Nano Banana Pro prompts for food photography respond well to three things: light source direction, surface material, and a single focal point. When you give the model all three, it stops guessing and starts making deliberate composition decisions. The 8 prompts below cover common food photography jobs, from restaurant menu stills to seasonal campaign hero shots. On 8frame, Nano Banana Pro costs 21 credits per image (42 at 4K).
Correction, 2026-09-30: the first version of this page presented these prompts as tested, with generation times and run-by-run results we cannot reproduce, and said Nano Banana Pro is faster and cheaper than Flux (on 8frame, FLUX.2 [pro] is 5 credits and Nano Banana Pro 21). The prompts and the formula stay; the test claims are gone.
TL;DR
- Nano Banana Pro is 21 credits per image on 8frame, 42 at 4K; Nano Banana 2 is 11 at 1K for high-volume menu work
- Include a surface descriptor ("matte black slate," "worn oak board") and a light direction in every prompt for sharper compositional control
- Action shots (pours, drizzles) need the word "motion frozen" or you get blur
- Overhead flat-lays need an explicit lens equivalent ("shot overhead on 50mm equivalent") to prevent the barrel distortion that makes plates look stretched
When to use Nano Banana Pro for food photography
Nano Banana Pro fits food work that needs many consistent images: menu refreshes, recipe blogs, delivery-app dish stills. Lock the surface and light wording, swap only the dish, and a set reads as one shoot. For volume, Nano Banana 2 at 11 credits (1K) is the cheaper sibling; Nano Banana Pro is the one we recommend when words on the image must be spelled right.
Limits worth planning for: busy tablescapes with many distinct objects are harder to control, and packaging label text is safest composited in the edit. For an image quality head-to-head, see the nano banana vs seedream vs flux comparison.
The prompt formula
[Subject] on [surface], [light source and direction], [camera angle and lens], [mood or grade], [any action or texture detail]
Everything before the comma is what the model renders. Everything after shapes how it renders it. Skip any slot and the model makes a default choice, usually the wrong one for your use case.
8 prompts for food photography
1. Plated dish on dark surface
Prompt:
Pan-seared salmon fillet on a matte black slate surface, single soft side light from the left casting long shadow, close 3/4 angle, muted restaurant palette, droplets of herb oil on plate edge, no garnish clutter
What to check: The slate texture and the oil droplets on the plate edge. Glossy liquid is where a food image looks real or doesn't.
2. Overhead flat-lay spread
Prompt:
Overhead flat-lay of a Mediterranean mezze spread, worn oak board, natural window light from the top right, shot overhead on 50mm equivalent to minimize distortion, warm afternoon tone, seven distinct dishes with negative space for text placement on the left third
What to check: That the plates stay round (the 50mm cue is there to prevent stretching), that negative space lands in the left third, and that the seven dishes don't stack.
3. Action shot: pour
Prompt:
Dark maple syrup pouring onto a short stack of fluffy buttermilk pancakes, motion frozen mid-pour, warm studio key light from above right, tight 3/4 angle, shallow depth of field, steam rising from pancakes, deep amber tones
What to check: That the pour is frozen rather than smeared, and that the steam reads as wisps, not haze. If the pour blurs, keep "motion frozen" early in the prompt.
4. Ingredients flat-lay
Prompt:
Flat-lay of fresh pasta-making ingredients: tipo 00 flour mound with well center, three cracked eggs, semolina dusting, wooden rolling pin, on white marble, overhead, diffuse studio light, minimal shadows, clean editorial look, no text
What to check: That the marble stays white and the flour well holds its shape.
5. Restaurant-style 3/4 angle
Prompt:
Grilled dry-aged ribeye on a warm ceramic plate, 3/4 angle at 30 degrees, candlelit restaurant atmosphere, warm amber practical lights in background bokeh, char marks visible, small compound butter pat melting on top, dark moody grade
What to check: That the plate rim stays visible at the 30-degree angle and the char marks look irregular rather than a uniform grid.
6. Beverage hero: cocktail
Prompt:
Negroni in a crystal rocks glass, single large ice cube, orange peel garnish, backlit on a marble bar top, condensation on glass, dark bar background, product photography lighting, 90mm equivalent macro crop, deep crimson liquid
What to check: Translucency in the backlit liquid, the shape of the single ice cube, and condensation that still reads at thumbnail size.
7. Dessert close-up texture
Prompt:
Extreme close-up of a cross-section cut of a dark chocolate lava cake, molten center beginning to flow, fork on left edge of frame, matte black plate, overhead side lighting raking across the surface to reveal texture, chocolate gloss and matte contrast, no background elements
What to check: That the raking light actually shows texture. If the cake looks flat, strengthen the light direction before changing anything else.
8. Seasonal / holiday food scene
Prompt:
Roasted whole turkey on a wooden farmhouse table, Thanksgiving scene, late afternoon golden hour window light from the left, cranberry sauce and roasted vegetables visible in background at soft focus, shallow depth of field focused on breast, warm amber grade, editorial food photography look, no props outside the table setting
What to check: That nothing outside the table setting crept in, and that the background dishes stay recognisable while out of focus.
Common failures
Missing surface material. "On a table" gives the model nothing. "On worn oak" or "on matte black slate" gives it texture behavior, light interaction, and a color relationship. Surface is not optional.
No light direction. The model defaults to flat frontal studio light. For most food categories this looks like a supermarket flyer.
Action shots without "motion frozen." Pours and drizzles default to blur unless you explicitly freeze them. Use "in motion" for flow, "motion frozen" for suspension.
Overcrowded prop lists. Above five distinct elements, the model compresses spatially. Name the hero element first, limit supporting items to three.
Expecting readable label text. No image model renders legible packaging text reliably. Composite it in post.
Step-by-step on 8frame
- Open the 8frame canvas and add an image node with Nano Banana Pro.
- Set the aspect ratio first: 4:5 for Instagram feed, 3:2 for editorial, 1:1 for delivery app tiles. Nano Banana offers all three.
- Paste your prompt using the formula above: subject, surface, light, angle, mood, action detail.
- Run at default settings first. If light direction or surface texture is wrong, add one correction rather than rewriting the full prompt.
- For batch menu work, lock the surface and lighting cues and swap only the dish name across runs.
FAQ
Can Nano Banana Pro handle multiple dishes in one shot?
Yes, but the more distinct dishes you ask for, the harder spacing is to control. Name the hero dish first, give an explicit spatial cue ("negative space on the left third"), and for very large spreads composite several frames in the edit.
Does the model handle steam and condensation accurately?
Ask for them explicitly: "steam rising", "condensation on glass". Shorthand adjectives like "dewy" or "fresh" leave more to chance. Explicit nouns beat adjectives.
How does Nano Banana Pro compare to Flux for food photography?
On price, FLUX.2 [pro] is cheaper on 8frame: 5 credits per image against 21 for Nano Banana Pro. Try the same prompt on both for your dish and compare; a few test images cost less than guessing. Full prices in AI image generator price per image.
Sources
- 8frame prices and aspect ratios: the 8frame tool registry, 2026-09-30
Lock the surface and light, swap the dish. The 8frame canvas is free and unlimited; generation is paid from $19/month.