logo across forty posts, and generated art reads as generated — the opposite of what building authority needs. Your images are built from type, your colours, and geometry.
• Nothing is posted for you. Zsper renders and offers. You download or attach it yourself.

Related: LinkedIn posts and newsletters · Import your writing