Blog to carousel with AI: plan five consistent social cards

Paste your blog post into the creative assistant, or supply a readable public URL and verify that the text was retrieved. State the audience and the single thing they should be able to do afterwards. Keep the original link and use material you are entitled to adapt.

Start with copy only. The carousel workflow can help plan and generate individual images, but it does not publish or schedule posts for you.

Here is a five-card plan for our own reference-image guide. Read the source before using the outline.

Remove repetition between cards 02–04. The closing page should help the reader act, without adding a new statistic or promising a guaranteed result.

Turn the source text below into exactly five social cards for beginner image creators.
Sequence: question, subject reference, style reference, composition reference, next action.
For each card, provide its number, one short headline, no more than two short sentences, the supporting source passage and a separate image prompt.
Preserve the difference between keeping a subject and copying a style. Do not invent guarantees, statistics or testimonials.
Use one visual system: cream background, dark purple headings, muted green accents, equal margins and a page number in the same corner.
Plan a 4:5 portrait series. Each card must be a separate file, not a five-panel contact sheet.
Only give the outline and complete page prompts now. Do not generate images until I approve the copy.
Source text: [paste the article here]

After review, request card 01 only with the approved copy and your selected image settings. Use the accepted card as an explicit reference when requesting the remaining pages. Confirm the actual export dimensions; a ratio in a prompt is not an exact pixel-size setting. Keep ordered filenames such as 01-cover and 02-subject.

Revise card 03 only. Use the approved card 01 as the visual reference.
Headline: Style is a separate instruction.
Body: Describe the palette and atmosphere. State which subject details must stay.
Keep the cream background, purple heading, green accent, margins and page-number position.
Do not change cards 01, 02, 04 or 05. Return one revised card, not a new full set.

Check the new result against its neighbours. If you need exact repeated typography, assemble text and imagery in the image editor and keep editable projects. See text-layer editing; a generated image does not automatically contain editable text.

Arrange the pages in order on the canvas or in your file viewer. Read them at phone size, then inspect each at full resolution.

Check the destination platform’s current upload requirements and preview its crop before publishing. Five portrait cards are this exercise’s design choice, not a claim about every platform’s format or page limit.

Can I make a LinkedIn PDF carousel? This workflow gives you separate image cards. For a document post, assemble them in a compatible document tool, export a PDF and inspect the page order and rendering before uploading. Do not rename an image to .pdf.

Can I ask for just the copy? Yes. Text work still follows assistant billing; image generation only happens when requested. Each new generated page or variant uses the applicable Credits.

Will the series keep exactly the same design? References help, but generated layouts can drift. Reusable text layers and manual alignment give you more control.

For a single visual summary, try article to infographic. For a seasonal campaign with several media formats, see the Black Friday content kit.