Beverage photography lives or dies on what is happening inside the glass: ice refraction, foam texture, a swirl, rising bubbles. A “clean” drink reads as plastic. The 10 prompts below pin down beverage state, container material, garnish, and lighting so the model has nothing to invent. Each is ready to paste into Midjourney, Flux, or GPT Image and tuned for cafes, bars, tea shops, and brand campaigns.
TL;DR
- The single biggest fix for “dead” drink images is naming a dynamic state: condensation, steam, ripple, or a pour in motion.
- For translucent subjects (glass, ice, beer foam), pick your model by the surface: Imagen 4 Ultra renders glass and reflections most convincingly, Flux 2 Pro gives camera-accurate optics fast and cheap, and Midjourney V8.1 wins on stylized mood.
- Shoot drinks at 50mm f/2.8–f/4 for full scenes and 100mm macro for foam and ice close-ups. Skip wide angles.
- Copy any of the 10 prompts, swap the drink and garnish, keep the lighting and lens, then iterate.
Which model for which drink (June 2026)
The big three photoreal models handle beverages differently. Glass, ice, and foam are exactly the “translucent material” cases where they diverge most.
| Model | Version (June 2026) | Best for beverages | Notes |
|---|---|---|---|
| Imagen 4 | Imagen 4 Ultra | Glass, ice cubes, wine reflections, hero shots | Highest fidelity on translucent surfaces; ~$0.04–0.08/image via API |
| Flux 2 Pro | Flux 2 Pro | Volume product shots, realistic optics | Camera-accurate depth of field and grain; ~$0.03–0.05/image, ~3s render |
| Midjourney | V8.1 (V7 still default) | Stylized, moody, editorial mood | Add --v 8.1; renders 2K HD without upscale, ~4–5x faster than V6 |
| GPT Image 2 | In ChatGPT + API since Apr 2026 | Quick iterations with overlay text | Strongest in-image text if you need a label or headline |
For most cafe and tea-shop work, start with Flux 2 Pro for speed, then re-run your two favorites on Imagen 4 Ultra for the final hero. Midjourney is the call when you want a magazine-ad feel over literal accuracy. New to these tools? Start with our Midjourney beginner guide.
What a high-quality beverage prompt contains
Six elements do almost all the work:
- Beverage state:
with condensation drops/gentle steam/slight ripple on surface/mid-pour - Container material:
frosted glass/clear crystal/white ceramic— this defines the reflection - Garnish: 1–2 elements (coffee bean / mint leaf / lemon slice). More than two clutters the frame.
- Lighting: coffee = warm single source; cocktails and beer = high-contrast hard light; juice and tea = soft natural light
- Texture words:
crystalline ice,velvet microfoam,golden caramel pour - Lens: 50mm f/2.8–f/4 for scenes; 100mm macro for close-ups
10 copy-ready prompt templates
1. Iced americano counter
Best for: Cafe hero, delivery detail
A tall glass of iced black coffee with condensation drops, ice cubes refracting light, single coffee bean on marble counter, soft north-window light, slight blurred warm wood interior behind, 50mm f/2.8 cafe commercial, --ar 4:52. Latte art macro
Best for: Cafe new-arrival, social hero
A close-up macro of latte art (rosetta) on top of espresso, smooth velvet microfoam, white ceramic cup, slight steam rising, plain warm wood background, single side warm light, 100mm macro f/4, --ar 1:13. Red-wine pour moody
Best for: Wine brand, dinner campaign
A glass of red wine being poured into a crystal glass on a dark wood table, single warm candle-light from camera-left, slight ripple on surface, deep moody luxury commercial, 50mm f/2.8, --ar 4:54. Matcha latte daylight
Best for: Japanese-style cafe, healthy drink brand
A green matcha latte in a small white ceramic cup with a swirl of foam, single matcha leaf garnish, beige linen surface, soft natural daylight from camera-left, lifestyle health-cafe commercial, 50mm f/4, --ar 4:55. Cold beer foam
Best for: Beer brand, pub vibe
A frosted glass mug of cold beer with thick white foam, droplets running down the side, single hard side spotlight, dark moody pub background, deep contrast beer commercial, 50mm f/2.8, --ar 4:56. Fresh orange juice morning
Best for: Juice / healthy drink brand
A glass of fresh orange juice on a sunny breakfast table with sliced oranges, soft golden morning light from camera-left, light blurred kitchen behind, lifestyle morning beverage, 50mm f/4, --ar 4:57. Layered blue cocktail
Best for: Bar / cocktail brand
A cocktail glass with vibrant blue layered drink, citrus peel garnish, single cocktail stirrer, single warm side light from camera-right, dark moody bar background, anamorphic 50mm cocktail commercial, --ar 4:58. Lemon iced tea summer
Best for: Tea brand, summer campaign
A clear glass cup of lemon iced tea with mint leaves and lemon slices, condensation, single soft top light, light pastel backdrop, fresh summery feel, 50mm f/4, --ar 4:59. Held boba street candid
Best for: Boba brand lifestyle
A handheld iced milk tea boba cup, condensation on plastic, half-blurred sunny city street behind, candid lifestyle moment, 35mm f/2.0 documentary feel, --ar 4:510. Cold-brew milk swirl
Best for: Premium coffee brand ad
A glass jar of cold brew coffee with milk swirling in from above, slow-motion cream trail, plain dark grey background, single soft top light, premium coffee brand commercial, 100mm macro f/5.6, --ar 4:5Common mistakes
- “Clean” inside the glass — no ice, no bubbles, no swirl reads as a dead, plastic image
- Asking the model for a logo or text on the cup — it almost always comes out garbled, except in Ideogram v3 or GPT Image 2
- 3+ glasses in one frame — the subject scatters and nothing is the hero
colorful drink— empty filler; the model picks chaotic colors. Name the exact color instead.- 24mm wide-angle — beverages need 50–100mm to hold texture and avoid distortion
How to push results further
- Coffee:
velvet microfoamandcrema layerare the soul of latte texture; without them the surface looks flat - Alcohol: single light + dark backdrop + a reflection-specific word (cocktail glass reflects differently than beer foam)
- Juice and tea: natural light + light backdrop + one fresh garnish
- Street lifestyle: template 9 — 35mm f/2.0 plus a blurred street
- Luxury commercial: templates 3 and 10 — single light, dark backdrop, 100mm macro
FAQ
Q: Why do my drinks look “dry” or flat?
A: There is no dynamic state in the prompt. Add with condensation drops, gentle steam, or slight ripple on surface and the drink comes alive.
Q: AI ice cubes look weird. What fixes it?
A: Write crystalline ice cubes refracting light instead of ice cubes. As of June 2026, Imagen 4 Ultra and Flux 2 Pro handle glass and ice more cleanly than older Stable Diffusion checkpoints; Midjourney V8.1 also improved hard-edge refraction over V6.
Q: Can I get stable latte art?
A: Partially. rosetta latte art on top or heart latte art cues the attempt, but the pattern can still warp. For a perfect rosetta, generate a few and pick, or use a video model (Sora / Veo 3) for a pour clip.
Q: How do I write a layered cocktail?
A: Name the order explicitly, e.g. layered drink: bottom blue, middle clear, top yellow with foam. Vague phrasing makes the model blend the layers.
Q: What gets the clean “Xiaohongshu viral” look?
A: Light backdrop + natural light + clean composition + one fresh garnish. Templates 6 and 8 are built for it. Flux 2 Pro tends to nail this aesthetic in one or two runs.
Related articles
- Product Photography AI Prompts
- Food Photography Image Prompts
- Skincare Product Image Prompts
- Food Photography sub-hub
- All AI Image Prompts
- Asian Cuisine Image Prompts: 10 Regional Dish Templates
Tags: #Food #Product photography #Commercial #Image generation