The MAI Image 2 prompt generator gives you 20 free, copy-ready prompts for Microsoft's #3-ranked AI image model. Best-in-class photorealism, natural lighting, and in-image text accuracy — powering Bing Image Creator and Copilot since March 2026.
The MAI Image 2 prompt generator on this page provides 20 free, professionally written prompts for MAI-Image-2 — Microsoft's first fully in-house AI image model, released March 18, 2026. MAI Image 2 is now the engine behind Bing Image Creator, Microsoft Copilot's image generation, and the MAI Playground, giving it one of the largest distribution footprints of any AI image model in 2026.
MAI Image 2 ranks #3 on the Arena.ai text-to-image leaderboard (May 2026), behind GPT Image 2 and ahead of all other competitors. Its headline strength is photorealistic portraiture and environmental photography — it produces the most convincingly natural skin tones, lighting, and physical environments of any model currently available. It also supports text-in-image generation significantly better than its predecessors, with readable storefront signs, poster headlines, and infographic labels.
All images generated by MAI Image 2 include automatic C2PA-compliant Content Credentials — a digital watermark standard that verifies AI-generated content. This makes it the preferred choice for editorial, journalism, and commercial workflows where content provenance matters. Every prompt below is optimised for MAI Image 2's specific strengths in photorealism, lighting, and typography.
MAI Image 2 responds to descriptive, scene-oriented prompts. Use this structure for best results:
Click any prompt to copy — paste into Bing Image Creator, Microsoft Copilot, MAI Playground, or Azure AI Foundry
A golden retriever puppy playing in autumn leaves, shallow depth of field, natural afternoon sunlight filtering through trees, photorealistic, 35mm lens. The light is warm amber, the leaves are red and orange, the dog is fully sharp against a softly blurred background.
A traditional Italian delicatessen storefront on a narrow European street. Above the entrance, a hand-painted wooden sign reads 'SALUMERIA TONELLI — Est. 1947' in dark green serif lettering on a cream background. The sign is weathered but clearly legible. Afternoon light. Photorealistic street photography.
A professional corporate headshot of a woman in her 40s, wearing a navy blazer, white blouse. She faces the camera directly with a composed, confident expression. Studio lighting: a key light from camera-left, soft fill from camera-right, no shadows on background. Background is medium grey. Photorealistic, 85mm lens equivalent, no retouching artefacts.
A dark amber glass perfume bottle on a black velvet surface, photographed from a 45-degree angle. A single overhead spot highlights the bottle's shoulder, creating a crescent of reflected light. The liquid inside glows in the backlight. The background fades to black. Luxury fragrance commercial photography style. Photorealistic.
A bright Scandinavian-style living room on a grey winter morning. White walls, pale oak floors, a linen sofa with a knitted throw, a single potted olive tree. Natural light from a wide north-facing window. The room is tidy but lived-in — a book on the arm of the sofa, a ceramic mug on the side table. Photorealistic interior photography.
A wide establishing shot of the Scottish Highlands at dawn: rolling moorland in the foreground, a dark loch in the middle distance, mountains with snow on their upper slopes behind. Low mist in the valley. The sky is pale pink and grey, the sun just cresting a ridge. No people, no buildings. Photorealistic landscape photography, ultra-wide lens.
A clean data visualization poster on a white background. Title at top: 'Global Renewable Energy Growth 2020–2026' in bold sans-serif, dark grey. Below: a stylized bar chart in blue and green showing rising values. Three key stats shown as large pull-quote numbers: '47%', '2.3 TW', '140 countries'. Caption text at bottom in small grey: 'Source: International Energy Agency, 2026'. Editorial data journalism style.
A minimalist fashion editorial shot: a female model, early 20s, wearing an oversized cream linen blazer and wide-leg trousers, standing against a pure white studio wall. Her posture is relaxed, one hand in a pocket. The lighting is entirely flat and even — no shadows, no gradients. The composition is centered. The image has the stillness of a Jil Sander campaign. Photorealistic.
A wet New York City street at 11 PM: yellow taxi cabs reflected in the rain-soaked pavement, steam rising from a manhole cover, pedestrians blurred at 1/30s as they cross an intersection. Shot as if on a full-frame camera at ISO 3200, slight digital grain, colour palette of amber street lamps against deep blue night sky. Photorealistic.
A breakfast flat-lay from directly overhead: a stack of three ricotta pancakes on a white ceramic plate, a small ramekin of maple syrup, fresh blueberries scattered around, a sprig of mint on top. Light source from top-left — soft, diffused, slightly warm. The surface is light grey linen. Clean, minimal, food editorial style. Photorealistic.
A documentary portrait of a fisherman, mid-60s, sitting on the edge of a wooden pier. He looks directly at the camera, weathered hands resting on his knees, wearing a faded blue windbreaker. The harbour is behind him, boats visible at low tide. Overcast flat light, no fill. Shot on a 35mm lens. No posed quality — it should feel observed, not directed. Photorealistic.
A printed restaurant menu card on a marble table, photographed from above. The menu reads: 'TONIGHT'S MENU' at the top in large classical serif, then three sections — 'Starters', 'Mains', 'Desserts' — each with two items listed in smaller body text. The card is cream paper with dark ink. A candle in the background provides warm blur. The text is fully legible. Fine dining aesthetic.
Exterior of a contemporary art museum building at dusk: angular white concrete volumes with deep-set windows, surrounded by a reflecting pool. The building's interior is lit warm amber, glowing through the windows against the blue dusk sky. Two small figures walk along the entrance path, providing scale. Shot from across the reflecting pool. Architectural photography.
A cosmetics product flat-lay: three skincare items — a glass serum bottle, a ceramic moisturiser pot, and a pump bottle — arranged on a pale pink marble slab with scattered dried white flowers. Overhead shot from directly above, perfectly even lighting with no harsh shadows. The product labels are clearly readable. Beauty editorial style. Photorealistic.
Macro photograph of a single tropical monstera leaf, shot with a 100mm macro lens. A water droplet sits on the surface, perfectly round, reflecting the canopy above in a tiny sphere. The leaf's veins are rendered in sharp detail. The background is other leaves in soft green bokeh. Overcast diffused light. Photorealistic botanical photography.
A luxury residential kitchen for a real estate listing: marble worktops, navy-painted cabinetry, brass fixtures, a large island with bar seating for four. Pendant lights hang above the island. Large windows look onto a garden. Natural morning light fills the room evenly. No clutter on surfaces — show-home presentation. Wide-angle lens, architectural interior photography. Photorealistic.
A music event poster, portrait orientation, dark background. The headline text reads 'NIGHT FOREST FESTIVAL' in large condensed sans-serif, white, all caps, centred at the top. Below: the dates '21–23 August 2026' in a smaller weight. The background image shows a forest silhouette against a full moon in cool blue tones. At the bottom: 'Tickets at nightforestfest.com' in caption size. The text is fully legible.
A candid portrait of a child aged 5, laughing, outdoors in a garden on a summer afternoon. She is in mid-movement, hair caught in motion. The light is backlit golden hour — the sun behind her creates a soft rim, her face lit by ambient fill from the blue sky in front. Shot with a 50mm lens at f/2, background blurred. Photorealistic family photography.
A clean tech product shot: a slim open laptop on a white oak desk in a light-filled home office. The screen shows a code editor with dark theme and colourful syntax highlighting. A ceramic mug of coffee is to the right, slightly blurred. The desk surface has only these two objects. Natural window light from the left. Flat lay style, shot from above at 45 degrees. Photorealistic.
A square social media image for Instagram. Background: a soft gradient from warm peach to light rose. Centred text in large serif typeface: 'Do the work no one sees. Reap the results everyone notices.' In smaller text below: '@productivitydaily'. The text is the focal point — no photographs or illustrations, typography only. The words are fully legible, elegantly typeset. Clean lifestyle brand aesthetic.
How Microsoft's MAI Image 2 compares to the current top-ranked image generators:
| Model | Photorealism | Arena Rank | Best For |
|---|---|---|---|
| GPT Image 2 (OpenAI) | Excellent | #1 (score: 546) | Product photography, UI mockups, versatility |
| MAI Image 2 (Microsoft) ★ | Best natural lighting | #3 globally | Portraits, documentary, natural light photography |
| Nano Banana Pro (Google) | High — 4K native | Top 5 | Typography, 4K editorial, multi-reference composition |
| Imagen 4 (Google) | High | Top 5 | Product photography, architectural visualization |
| MidJourney v7 | Stylized | — | Artistic, conceptual, stylized illustration |
★ MAI Image 2 leads specifically on portrait photorealism and natural lighting. All outputs include C2PA Content Credentials (AI watermark) by default.
The MAI Image 2 prompt generator on this page provides 20 free, copy-paste prompts for MAI-Image-2 — Microsoft's in-house AI image generation model, developed by Microsoft's superintelligence team and released March 19, 2026. MAI Image 2 ranks #3 on the Arena.ai leaderboard for text-to-image generation (May 2026) and is the default image model powering Bing Image Creator, Microsoft Copilot, and the MAI Playground.
MAI Image 2 (formally MAI-Image-2) is Microsoft's first fully in-house AI image generation model, developed by Microsoft's superintelligence team and released on March 18, 2026. Prior to this, Microsoft's image products ran on third-party models (DALL-E 3, then FLUX.1). MAI Image 2 is Microsoft's own foundation model. It is distributed across Microsoft Copilot, Bing Image Creator, Microsoft Designer, and is available directly via the MAI Playground at microsoft.ai. It ranks #3 globally on the Arena.ai leaderboard and includes a lightweight variant, MAI-Image-2-Efficient, for faster and cheaper API inference.
MAI Image 2 responds best to descriptive, specific prompts that name the lighting conditions, camera settings, and mood — rather than keyword lists. Three things improve output quality significantly: (1) Specify the time of day and light source — 'golden hour sunlight filtering through trees' outperforms 'good lighting'; (2) For portraits, describe the expression, posture, and emotional quality, not just appearance; (3) For text-in-image, quote the exact text and describe its typeface style, size, and colour. MAI Image 2 excels at photorealistic natural lighting, so prompts that describe real-world lighting scenarios (window light, overcast sky, studio softbox) produce the strongest results.
As of May 2026: GPT Image 2 (OpenAI) leads the Arena leaderboard overall with the strongest product photography and UI mockup output. Nano Banana Pro (Google) leads on in-image text rendering accuracy and native 4K output. MAI Image 2 leads on photorealistic portrait quality, natural skin tones, and authentic environmental lighting — it produces the most convincingly 'real' photographs of the three. MAI Image 2 also adds C2PA-compliant watermarks automatically (Content Credentials standard), which is unique among major models and important for editorial and commercial use cases requiring content provenance.
MAI Image 2 is accessible in three ways: (1) Bing Image Creator — free with a Microsoft account, limited daily generations; (2) Microsoft Copilot — integrated into the Copilot sidebar on Windows 11 and Copilot.microsoft.com; (3) MAI Playground at microsoft.ai — direct access with more configuration options. The higher-fidelity MAI Image 2 model (non-Efficient variant) is also available via the Azure AI Foundry API for developers, billed per generation. Copilot Pro subscribers receive faster generation and priority queue access.
MAI-Image-2-Efficient is a smaller, faster, and cheaper variant of MAI Image 2 released alongside the main model. It delivers comparable creative quality at 41% lower inference cost and significantly faster generation times — making it practical for high-volume workflows, batch generation, and experimentation. The difference in visual quality is most noticeable in fine-grained detail (skin texture, complex backgrounds) where the full MAI Image 2 model produces sharper, more photorealistic output. For social media, concept generation, and most commercial use cases, MAI-Image-2-Efficient is the recommended starting point.
OpenAI's #1 Arena-ranked image model — product photos and UI mockups
Google's Gemini 3 Pro image model — 4K, best text rendering, 14 reference images
Google's photorealism specialist — product photography and architecture
xAI FLUX.1-powered — portraits, fantasy art, architecture
Professional MidJourney prompts for stylized AI art
Generate prompts for any AI image model — aesthetic, fantasy, photography