Ideogram V4 Prompting Guide

Ideogram V4 prompting is typography and layout work, not a mood-board caption. Treat the box like a lettering brief: deliverable, quoted copy, hierarchy, placement, palette, and what must stay out. On Versely you run Ideogram V4 at 5 credits per generation (flat rate on the live model page), then iterate the brief instead of restacking adjectives.

This guide is the technique hub for posters, labels, logos, and any still where the words are part of the picture. For row routing when lettering is the job, see Ideogram owns words in the frame. Stills siblings: Seedream 5 prompting guide and GPT Image 2 prompting guide. A Recraft prompting guide is upcoming.

Plain language vs JSON layout

Ideogram V4 was trained on structured captions. You have two honest ways to feed it.

Plain language is a normal sentence or labeled brief. Good for first ideas, loose comps, and hunting a direction. On Ideogram's own stack, Magic Prompt (expansion) can rewrite a short plain prompt into a full JSON caption before the model sees it. Valid, well-formed JSON usually passes through as-is.

JSON layout prompting is the production dialect: labeled scene, style, background, and elements. Text elements carry a literal text field. Optional bbox pins placement on a 0-1000 grid. Hex palettes lock brand color. Use JSON when the headline band, logo zone, or palette cannot freestyle.

Job	Prefer
First idea, exploration	Plain language (+ expansion when available)
Brand poster / packaging with fixed copy	JSON + text elements
Exact placement across seeds	JSON + bbox
Exact brand hex	JSON + uppercase color_palette
Campaign look lock	JSON + style code from a hero (when the UI exposes it)

On Versely, paste either form into the Ideogram V4 prompt on /tools/text-to-image. If the composer later exposes a Magic Prompt or expansion control, treat it as exploration only. For non-negotiable layout, write the JSON yourself so nothing invents extras.

Worked prompt 1 (plain language poster)

DELIVERABLE: 9:16 event poster, high contrast.
SCENE: Single silhouetted figure on an empty street under one lamp at twilight. Deep blues and warm amber.
TEXT: Large serif gold headline across the lower third reads "THE NIGHT BEGINS AT EIGHT". Smaller sans white tagline below reads "A STORY ABOUT WAITING". Two lines only. Quote case as written.
CONSTRAINTS: No other text, no watermark, no logos, no QR codes.

Worked prompt 2 (JSON layout sketch)

{
  "high_level_description": "9:16 event poster, silhouetted figure under a street lamp at twilight",
  "style_description": {
    "aesthetics": "editorial, high contrast, cinematic",
    "lighting": "single warm lamp from upper right, cool blue fill",
    "medium": "graphic_design",
    "art_style": "contemporary poster design"
  },
  "compositional_deconstruction": {
    "background": "empty wet street, soft reflections, deep blue twilight sky",
    "elements": [
      {
        "type": "obj",
        "bbox": [180, 350, 720, 650],
        "desc": "single silhouetted standing figure, mid-frame, facing away"
      },
      {
        "type": "text",
        "bbox": [720, 80, 820, 920],
        "text": "THE NIGHT BEGINS AT EIGHT",
        "desc": "large gold serif capitals across the lower third, generous tracking"
      },
      {
        "type": "text",
        "bbox": [830, 120, 900, 880],
        "text": "A STORY ABOUT WAITING",
        "desc": "smaller white sans under the headline, centered"
      }
    ]
  }
}

Keep elements lean (about five or fewer). Bounding boxes are layout guides, not pixel crops. Leave margin. Floor and pavement belong in background, not as boxed objects.

Quoting text so letters stay literal

Ideogram V4 treats quoted strings as render instructions. Three habits:

1. Put required wording in straight double quotes in plain prompts, or in a JSON text field.
2. Describe treatment in a separate clause (or in desc): weight, case, color, size feel, placement.
3. Fence the frame: render verbatim, exactly as written, no extra characters and no other text anywhere in the frame.

Keep each quoted string short. Accuracy is strongest on large, short English lines. Long paragraphs, tiny legal microcopy, and dense menus are weaker inside the still. Prefer two clean lines. Spell unusual brand names carefully and proofread before you ship.

Do not rely on naming a commercial typeface by product name. Describe the feel instead: condensed bold sans, soft script, slab serif display, mono label lettering.

Weak: A poster that says something about summer vibes.

Stronger: headline "SUMMER SALE" in bold condensed cream sans across the top third; subline "ENDS SUNDAY" in small italic under it; two lines only; no other text.

For tiny numeric labels, spelling out the word (forty-two percent) often holds better than a micro 42%. Large display numerals are usually fine as digits.

Posters and labels

Posters and product labels are where Ideogram V4 earns the row. Name deliverable and aspect first (9:16 story, 4:5 feed, 1:1 pack front). Then hierarchy: one hero line, one support line, optional corner mark. Give type a quiet band so illustration does not chew the glyphs.

Poster checklist: aspect and safe margins; hero string quoted, short, high contrast; shorter support line; subject off the type band; palette named or hex in JSON; no extra text, watermark, or invented logos.

Label checklist: SKU shape and finish before vibe; brand and flavor as separate quoted strings; quiet face for the lockup; no invented barcode / no rewritten ingredients block when those stay outside the model; short brand lines in-image, long ingredient panels often composite better externally.

Worked prompt 3 (pack label)

DELIVERABLE: 1:1 packaging hero, soft gray seamless.
SUBJECT: Matte navy coffee bag with a cream valve, front-facing, slight three-quarter tilt.
TEXT: Large condensed cream headline on the bag reads "ALTITUDE". Smaller line under it reads "SINGLE ORIGIN". Two lines only. Quote case as written.
LIGHT: Softbox from upper left, gentle fill from the right, realistic fabric grain.
CONSTRAINTS: No other text, no barcode invent, no watermark, no props that hide the lockup.

When placement must survive seed changes, move to JSON and pin headline and subline with bbox. Soft placement (across the top third) is enough for exploration; coordinates are for finals.

When Magic Prompt / expansion helps

Magic Prompt (and similar expansion) rewrites a short idea into a structured caption. Use it for sparse sentences, early style hunts, and JSON-quality structure without writing braces yet.

Skip or override expansion when hex, exact wording, and zones are non-negotiable, when you already have validated JSON, or when expansion keeps inventing props and committing your hedging for you.

Expansion is opinionated: it may fill empty space, pick one value when you wrote two, and strip soft language. Useful for drafts, risky for brand finals. On Versely, explore in plain language, then rewrite the winner as JSON (or a tighter labeled brief) before hero spends. If no Magic Prompt control is visible, the same fork is what you paste: short plain text to explore, structured JSON to lock.

Failure modes

Failure	Likely cause	Fix in the brief
Gibberish or rewritten type	Unquoted copy, long strings, busy zone	Quote short strings; quiet band; verbatim fence
Headline drifts every seed	Only soft placement language	JSON bbox on text elements; fewer elements
Palette freestyles	Mood words only	Uppercase #RRGGBB palette in JSON; include bg and shadow
Extra props and fake logos	Expansion or thin constraints	Write JSON; constraints: no other text, no watermark, no logos
Tiny legal soup	Asking for a paragraph in-frame	Short brand lines only; composite microcopy outside
Generic centered poster	No aspect, hierarchy, or negative space	Name ratio, hero/support roles, empty type band
Soft edges after approval	Asking the model to "fix" cleanup	On-device /free-tools/image-upscaler or /free-tools/background-remover

Exclusions live in the prompt. Targeted absences beat long boilerplate stacks.

Skip the dialect: brief the Versely agent

You can learn this stack. On Versely you can also skip memorizing every Ideogram dialect and brief the agent instead of writing prompts: content type, photos or references, edits you will accept, target platforms, and a budget ceiling. Name the row when it matters ("use Ideogram V4"). The agent plans the job; you approve the plan.

A prompt still wins when you are hand-tuning one hero poster. A brief wins when the job is bigger than one generate: variants, localization, cleanup, and a post path with a spend cap.

After the take: edit, post, collections

When the Ideogram V4 pass is close enough:

1. Finish cutouts and sharpening on-device with /free-tools/background-remover and /free-tools/image-upscaler when those are the remaining gaps. Use /free-tools/object-eraser for stray props.
2. Optional motion: send a winning still into /tools/ai-video-generator as image-to-video. Prompt motion on the video row; do not ask Ideogram for a timeline.
3. Upload or schedule to connected social accounts, or export PNG/JPEG and post from the native app.
4. Save keepers into a Versely collection so the next brief reuses brand hex lists, winning JSON skeletons, quoted lockups, and approved posters.

Honest limit: Ideogram V4 is a stills generator tuned for text and layout, not desktop publishing and not a video timeline. Board separate generates for separate setups. Credits apply on every generate.

FAQ

What is the best Ideogram V4 prompt formula?

Deliverable and aspect first, then scene, quoted text with hierarchy, placement (or bbox), style or palette, and constraints. Plain language to explore; JSON when layout and color must stick.

When should I use plain language vs JSON?

Plain language (and Magic Prompt / expansion when available) for first ideas. JSON for posters, labels, brand hex, and repeatable placement. Valid JSON should pass through unchanged.

How do I get readable text in the image?

Quote the exact string, keep it short, describe treatment and placement, and fence with verbatim / no other text. Prefer large hero lines over dense paragraphs. Proofread before you ship.

How much does Ideogram V4 cost on Versely?

Ideogram V4 is listed at 5 credits per generation (flat rate) on the live model page. No free generation tier on this row. Confirm the composer total before a batch.

Where do I run this on Versely?

Open text-to-image with Ideogram V4 selected, or start from the model page. Animate winners later in the AI video generator via image-to-video.

How is this different from Seedream 5 or GPT Image 2 prompting?

Same stills-brief discipline, different strengths. Ideogram V4 leans lettering, posters, and layout control (including JSON). Compare Seedream 5 prompting guide and GPT Image 2 prompting guide. A Recraft prompting guide is upcoming for the design-system lane.

Takeaway

Prompt Ideogram V4 like a lettering and layout brief, not like a vibe caption. Explore in plain language, lock finals in JSON when placement and hex matter, quote every load-bearing string, and keep type short and high contrast. Run the 5-credit row on Versely, or hand the agent a job brief when you would rather not memorize the dialect. Clean winners on-device when upscale or remove-bg is enough, animate keepers via image-to-video when motion is next, and file keepers in a collection so the next poster starts warmer.