Text in Image, Covers & Carousels

What you will walk away with

By the end of this lesson you will be able to make images that carry legible, on-brand text — the thumbnail/cover that earns the click and a 5-slide carousel that gets saved — using Ideogram and Adobe Firefly, the tools built to spell words correctly. You'll produce one cover plus a five-slide carousel for a single pillar idea, and own a prompt that outlines the carousel before you make a single slide.

Watch the class

Primary video: (own screen-demo — to be produced) — a 5-8 minute Academy walkthrough: writing a cover in Ideogram, then building a 5-slide carousel with consistent type and color, end to end.

• Status: PRODUCTION PENDING (own demo). Until it ships, this written manual is the complete standalone class — you can do everything below without any video.

Assigned fallback: a validated external beginner walkthrough on Ideogram / text-in-image — to be sourced and embed-verified in G1/G3 (G1 found no standout to lock yet). Do not treat as present until a real ID passes the video-first checklist.

The concept: text-in-image is its own skill

The cover is the ad for your content. On a feed of thumbnails, a viewer reads three or four huge words in a fraction of a second and decides to tap or scroll — so a cover is not "an image with a title on it", it's a few words engineered to be read instantly, with color, contrast, and focal point all serving that read.

Carousels are the other high-value text format: multi-slide posts people swipe and save, and saves are a strong ranking signal (especially on Instagram). The 5-slide spine turns one idea into an arc: Hook → Problem → Insight → Steps → CTA. Same font, same 3-4 colors, same layout grid makes them feel like one designed piece. Keep words short — AI text stays clean at 1-6 words; put the detail in the caption.

Why dedicated tools (Verified: Jul 2026): general image models treat text as decoration and misspell it. Ideogram was built to render readable text; Firefly does too, is trained on licensed/"commercially safe" data, and drops into Photoshop. For anything with words on it, start in one of these — and the durable move is to generate the layout with AI, then set the final text yourself (in Canva or the editor) so spelling is never left to chance. If a label has moved, search the same words.

Step by step

You need: a free Ideogram account (https://ideogram.ai) and/or Adobe Firefly (https://firefly.adobe.com), one pillar idea to build around, and optionally Canva (https://www.canva.com) to finalize text.

Part A — The cover / thumbnail

1. In Ideogram, type your cover prompt, stating the exact words in quotes and the layout. Set the aspect ratio with the ratio control ("1
   " for a carousel cover or feed post, "16
   " for a YouTube thumbnail, "9
   " for a story/Reel cover).
2. Click "Generate". Ideogram returns a set of options — pick the one where the text is cleanest and the focal point strongest.
3. Check every letter. If a word is misspelled or cramped, regenerate, shorten the text, or — most reliable — generate the background clean and add the headline yourself in Canva with a bold font.
4. Download your cover (the download icon).

Part B — The 5-slide carousel

1. Run Prompt 1 below to get your five-slide outline (the words for each slide).
2. Build slide 1 (the cover) using Part A — this is the swipe-earning frame.
3. Build slides 2-5 in the same tool, same style: keep the aspect ratio identical ("1
   " or "4
   " for feed), reuse the same 3-4 colors and font family, and keep each slide to one idea in a few words.
4. To guarantee consistency and correct spelling, the pro move: generate clean backgrounds for all five in Ideogram/Firefly, then place the text on a shared template in Canva so type, size, and color are pixel-identical across slides.
5. Export all five in order. Slide 1 leads; slide 5 is the CTA.

Part C — Order and export

Read the five slides as a swipe: does slide 1 make you want slide 2? Does the last slide tell the viewer exactly what to do? Reorder or tighten, then export at the platform's size (1080x1350 for a 4

Prompts (copy and use)

Prompt 1 — Carousel Outline Generator

Purpose: turn one pillar idea into a ready-to-build 5-slide carousel — the exact short words for each slide — before you open any image tool. When to use: Part B, Step 1. Works in ChatGPT, Claude, or Gemini.

Turn this idea into a 5-slide social carousel outline.

IDEA: [IDEA — one sentence, the single takeaway]
AUDIENCE: [AUDIENCE — who it's for]
PLATFORM: [PLATFORM — Instagram / LinkedIn]

Use this 5-slide spine and give me the ON-SLIDE text for each (short — a headline plus
at most one line):
1. HOOK — a scroll-stopping cover line (max 6 words) + a subtitle line
2. PROBLEM — the pain or mistake the audience feels
3. INSIGHT — the reframe / the "aha"
4. STEPS — 2-3 concrete steps, one line each
5. CTA — one clear action (save, follow, comment a word)

Rules:
- On-slide text only — keep every line short enough to read at a glance; the long
  explanation goes in the caption, not on the image.
- Keep a consistent voice across slides.
- After the 5 slides, write the post CAPTION (2-4 sentences) and 5 hashtags.

Example (filled in): [IDEA] = "batching content beats posting daily" · [AUDIENCE] = "overwhelmed solo creators" · [PLATFORM] = "Instagram". Typical output — Slide 1: "STOP POSTING DAILY" / "there's a calmer way" · Slide 2: "Daily posting = burnout by week 3" · Slide 3: "Batch once, publish all week" · Slide 4: "1) Pick a day 2) Film 5 in a row 3) Schedule them" · Slide 5: "Save this + follow for the system" — plus caption and hashtags.

Variations: ask for a 7-slide version; ask for ES/PT slide text; ask it to also suggest a visual for each slide (background idea, one icon) so your image prompts are ready.

Warnings: the outline is a starting draft — cut any slide that doesn't earn its swipe. Don't paste unpublished client strategy into a tool you don't control. If a slide shows a realistic AI depiction of a real person/event, disclosure may apply (Module M8).

Example: cover before vs after

Before: a general image generator, "motivational post about consistency with the text 'Consistency Wins'" → a pretty gradient with "Consistancy Wns" — misspelled, unusable. After: Ideogram, "bold poster, huge white text 'CONSISTENCY WINS' centered on a deep navy background, high contrast, 1

Common errors and how to fix them

Exercise

Make one cover + a 5-slide carousel for a single pillar idea:

1. Run Prompt 1 on one idea from your pillars (lesson 0.2) to get the slide text.
2. Build the cover (slide 1) in Ideogram or Firefly — legible, high-contrast, one big idea.
3. Build slides 2-5 in the same style: same colors, same font, same aspect ratio.
4. Finalize the text (in-tool or on a Canva template) and export all five in order, plus the caption from Prompt 1.

Expected result: a 6-asset set — one cover and a five-slide carousel — that reads as one designed piece: consistent color and type, correctly spelled words, a cover that earns the tap, and a last slide with one clear CTA. This is a real, publishable post; it also seeds your Module M9 content calendar.

Final checklist

• Used a text-capable tool (Ideogram/Firefly), not a general image model, for words
• Cover = one big idea, few huge words, high contrast, readable at thumbnail size
• Every word spell-checked (or text set manually in Canva)
• Carousel follows Hook → Problem → Insight → Steps → CTA
• Same 3-4 colors + one font + one layout across all slides
• Correct aspect ratio set before generating (1
  / 4
  / 16
  )
• Caption + hashtags written; long text lives in the caption, not on-image
• Five slides exported in order

Toolkit for this lesson

Required

Also useful

Free-tier limits and prices change fast — dated benchmarks, not permanent facts.

Next step

Module M3 — Video with AI, starting with "The 2026 map: Veo, Kling, Runway (and why not Sora)" (3.1): your images are done — now you'll bring them to life. You'll generate your first AI video clips and, in 3.2, animate one of the consistent-character stills you made in this module into motion.