How to Prompt Nano Banana 2
A complete guide to writing prompts for Nano Banana 2, covering the six-part brief, camera, lighting, style, materials, text, references, image editing, and real-time images.
Nano Banana 2 is a reasoning image model. Before it renders, it plans. That means your prompt is not a list of keywords. It is a brief.
A good brief tells the model what is in the frame, where it is, how it is framed, how it is lit, and what it should feel like. The better the brief, the less the model has to guess.
This guide will show you how to write that brief. It is organized around the six parts every prompt needs, the camera and lighting controls, and the special cases that make Nano Banana 2 different from other image models.
What You Can Make
- Fashion and editorial portraits
- Product and beauty photography
- Marketing images, ads, and posters
- Packaging and labels with text
- Infographics and exploded views
- Storyboards and multi-panel layouts
- Reference-driven consistency
- Real-time, web-grounded visuals
The Six Parts of Every Prompt
Every useful prompt has six parts. You can write them in one paragraph or one line at a time, but the parts stay the same.
- Subject - who or what is in the frame
- Setting / Scene - where it is and what surrounds it
- Composition - shot type, angle, placement, focal length
- Style / Mood / Color - visual medium, era, palette, emotional feel
- Lighting - source, direction, hardness, color
- Camera / Lens - focal length, aperture, depth of field, camera type
Aspect ratio and resolution are output controls. Put the ratio at the end of the prompt. Set the resolution in your dashboard if the interface supports it.
Bad Prompt vs. Good Prompt
Here is the same idea written two ways.
The Bad Version
A fashion model, realistic, 4K, beautiful, cinematic.If you use this, the model will do its best: a plain model under flat light, no pose, no backdrop, no story. The words "realistic," "4K," and "cinematic" do not tell the model what she is wearing, where she stands, how she is framed, or what the light should do.
The Good Version
A fashion model in a floor-length emerald-green silk dress with a high slit, standing against a weathered concrete wall. Strong golden-hour light from the left, deep shadows on the right. Her hair pulled back, one hand resting on the wall, the other at her waist. 50mm lens, f/1.8, medium shot, model slightly off-center to the right. Editorial fashion photography, rich green and amber palette, subtle film grain. 4:5.Now the model knows the subject, the setting, the composition, the light, and the style. It is the same idea. The second prompt simply fills the six parts the first one left empty.
Bad Prompt Vs Good Prompt
Build a Prompt Step by Step
The fastest way to learn the structure is to watch the good fashion prompt grow. We will add one part at a time.
Step 1 - Subject
A fashion model in a floor-length emerald-green silk dress with a high slit.This is already better than "fashion model." It gives the clothing, material, and a defining detail. But the model still has to guess the setting, framing, style, light, and camera.
Step 2 - Setting / Scene
... standing against a weathered concrete wall. One hand resting on the wall, the other at her waist.The setting is now specific. The concrete wall and the pose give the model a place and a body position.
Step 3 - Composition
... 50mm lens, medium shot, model slightly off-center to the right.Now the image has a camera position. The medium shot shows the dress and the pose. The off-center placement gives the composition tension.
Step 4 - Style / Mood / Color
... editorial fashion photography, rich green and amber palette, subtle film grain.This pulls the model away from a generic fashion render and toward a deliberate, editorial look. The color palette locks the dress, the concrete, and the golden light together.
Step 5 - Lighting
... strong golden-hour light from the left, deep shadows on the right, highlights on the silk.Lighting is what gives the dress shape and mood. The strong side light makes the silk come alive and creates depth on the concrete.
Step 6 - Camera / Lens
... 50mm lens, f/1.8, shallow depth of field, sharp focus on the eyes.The 50mm and shallow depth of field give the image the believable compression of a real fashion lens.
Final prompt
A fashion model in a floor-length emerald-green silk dress with a high slit, standing against a weathered concrete wall. Strong golden-hour light from the left, deep shadows on the right. Her hair pulled back, one hand resting on the wall, the other at her waist. 50mm lens, f/1.8, medium shot, model slightly off-center to the right. Editorial fashion photography, rich green and amber palette, subtle film grain. 4:5.When you read it back, you should be able to picture the image. If you can picture it, the model can too.
Three Ways to Write the Same Prompt
The six parts do not change. Only the way you write them changes. Choose the format that matches the image you are trying to make.
One Paragraph
Use this for fast, single-subject prompts. The whole brief is written as a flowing sentence or paragraph.
When to use it: portraits, product shots, quick ideas, single scenes.
The bad version
A premium watch, realistic, 4K, luxury, cinematic, detailed.What the model does: It will give you a generic silver watch on a plain background with soft light. It will look like a stock product render.
The good version - product
A 42mm brushed steel automatic diver's watch with a black ceramic bezel and tan rubber strap, resting on a weathered wooden sailboat deck at golden hour. Overhead 85mm shot, f/5.6, deep depth of field keeping the deck grain and watch details sharp. Warm amber sunlight from the right, soft shadows, fine salt crystals on the case and buckle. 1970s tool-watch photography, muted teal and amber palette, subtle film grain. 1:1.Why it works: It covers the six parts in one paragraph. The subject is specific. The setting, light, camera, and style are named.
The good version - portrait
A striking fashion model wearing a tailored brown dress, sleek boots, and holding a structured handbag, posing with a confident, statuesque stance, slightly turned. A seamless deep cherry red studio backdrop. Medium-full shot, center-framed. Fashion magazine editorial, shot on medium-format analog film, pronounced grain, high saturation, soft key light from the left. 4:5.Why it works: The subject, the color, the camera, and the medium are locked in.
Layer by Layer
Use this when the environment is as important as the subject. It breaks the image into foreground, midground, and background.
When to use it: complex environments, cinematic scenes, images with clear depth.
The bad version
A city street at night with neon lights, people, skyscrapers, cinematic, 4K.What the model does: It will likely center a person under generic neon, with random signs and a blurry skyline. Nothing will be anchored.
The good version
A busy Bangkok street at night after rain. Foreground: a metal street-food cart with skewers grilling over charcoal, smoke rising, a vendor in a red apron seen from behind, turning the skewers. Midground: one customer with their back to the camera, holding a steaming bowl. Background: blurred figures walking past neon signs in red, pink, and blue, their glow reflecting on the wet pavement. 35mm, f/2, Cinestill 800T night photography, soft grain, warm neon highlights, deep shadows. 16:9.Why it works: It tells the model what belongs in each depth layer. The foreground, midground, and background are named.
Section by Section
Use this for complex images where each part needs its own line. It makes the prompt easy to edit later.
When to use it: complex scenes, technical shots, prompts you will iterate.
Example
Subject: A sun-kissed blonde girl standing ankle-deep in crystal-clear turquoise water.
Setting: A beach with dramatic limestone cliffs and a flawless blue sky.
Wardrobe: Oversized white T-shirt knotted at the waist, black bikini bottom, sheer rust-orange sarong, golden hoop earrings, retro sunglasses, snakeskin-textured crossbody bag, messy bun.
Camera: 50mm lens, f/2.0, face in sharp focus.
Lighting: Golden tropical daylight, slightly backlit for glow on hair.
Mood: Relaxed, warm, luxurious escape.
Style: Vacation-core, Jacquemus-inspired, 90s casual glamour.
Quality: Sharp detail, fashion campaign finish.Why it works: Each part of the brief is a separate line. The model can reason through the subject, wardrobe, light, and style without mixing them.
Camera and Shots
Nano Banana 2 understands camera language. Use it in your prompts.
Shot types
- Extreme close-up - a single detail: an eye, a watch dial, a texture
- Close-up - head and shoulders, a product on its own
- Medium shot - waist up, good for fashion and portraits
- Full shot - the whole subject, head to toe
- Wide shot - subject plus the environment
- Aerial / overhead - looking straight down, good for food and products
- Low angle - looking up, gives power and scale
- High angle - looking down, makes the subject feel smaller
Focal length and depth of field
- 24mm - wide, distorted, good for action and architecture
- 35mm - natural, street photography look
- 50mm - close to human eye, versatile for portraits and product
- 85mm - portrait, shallow depth of field, soft background
- 100mm macro - extreme detail, product and food
Use f/1.8 or f/2 for shallow depth of field and soft background. Use f/8 or f/11 to keep everything sharp. Use f/4 to f/5.6 for product shots with controlled focus.
Example - action with wide angle
A skateboarder grinding a metal rail in a concrete bowl, backlit by harsh late-afternoon sun, face hidden in shadow. Low angle, 24mm lens, distorted perspective, deep shadows, warm dust in the air. 1990s skate photography, 16:9.Example - macro product
A chef's hands placing microgreens on a white plate. 100mm macro, f/2.8, natural window light, shallow depth of field. 4:5.Example - wide environmental portrait
A motorcycle mechanic seen from behind, leaning over a disassembled engine on a greasy workbench in a cluttered garage. Tools hang on a pegboard behind him, a vintage motorcycle frame against the wall. Wide shot, 35mm, f/4, deep depth of field. Natural window light from a dirty window, muted colors, 1970s documentary photography, 16:9.Lighting: Source, Direction, and Quality
Lighting is what separates a flat image from a dimensional one. Tell the model where the light comes from, how hard or soft it is, and what color it is.
Light source
- Natural window light - soft, directional, common for product and portraits
- Golden hour sun - warm, low, long shadows
- Overhead tungsten work light - warm, harsh, creates deep shadows
- Three-point softbox setup - even, clean, common in product photography
- Single harsh side light - dramatic, high contrast
- Neon signs - colored, atmospheric, cinematic
Light direction
- Front light - even, flat, good for beauty products
- Side light - creates texture and depth
- Backlight - creates a rim or silhouette
- Top light - dramatic, sculptural shadows
- Underlight - unnatural, horror or thriller mood
Light quality
- Soft light - gentle shadows, good for beauty and products
- Hard light - sharp shadows, good for drama and texture
- Diffused light - no harsh shadows, overcast or studio softbox
Example - product with three-point light
A matte black cylindrical perfume bottle with a brushed gold cap on a slate surface. Three-point softbox setup, soft even light, subtle reflections, clean shadows. Clean studio product photography. 1:1.Example - dramatic chiaroscuro
A single red wine bottle with a blank dark label on a dark oak table. One harsh side light, deep shadows, high contrast, chiaroscuro. Dramatic still life. 1:1.Example - golden hour backlit
A picnic blanket on a meadow with a wicker basket and a clear glass bottle of lemonade. Golden hour backlighting, long shadows, warm and relaxed. 35mm, f/4, deep depth of field. 1970s summer photography, soft film grain. 16:9.Style, Mood, and Color
The visual language of the image. Use named references the model can understand.
Film stocks and eras
1980s color film, slightly grainy- nostalgic, warmKodak Portra 400- skin tones, soft grainFujifilm Velvia- saturated, punchy colorsDisposable camera, direct flash- raw, lo-fiMedium-format analog film- editorial, grain
Color palettes
Muted teal and amber- moody, cinematicWarm brown and teal- cozy, naturalRich green and amber- fashion, sunsetBlack and gold- luxury productPastel pink and cream- soft, dessert, beauty
Mood words
Use mood words that describe a feeling, not a quality: quiet, energetic, romantic, tense, carefree, mysterious. Avoid realistic, 8K, masterpiece on their own.
Example - 1980s film
A couple dancing in a neon-lit ballroom, the woman's red dress mid-twirl, the man in a dark suit, both faces partially hidden by the turn. 1980s color film, slightly grainy, warm highlights, cool shadows, romantic and energetic. 4:5.Example - moody cinematic
A foggy pier at dawn, a lone figure in a long coat walking toward the water, a small fishing boat in the distance. Wide shot, 35mm, f/4, deep depth of field. Muted teal and amber palette, soft mist, cinematic dawn light. 16:9.Materials and Texture
Do not ask for a jacket. Ask for a navy blue tweed blazer with visible weave. Do not ask for a mug. Ask for a matte ceramic coffee mug with a hand-painted blue rim. The model needs the material to make the object feel real.
Example - fabric
A navy blue tweed blazer on a wooden hanger, visible weave texture, soft window light from the left. 1:1.Example - ceramic
A matte ceramic coffee mug with a hand-painted blue rim, on a wrinkled linen napkin, soft morning light. 1:1.Example - metal and patina
A brass desk lamp with a green patina, on a walnut table, soft side light, industrial still life. 1:1.Text and Typography
Nano Banana 2 can render sharp, legible text. Follow these rules:
- Put exact words in quotation marks.
- Name the font or style.
bold white sans-serif,flowing brush script,small monospaced font. - Describe the layout. Where does the text sit and what is behind it?
- For long copy, design the text first, then the image.
Product label
A matte green shampoo bottle on a white marble shower shelf. Three lines of text on the bottle: 'PURE' in a thin, minimalist sans-serif, 'BOTANICAL' in a flowing brush script, '500 ML' in a small, monospaced font. Soft natural light, 1:1.Poster
A typographic poster for a summer jazz festival. Solid deep indigo background. The word 'LIVE' in large, bold white serif letters fills the center. The text acts as a cut-out window. A saxophone and trumpet are visible inside the letterforms. 2:3.Storefront
A weathered Parisian bakery storefront at 7am. A hand-painted wooden sign reads "GOLDEN CRUST" in gold Art Nouveau serif lettering. A smaller window decal reads "FRESH DAILY". Wicker baskets of baguettes in the window, warm morning light on cobblestones. 35mm, f/2.8. 4:5.Reference Images
You can use up to 14 reference images in one prompt. Tell the model exactly what the reference is for and what must stay the same.
Product in a new environment
A high-end luxury fashion campaign featuring the exact handbag from the reference image, kept identical in design, warm caramel-brown color, texture, stitching, gold-tone hardware, handles, zipper details, proportions, and overall shape. A fashion model carries the handbag as the main focus, wearing a camel wool coat, in a premium hotel lobby with marble floors and tall arched windows. Soft cinematic lighting, shallow depth of field, focus on the handbag, subtle background bokeh. Editorial Vogue-style aesthetic. The handbag must stay identical to the reference: keep its original design, color, shape, and hardware unchanged.Why it works: It names the reference, lists the details to preserve, and then builds the new scene around the unchanged product.
Face consistency
Cinematic 3:4 vertical portrait using the uploaded reference for 100% face alignment. A confident man in a cream double-breasted linen suit and tortoiseshell sunglasses pours cocktails behind a wooden bar at golden hour. Warm, cinematic key light from the left. Leica SL2-S 85mm f/1.4, f/1.8, shallow depth of field, lomography grain, golden diffusion.Why it works: It says the reference is for 100% face alignment, then builds the new scene around that fixed face.
Image Editing with Words
Editing is different from generating. You already have a base image. Your prompt must say what changes and what stays the same.
Remove an element
Remove the price tag from the product shot. Keep the bottle, shadows, and background exactly the same.Style transfer
Transform this photo of a modern city street into a 1970s pulp-noir illustration. Keep the buildings, cars, and signs unchanged. Apply the pulp-noir style, muted warm color palette, and rough newsprint texture.Add an element
Add a small potted fern to this empty corner. Match the warm afternoon light, the oak floor texture, and the muted earth tones.Real-Time Web-Grounded Images
Nano Banana 2 can search the web and use live data in the image.
Live weather app
Search for the current weather and time in Tokyo, then use that data. If it is raining, show a person with a transparent umbrella crossing Shibuya. If it is sunny, show a picnic in Yoyogi Park. Visualize this as a miniature city-in-a-cup inside a clean smartphone weather-app interface.Aspect Ratios and Resolution
Nano Banana 2 supports 1:1, 3:2, 2:3, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9, and extra tall/wide ratios like 1:4, 4:1, 1:8, and 8:1. Resolutions are 0.5K, 1K, 2K, and 4K.
Always add the ratio at the end of the prompt. The model respects it.
Why it works: The scene is specific, and the output request is explicit. The model can render the same content in different crops.
Choosing the Thinking Mode (MINIMAL vs. HIGH)
Nano Banana 2 has two thinking modes:
- MINIMAL - fast, good for simple scenes and single subjects.
- HIGH - slower, better for complex scenes with many objects, text labels, or reference images.
Use HIGH for: multi-panel storyboards, exploded infographics, counted objects, multiple references, dense text.
Use MINIMAL for: simple portraits, single product shots, quick variations, scenes with one subject and one setting.
Prompts and their results using Nano Banana 2
Typography and Posters
Coloring page with text
A black and white line drawing features the text "DOING MY BEST RESULTS MAY VARY" in large, bold, bubble-style lettering. The text is surrounded by various cute and whimsical elements, including smiling stars, a cat face, clouds, daisies, hearts, cupcakes, a boba tea, and sparkly starbursts. The overall composition is dense and playful, designed as a coloring page.Why it works: The text is quoted and the style of the lettering is described. The decorative elements are listed, giving the model a clear layout.
Movie poster with title and tagline
Cinematic movie poster for a mysterious fantasy science-fiction film titled "THE LAST LIGHTKEEPER". A solitary lighthouse keeper stands on a massive cliff above a dark, stormy ocean at night, holding an old brass lantern. Behind him rises an enormous weathered lighthouse, its beam cutting through thick clouds and fog. Above the ocean, the stars are disappearing from the night sky one by one, leaving vast darkness between the remaining constellations. A faint cosmic glow beneath the horizon suggests something enormous approaching from beyond the ocean. Wind violently moves his coat, waves crash against the cliffs, distant lightning illuminates the clouds, and the lighthouse beam cuts a dramatic path through the darkness. The keeper feels small against the immense landscape, creating a powerful sense of isolation and scale. Sophisticated theatrical poster composition, strong visual hierarchy, dramatic lighting, deep environmental detail, cinematic color grading, subtle film grain, intriguing mystery. Professional movie-poster typography: large title "THE LAST LIGHTKEEPER" and small tagline "WHEN THE STARS GO DARK, SOMEONE HAS TO KEEP WATCH." Only the title and tagline; no other text, logos, watermarks, or credits. 2:3.Why it works: It is long, but every sentence is doing a job. The poster dimensions, title, tagline, and "no other text" constraint are explicit.
Infographics and Exploded Views
Scientific educational infographic
A detailed scientific educational infographic about a capybara. The layout features a large naturalistic illustration of the capybara in the center with a calm expression. Four modular sections around it: a 3D skeletal diagram, a skull anatomy drawing, a size comparison silhouette next to a human, and a world map showing its habitat. The background is a clean, off-white parchment texture. Bold, clean sans-serif headings: "ANATOMY," "DIET," "HABITAT," and "SIZE." Professional and informative, resembling a page from a natural history encyclopedia. 4:5.Why it works: It names the layout sections, the typography style, and the encyclopedia reference. The model knows how to structure the page.
Exploded milkshake infographic with labels
An exploded vertical infographic composition of a strawberry milkshake. Top to bottom structure: Fresh Strawberry Garnish, Whipped Cream, Strawberry Syrup, Creamy Strawberry Shake, Crushed Ice, Transparent Glass Base. All elements perfectly centered, evenly spaced, and vertically aligned. Strawberry texture, fluffy whipped cream, glossy syrup, thick milkshake consistency, translucent ice, and glass reflections. Soft pastel pink seamless studio background with elegant café lighting and subtle shadows beneath each floating element. DSLR macro beverage photography, clean infographic composition. Add clean minimalist infographic text labels with thin pointer lines: "Strawberry" "Whipped Cream" "Strawberry Syrup" "Milkshake" "Ice" "Glass". 1:1.Why it works: The vertical order, alignment, and labels are explicit. The model knows the structure before it renders the parts.
Product and Packaging
Smartphone ad with brand colors
Commercial advertisement featuring [Insert Product: e.g., a modern sleek smartphone] as the main hero. Surrounding scene in a vibrant 2D/3D hybrid pop illustration style, with stylized character hands reaching in, playful floating geometric shapes, dynamic motion speed lines, and abstract vapor clouds. The entire environment, highlights, and graphic accents derive from the product's primary and secondary brand colors. Clean vector aesthetic, bold commercial illustration. 4:5.Why it works: It tells the model to build the whole scene from the product's brand colors. The placeholders are for the user to fill in.
Fashion and Editorial
Vintage Americana editorial
A dreamy editorial photograph of a young woman with long chestnut-brown hair spilling out of the open driver’s door of a vintage cream-colored pickup truck, reclining across the front bench seat. She wears knee-high white socks, an oversized white sleeveless tank top, and tiny faded charcoal athletic shorts. Her eyes are peacefully closed, one hand resting on her stomach, the other dangling toward the worn tan leather seat. The truck is parked in a forest clearing surrounded by tall oak trees and late-summer grass. Natural golden daylight filters through the leaves, casting soft dappled shadows across the interior. Sofia Coppola atmosphere, quiet countryside, dreamy editorial lifestyle photography. Shot on a Contax 645 with Kodak Portra 400, 80mm, f/2.8, subtle film grain, soft pastel color grading, creamy highlights, analog imperfections. 4:5.Why it works: The pose, the film stock, and the named director's mood give the image a clear style. It is not a generic portrait because the references are specific.
High-fashion beach portrait
Cinematic high-fashion beach portrait of a blonde girl standing ankle-deep in crystal-clear turquoise water, with dramatic limestone cliffs in the background under a clear blue sky. She wears an oversized white T-shirt knotted at the waist over a black bikini bottom, a sheer rust-orange sarong tied around her hips, and a mini snakeskin-textured crossbody bag in burnt orange over one shoulder. Golden hoop earrings and narrow retro sunglasses. Her blonde hair is pulled into a messy bun, a few strands catching the light. 50mm, f/2.0, shallow depth of field. Golden tropical daylight, slightly backlit. Relaxed, warm, luxurious escape. Vacation-core, Jacquemus-inspired, 90s casual glamour. 4:5.Why it works: The clothing, the brand reference, and the camera/light sections give the image a clear fashion direction.
UGC and Social Media
Candid street selfie
A candid, tight close-up street selfie surrounded by white-cream hydrangeas. The face is in three-quarter view, chin slightly lowered, head turned to the side, gaze shifting sideways with a thoughtful, sensual expression. Strong motion blur on the hair and flowers, the face in focus. Voluminous curls fall diagonally across the foreground, partially covering one shoulder. Light oversized light-wash denim jacket, bare shoulders visible along the lower edge of the frame. White-cream hydrangea blossoms and large dark-green leaves densely frame the figure, darkened and muted to create depth. Direct, hard Canon flash creates bright, high-contrast light on the face and hair, highlights on the hydrangeas, and deep shadows around. Glowing skin with visible pores, deep contouring, winged eyeliner, long fluffy lashes, plump lips with nude-pink gloss. Natural skin texture, strong grain. Short square nails with a nude base. 9:16.Why it works: The constraints are first. The face, the flash, the pose, and the "no plastic look" rule give the image a specific UGC feel.
Storyboards and Multi-Panel
8-panel matcha latte storyboard
Pixar-style 3D animated 8-panel horizontal storyboard, bright cozy modern home kitchen with morning sunlight through windows, fresh greens and whites. Top title: "HOMEMADE MATCHA LATTE" in warm green and gold. Panel 1 "Fresh Ingredients" with matcha powder, milk, bamboo whisk, bowl, and ice on a wooden counter. Panel 2 "Sifting Matcha" with hands sifting bright green matcha into a bowl. Panel 3 "Whisking to Froth" with hands whisking matcha into smooth foam. Panel 4 "Steaming & Frothing Milk" with milk steaming in a small pitcher. Panel 5 "Pouring Matcha" with thick green matcha poured over ice. Panel 6 "Adding Milk" with frothed milk layered over matcha. Panel 7 "Final Latte Art" with a finished iced matcha latte, mint garnish, condensation. Panel 8 "Ready to Enjoy" with hands holding the latte. Creamy textures, vibrant matcha green, soft bokeh background, Pixar charm, clean composition. Large clear white text on each panel. 16:9.Why it works: Every panel has a number, a label, and a visual. The model can build the structure before it fills the details.
Product commercial storyboard
TITLE: Premium Instant Noodles Product Commercial Storyboard. FORMAT: Single-page premium storyboard; 3:4 portrait; 8 cinematic product scenes. HEADER: Modern bold typography; duration 20 seconds, style cinematic food advertisement, product instant noodles, audio cooking ASMR + energetic beat; warm yellow and red accents. STORYBOARD: 1. Noodle packet standing upright with dramatic lighting; 2. Packet opening and noodles sliding into a bowl; 3. Boiling water pouring over noodles; 4. Noodles cooking with rising steam; 5. Seasoning powder being added; 6. Chopsticks lifting cooked noodles; 7. Extreme macro of glossy noodles with toppings; 8. Final hero bowl beside the original packet. EVERY PANEL: scene number, duration badge, visual, action, camera direction, product detail. CAMERA: macro food photography, overhead composition, steam close-up, noodle pull shot, dramatic product push-in, cinematic hero framing. STYLE: detailed food commercial, visible steam, noodle texture, glossy broth, vibrant ingredients, professional studio lighting, premium packaging. 3:4.Why it works: The title, format, and panel structure come first. The model knows the commercial format before it renders the food.
Common Mistakes
| Mistake | What the model does | Fix |
|---|---|---|
| Vague subject | Picks an average version of the noun | Name the age, material, clothing, and defining traits |
| Generic idea | Falls back on a polished but forgettable stock image | Combine specific subjects, locations, and styles; avoid the first idea |
| Keyword soup | Averages the words into a generic look | Write a brief in full sentences, one part at a time |
| Missing composition | Centers the subject, flat framing | Add shot type, angle, subject placement, focal length |
| Weak style words | Adds generic polish | Use named mediums, film stocks, and eras |
| Missing light | Flat, even lighting | Name the source, direction, and quality of light |
| Negative phrasing | May still draw the thing you said to avoid | Use positive framing: "empty road" instead of "no cars" |
| Long text in one prompt | Text becomes garbled | Keep text short or generate the layout first |
| Forgetting the ratio | Model falls back to a default crop | Add 1:1, 4:5, 16:9, etc. at the end |
| Not using HIGH for complex scenes | Drops objects, misplaces text, ignores references | Use HIGH for infographics, storyboards, multi-reference scenes |
What to Do Next
Start with the six-part brief. Add the aspect ratio at the end. Set the resolution in your dashboard, or add it to the prompt if your interface does not set it. Then generate.
If the first result is not right, change one part at a time. Adjust the light before you rewrite the whole prompt. Change the camera before you blame the model. The more specific your brief, the less the model has to guess.
Last updated on
How to Prompt Grok Imagine Video 1.5
Learn how to prompt Grok Imagine Video 1.5 for image-to-video, text-to-video, reference images, complex scenes, camera control, audio, and video extension.
How to Prompt Gemini Omni Flash 1.1 Video Model
Learn how to prompt Gemini Omni Flash 1.1 for text-to-video, image-to-video, first and last frames, reference images, video editing, video extension, audio, timing, and text.