OpenAI
Generate App Store and Google Play screenshots on Martini using GPT Image 2 — a strong narrative composition pick for screenshots that need to tell a feature story across the gallery, with reasonable in-image text rendering for short headlines. GPT Image 2 sits between Ideogram (typography champion) and FLUX.2 (literal product staging): it understands feature narratives better than either, which makes it useful for screenshots that frame a feature as a moment ("when your run finally clicks"). For pixel-perfect headlines, chain to Ideogram; for the device frame and feature staging, GPT Image 2 is solid.
GPT Image 2 reasons over scenarios. Skip "screenshot showing the leaderboard feature" and write: "A runner pauses on a coastal trail to check their phone, the screen displays a leaderboard with their name moving up the rankings, expression shifts from focus to satisfaction." GPT Image 2 composes the moment — phone, UI, person, expression, environment — as one coherent scene.
Save the visual language in a Text node: "Mobile app screenshot photography style, soft natural light, warm sunset palette, athletic apparel context, iPhone 15 Pro device frame in hand, clean modern UI design language." Wire it into every GPT Image 2 screenshot node; each concatenates the brand block with its feature-moment prompt. Brand voice stays consistent across the 8-screenshot set.
GPT Image 2 renders short in-image text reliably but Ideogram wins on legibility for App Store production. The cleanest pipeline: GPT Image 2 generates the feature-moment scene (device + person + UI + environment) without in-frame headline text, then a downstream Ideogram V3 node takes the GPT Image 2 output as a reference and adds the headline + CTA in legible bold typography. Two nodes, one canvas, App Store-ready output.
Apple requires 1290x2796 (iPhone 6.7"), 2064x2752 (iPad 13"), Google Play accepts 1080x1920. Pin target dimensions on each GPT Image 2 node before generation. GPT Image 2 composes per aspect — re-cropping later breaks the device-frame composition. Generate the screenshot at the App Store dimensions directly.
GPT Image 2 handles 5+ named elements per frame. For App Store screenshots with device frame + person + UI + environment + lighting + brand element, this is its sweet spot. Other Martini image models start dropping elements past 3-4 named items, which leaves screenshots looking incomplete or off-brand.
For multilingual stores, duplicate the entire 8-screenshot canvas per locale. The GPT Image 2 layer (scene, device, environment) stays identical across locales — only the downstream Ideogram headline changes. Or, regenerate the GPT Image 2 layer per locale if the screenshot needs locally-relevant cultural context. Same canvas, different headline strings, multiple locales shipped from one workflow.
Hero feature-moment screenshot. The "no in-image headline text" instruction keeps the layer clean for downstream Ideogram overlay.
[App brand block prefix] + Frame: a runner pauses on a coastal trail at sunset to check their iPhone 15 Pro, the screen displays a clean fitness app leaderboard UI with their name moving up the rankings, expression shifts from focus to satisfaction, soft golden palette, App Store 1290x2796 portrait, no in-image headline text (text added downstream by Ideogram).
Feature-summary screenshot framed as a quiet moment. GPT Image 2 reasons through the scene composition.
[Brand block] + Frame: a runner stretching on a sunny morning patio, iPhone 15 Pro on a small table beside them displaying a workout summary with route map and pace chart, the runner glances at the phone with quiet satisfaction, soft morning light, 1290x2796 portrait, no in-image text.
iPad screenshot at the required dimensions. Same feature-moment framing scaled to tablet.
[Brand block] + Frame: an iPad 13" Pro on a wooden desk in a sunlit room, screen displaying the app dashboard with multiple workout cards visible, a hand reaching toward the screen to tap a workout, 2064x2752 portrait, no in-image text.
Social feature framed as a community moment. The dense composition (multiple people + device + environment + lighting) plays to GPT Image 2's strength.
[Brand block] + Frame: a group of runners gathered post-workout at a coastal cafe, one of them shows their iPhone screen displaying the app's social sharing feature, friends laugh and lean in, warm golden hour light, sense of community, 1290x2796 portrait, no in-image text.
Frame each screenshot as a feature moment, not as a feature description. GPT Image 2 reasons through stories; pure spec sheets waste its strength.
Generate without in-image headline text and chain to Ideogram for the typography layer. Two-node pipeline produces App Store-ready output.
Save the brand block once and wire into every screenshot node. Editing the block once propagates to all 8 screenshots.
Use 5+ named elements per prompt for dense compositions. GPT Image 2 handles density better than FLUX.2 / Midjourney.
Match dimensions on the canvas (1290x2796 iPhone 6.7", 2064x2752 iPad 13", 1080x1920 Android). Re-cropping breaks device-frame composition.
Set quality to "high" for hero screenshots, "standard" for variants and exploration.
GPT Image 2 returns 1024-2048+ outputs at the requested aspect with strong feature-moment composition and consistent brand voice across the 8-screenshot set when wired through a shared brand block. Generation time 30-60s per screenshot. Best at narrative feature framing; chain to Ideogram for App Store-grade headline typography. Output drops onto the canvas; export the final screenshot set after Ideogram overlay and upload to App Store Connect / Play Console.
Connect GPT Image 2 with other AI models on Martini's infinite canvas. No GPU required — start free.
Get Started FreeIdeogram
Ship pixel-perfect App Store and Google Play screenshots on Martini using Ideogram V3 — the only AI image model that renders headline copy, feature callouts, and CTA buttons legibly inside the frame. Mobile founders and product marketers ship 5-8 screenshots per locale at the required iPhone 6.7", iPad 13", and Android dimensions, with localized headlines that have to read cleanly in-image. Ideogram V3 is the production tool for this — no Sketch round-trip, no Figma headline overlays, the typography ships baked into the image.
View guideGenerate App Store and Google Play screenshots on Martini using Imagen 4 — strong photographic realism and prompt adherence make it the right pick for screenshots that need to look like real product photography rather than illustrated mockups. Imagen 4 renders specific device frames, lighting setups, and environment compositions with near-literal accuracy. For app marketers who need photo-realistic device-in-context screenshots (a phone in hand, a tablet on a desk, a watch on a wrist), Imagen 4 is the cleaner pick than illustrated alternatives. Pair it with a downstream Ideogram node for the headline typography pass.
View guide