Skip to content
explore
Ideogram logo

Ideogram

Visit Ideogram

Image generation with accurate in-image text, typography, and layout.

ImageDesignMarketing

How it works / How to use

Image generation with accurate in-image text, typography, and layout. Write a prompt, generate images, then refine with editing tools such as Magic Fill, extend, remix, upscale, or background removal. Plan credits cover app features; API accounts and credits are billed separately from Ideogram app subscriptions.

  1. Open Ideogram and start a new generation (or open an existing image to edit).
  2. Write a prompt with subject, layout, style, and any text that must appear in the image—spell text exactly.
  3. Generate, pick the closest result, then use edit tools (Magic Fill, extend, remix, upscale, background removal) one change at a time.
  4. Download or share the final image; use the API separately if you need programmatic generation.

Posters and layouts with readable text

Lead with the text to render, then describe the scene, style, and composition. Put required copy in quotes.

What to provide

  • The exact words that must appear in the image
  • Layout intent (headline placement, hierarchy, background)
  • Style, color palette, and aspect ratio or use (poster, slide, social post)

Details that improve the result

  • Spell every word, number, and punctuation mark exactly
  • Say whether text should be centered, stacked, or on a badge
  • Name exclusions: no watermark, no extra words, no misspellings

Example prompt

Poster for a local book club. Text must read exactly: "Summer Reads" on top and "Every Thursday · 7pm" below. Warm illustration style, stacked typography, cream background, portrait orientation. No other text, no watermark.

If the first output is not good

  • “Keep the layout; fix only the second line spacing.”
  • “Same text; switch to a flat vector style.”
  • “Regenerate with bolder headline weight.”

Common mistakes

  • Paraphrasing text instead of spelling it exactly
  • Asking for long paragraphs of tiny text
  • Combining several unrelated layouts in one prompt

Logos and wordmarks

State the wordmark text first, then describe the mark, colors, and background. One concept per generation pass.

What to provide

  • Brand or product name spelled exactly
  • Icon or symbol idea, if any
  • Style constraints (minimal, geometric, hand-drawn) and background

Details that improve the result

  • Say “wordmark only” if you do not want a full scene
  • Name 1–2 colors explicitly
  • Request a plain or transparent background when exporting

Example prompt

Minimal wordmark for "Northwind Labs". Clean sans-serif lettering, navy on white, small compass icon above the wordmark, centered, no tagline, no shadow.

If the first output is not good

  • “Keep the icon; simplify the letterforms.”
  • “Try a circular badge layout with the same name.”
  • “Remove the icon; wordmark only.”

Common mistakes

  • Expecting trademark-perfect logo files without manual cleanup
  • Adding a tagline you did not specify
  • Asking for multiple logo directions in one crowded prompt

General image generation

Describe one scene with concrete visual details. Put negatives (no text, no watermark) at the end.

What to provide

  • Subject, setting, and action
  • Visual style, lighting, and camera or art direction
  • Aspect ratio and anything to exclude

Details that improve the result

  • One subject and one setting work better than a montage
  • Name materials, time of day, and palette when they matter
  • Say “no text” if labels would be a distraction

Example prompt

A ceramic mug on a wooden desk by a rainy window. Soft morning light, shallow depth of field, muted greens, 4:5 aspect ratio. No text, no watermark, no people.

If the first output is not good

  • “Same scene; switch to evening lamp light.”
  • “Tighter crop on the mug.”
  • “More photoreal, less stylized.”

Common mistakes

  • Vague mood words with no visual anchors
  • Stacking unrelated scenes in one prompt
  • Forgetting aspect ratio until after several generations

Magic Fill and local edits

Select the area to edit, then describe the replacement as a delta—not a full re-description of the whole image.

What to provide

  • The source image (generated in Ideogram or uploaded, if your plan allows)
  • The region or element to change
  • What the filled area should become, and what must stay the same

Details that improve the result

  • One major edit per pass
  • Name locked elements (face, typography, color grading)
  • Use Magic Fill when you need a local fix, not a new composition

Example prompt

On this poster, replace the empty plate with a slice of lemon tart. Keep the headline text, background color, and hand-drawn border unchanged.

If the first output is not good

  • “The fill is close; make the tart smaller and more centered.”
  • “Try a simpler food item with fewer details.”
  • “Keep everything else; remove the prop entirely.”

Common mistakes

  • Using Magic Fill to change the entire concept
  • Not saying which text or layout elements are locked
  • Stacking several unrelated edits before checking the first result

Extend (outpainting)

Start from the closest crop, then describe what belongs in the extended canvas and what must match the original (lighting, style, perspective).

What to provide

  • The source image
  • Which edges or sides to expand
  • What should appear in the new area and what mood to preserve

Details that improve the result

  • Extend when you need more background, not when the subject is wrong
  • Say whether the subject should stay centered
  • Match aspect ratio to the final export target

Example prompt

Extend this portrait crop upward and sideways to 16:9. Add more sky and rooftop background in the same watercolor style. Keep the person, clothing, and face unchanged.

If the first output is not good

  • “Extend only to the left; keep the right edge fixed.”
  • “More negative space above the subject.”
  • “Stop before adding new people or text.”

Common mistakes

  • Extending to fix a bad subject instead of regenerating
  • Asking for new focal elements that fight the original crop
  • Forgetting to specify the target aspect ratio

Remix and variation

Point to the base image, then list 1–2 controlled changes. Remix works best when you want alternatives, not a unrelated scene.

What to provide

  • The source image or a strong prior generation
  • What should change (style, palette, props) vs stay (composition, text)
  • How far the variation should diverge

Details that improve the result

  • Say “keep the same text” when typography is critical
  • Request subtle vs bold variation explicitly
  • Generate a few options and pick before further edits

Example prompt

Remix this event flyer. Keep the exact headline text and date. Try three variations: neon nightlife palette, muted editorial palette, and high-contrast monochrome.

If the first output is not good

  • “Variation 2 is closest; remix again with warmer tones only.”
  • “Keep composition; change only illustration style to flat vector.”
  • “Discard text changes; preserve copy verbatim.”

Common mistakes

  • Using remix when you need a precise local fix (use Magic Fill instead)
  • Allowing text to drift when accuracy matters
  • Asking for totally different subjects while calling it a remix

Upscale and background removal

Finish composition and text first, then upscale or remove background as export prep—not as a substitute for fixing the prompt.

What to provide

  • The final or near-final image
  • Target use (print, slide, web hero)
  • Whether you need transparency after background removal

Details that improve the result

  • Upscale after you approve content and layout
  • Background removal works best on clear subject edges
  • Remember these tools consume plan credits

Example prompt

Upscale this approved product photo for a 3000px-wide web hero. Then remove the background and deliver a clean cutout with soft edge, no color fringe.

If the first output is not good

  • “Upscale only; do not change colors.”
  • “Background removal missed the handle—clean up the left edge.”
  • “Export with transparent background.”

Common mistakes

  • Upscaling early and then re-editing text-heavy layouts
  • Expecting perfect hair or glass edges without touch-up
  • Removing backgrounds before fixing spelling in the image

API and programmatic generation

Treat the API as a production pipeline: send structured prompts, handle async jobs if applicable, and validate text-heavy outputs before shipping.

What to provide

  • Separate API account and credits (not the same pool as app subscriptions)
  • Prompt, aspect ratio, and any rendering options your integration needs
  • Downstream storage, moderation, and retry rules

Details that improve the result

  • API accounts and credits are billed separately from Ideogram app subscriptions
  • Reuse the same text-spelling discipline as the web app
  • Log prompts and outputs when typography must be audited

Example prompt

Generate a 1:1 social asset via API. Prompt: 'Weekly menu board, text exactly: "Soup · Salad · Pasta", chalkboard style, centered typography, no other words.' Store image URL and reject if OCR detects misspellings.

If the first output is not good

  • “Retry with a shorter text string.”
  • “Pin style preset; vary only the headline copy.”
  • “Add a server-side check before publishing.”

Common mistakes

  • Assuming app subscription credits cover API usage
  • Sending long marketing copy without validation
  • No human or automated check on rendered text before publish

How to prompt

Ideogram responds best when you treat text as part of the spec: spell words exactly, describe layout and hierarchy, then add visual style. Iterate on the closest generation with edit tools instead of rewriting the whole scene each time.

Use: social ad, 4:5.
Text exactly: "Free workshop · Register now".
Visual: friendly illustration, laptop and coffee on a desk, warm daylight.
Exclude: no extra words, no watermark.

Put required text in quotes

Ideogram is strong at in-image typography when you spell the copy exactly and separate it from visual description.

Example

Event poster. Text exactly: "Open Mic Night" and "Friday 8pm". Minimal gig-poster layout, dark background, bold centered type.

Describe layout before decoration

Headline placement, stacking order, and aspect ratio reduce garbled type and awkward cropping.

Example

Stacked layout: headline top third, subhead middle, illustration bottom half. Portrait 4:5. No border text.

One scene, explicit negatives

Concrete exclusions (no watermark, no extra text, no logo) prevent common artifacts.

Example

Product still life on marble. No text, no hands, no watermark, no brand marks.

Edit with deltas, not full rewrites

Magic Fill, extend, and remix work best when you change one dimension and lock the rest.

Example

Keep the headline and colors. Replace only the background with soft gradient blue.

Separate app work from API work

App subscriptions and API billing are separate. Prompt the same way, but plan credits and monitoring for production pipelines.

Example

In the app, approve a poster layout manually. In the API, reuse that prompt template with swapped headline text only.

Best output tips

Treat text as a hard requirement

Put exact copy in quotes, including punctuation and capitalization. Ideogram is known for unusually reliable in-image text when the prompt is precise.

Lead with layout and aspect ratio

Say where text sits, how it stacks, and the target ratio (4:5, 16:9, square) before style adjectives. Layout first reduces rework.

Use one concept per generation

One subject, one setting, and one typographic hierarchy per prompt. Split unrelated ideas into separate runs.

Iterate on the closest result

Regenerating from scratch wastes credits. Select the best image, then use Magic Fill, extend, remix, upscale, or background removal for targeted fixes.

Lock elements during edits

When refining, name what must not move: headline copy, face, palette, border, or composition. Unlocked elements may drift.

Choose the right edit tool

Magic Fill for local replacements, extend for more canvas, remix for style or palette variations, upscale and background removal for export prep.

Plan credits for edit features

Magic Fill, extend, upscale, remix, and background removal consume plan credits on the Ideogram app. Budget for both generation and refinement.

Keep API usage separate

API accounts and credits are billed separately from app subscriptions. Use the same prompt discipline in production, plus validation for rendered text.

Validate before publish

Read every rendered word on text-heavy assets. For logos and posters, a quick manual check beats assuming the first generation is final.

Say what to exclude

Explicit negatives—no watermark, no extra words, no people, no logo—reduce cleanup time more than hoping the model omits clutter.

  • Spell every word that must appear in the image; Ideogram is strong at typography when copy is explicit.
  • Describe one scene, one layout, and one aspect ratio before generating.
  • Put negatives in the prompt: no watermark, no extra text, no misspellings.
  • Pick the closest generation, then refine with Magic Fill, extend, remix, upscale, or background removal.
  • Change one thing per edit pass—especially when text or layout must stay locked.
  • Finish content and spelling before upscaling or removing backgrounds.
  • Remix for controlled alternatives; Magic Fill for local fixes; extend for more canvas.
  • API accounts and credits are separate from Ideogram app subscriptions—check billing before production use.
  • Review text-heavy outputs before publishing; regenerate or edit if any word is wrong.

Try this AI

Try Ideogram

Product Details

Pricing, features, limits and latest updates

Ideogram

Image generation with accurate in-image text, typography, and layout.

Free / Paid · Free

Pricing Plans

Free

Free

Plus

$180/ Yearly

Pro

$504/ Yearly

Team

$240/ Yearly

Enterprise

Custom

Key Features

Image Generation

Generates images with unusually reliable in-image text for logos, posters, and layouts.

API

API accounts and credits are billed separately from Ideogram app subscriptions.

Image Editing

Magic Fill, extend, upscale, remix, and background removal use plan credits.

Limits

  • Prompt length: Plain-text prompts work best under about 150 words; API billing is separate from app subscriptions (Ideogram docs).

Ideas / Prompt experiences

Share a prompt that worked for you. Username and email are shown with your submission. External links are not allowed.

Example prompt

Poster with exact headline text

Prompt

Event poster for "NIGHT MARKET" headline in bold sans-serif, subhead "SAT 7 SEPT · 6PM" below, neon market scene background, high contrast, portrait layout

Short explanation

Ideogram typography guidance: spell text exactly, put important words early, avoid font names.

Example prompt

Magic Fill localized edit

Prompt

[select region] Replace the blank sign with the text "OPEN LATE" in white condensed lettering. Keep the wall texture and lighting unchanged.

Short explanation

Use edit tools one change at a time after picking the closest base generation.

Share your experience

Required fields are marked. Variation, result, and explanation are optional.

You may also like

Midjourney logo

Midjourney

Explore

AI image and video generation from text prompts and reference images.

ImagePaid

Runway logo

Runway

Explore

AI image and video generation and editing for motion, storytelling, and creative workflows.

ImageFree / Paid

Adobe Firefly logo

Adobe Firefly

Explore

Generative AI for images, video, audio, and design built for commercially safe creative work.

ImageFree / Paid

Back to Explore