Skip to content
CHICKENPIE.Start a project →
The prompt book

The Zero-Adjective Prompt: How Material Vocabulary Beats Aesthetic Labels in Gemini Imagen 3

Stop stacking 'moody,' 'atmospheric,' 'editorial' onto your prompts. One ceramics term — ash rim marks from wood-firing — is doing more than all of them combined. Here's why.

MADE WITH / Gemini Imagen 3 (gemini-3-pro-image-preview)

Read the paper ↓Get the prompt ↓
Gemini Imagen 3 output: terra cotta ceramic bowl with ash marks on rough concrete surface, dried eucalyptus branch leaning against it, pampas grass stem, indirect window light from upper left
A PAPER FROM THE DESK
THE DETAILS
At a glance
CategoryThe prompt book
PublishedJun 04, 2026
Read time7 min
Covered here
  • AI Tools
  • Design
  • Football
THE PROMPT / Gemini Imagen 3 (gemini-3-pro-image-preview)
Ceramic bowl with matte terra cotta glaze, ash rim marks from wood-firing, resting on rough hand-cast concrete slab surface, dried eucalyptus branch leaning against the left side of the bowl, single stem of dried pampas grass crossing behind, indirect window light from upper left, soft shadow gradient across concrete, camera positioned slightly above and to the right, Canon EOS R5, 85mm f/2.0, ISO 200
Ceramic bowl with matte terra cotta glaze, ash rim marks from wood-firing, resting on rough hand-cast concrete slab surface, dried eucalyptus branch leaning against the left side of the bowl, single stem of dried pampas grass crossing behind, indirect window light from upper left, soft shadow gradient across concrete, camera positioned slightly above and to the right, Canon EOS R5, 85mm f/2.0, ISO 200

Most Gemini Imagen prompts have the same problem: too many aesthetic labels. "Moody," "atmospheric," "editorial," "golden hour." Stack enough of those together and you get the same output every time — a specific kind of soft-focus Instagram aesthetic that's recognizable and ultimately generic.

This prompt has none of that. What it has instead is a half-word change that turns a generic ceramic bowl into something that looks like it came out of an anagama kiln in Bizen. And no — that's not a flex. It's just how material vocabulary works on Gemini Imagen 3.

Why It Works — Five Clauses

"matte terra cotta glaze, ash rim marks from wood-firing"

Two clauses doing two distinct jobs. "Matte terra cotta glaze" activates a specific texture cluster — earthy, slightly chalky surface with irregular depth in the color. "Ash rim marks from wood-firing" is the critical addition: this phrase references a real ceramics technique (anagama firing, where wood ash settles on the shoulder and rim during the firing process and leaves natural deposits). Gemini Imagen 3 has been trained on enough studio photography of artisan ceramics that this phrase pulls toward a specific visual signature — speckled, slightly rough rim deposits on an otherwise smooth glaze.

Without the "ash rim marks" qualifier, you get generic ceramic. With it, you get something that reads as handmade.

"rough hand-cast concrete slab surface"

"Concrete" alone in a prompt produces smooth poured concrete — commercial, uniform, grey. "Hand-cast" is what changes this. Hand-cast concrete is made by hand-packing concrete mix into a form without industrial vibration equipment — the result has visible air pockets, uneven aggregate distribution, and slight irregularity. "Hand-cast" vs "poured" vs "polished" are meaningfully different inputs that produce meaningfully different surfaces.

"dried eucalyptus branch leaning against the left side"

Two decisions here: the species and the position. "Dried" matters because it shifts the eucalyptus from green-and-waxy to silver-grey and papery — a completely different color temperature than fresh botanicals. "Leaning against the left side of the bowl" is a physical staging instruction. Gemini interprets spatial relationships better when they're described physically rather than compositionally. "Leaning against" is better than "to the left of" because it defines a gravitational relationship — there's a surface for the branch to lean against and a direction the lean would naturally fall.

"indirect window light from upper left"

The compass-direction plus quality combination. "Natural light" produces inconsistent results — Gemini interprets it differently every run. "Indirect window light from upper left" constrains the model's light interpretation to no direct sun, a diffused source above and to the left of frame, soft shadow gradients. "Upper" is doing real work — it specifies that the light source is elevated, which changes shadow direction and reduces hot spots on curved surfaces. Most still life product photography in the model's training data was shot under exactly this condition: north-facing studio window or skylights.

"Canon EOS R5, 85mm f/2.0, ISO 200"

This is a technical vocabulary instruction for color science and depth. "Canon EOS R5" carries embedded color science associations — Canon color rendering tends toward warm tones and slightly elevated reds. "85mm f/2.0" tells the model to produce slight background compression and shallow depth of field without full bokeh blur. "ISO 200" signals low-noise, high-detail rendering. Together these three specs are shorthand for "a clean, warm, slightly compressed product photograph with one sharp plane of focus." That's more precise than any aesthetic label.

3–5 Variations

Variation 1 — Swap the ceramic for glass

Replace "ceramic bowl with matte terra cotta glaze, ash rim marks from wood-firing" with "small hand-blown glass vessel with inclusion of air bubbles and slight amber tint, walls 4mm thick." Glass completely changes the light interaction — refractions, transparency, and caustics replace matte absorption. The concrete and botanical instructions hold; the camera spec holds. What changes: instead of surface texture, you're asking the model to handle transmission and refraction.

Variation 2 — Add water

Add "small amount of water in bowl, surface slightly disturbed by the eucalyptus branch stem" after the bowl description. Water in AI-generated still life is a reliable test of whether a model can handle reflections within a composition. Gemini Imagen 3 handles this better than most models — but you need to specify the water quantity and the disturbance source or you'll get a perfectly still water mirror that doesn't read naturally.

Variation 3 — Hard morning light

Replace "indirect window light from upper left" with "direct morning sunlight through east-facing window, hard shadows with sharp edges, high contrast between lit and shadow surfaces." This tests whether the no-aesthetic-labels approach holds under dramatic lighting. Hard direct light produces completely different shadow geometry and requires Gemini to commit to high-contrast rendering rather than soft gradients.

Variation 4 — Scale up to a full surface composition

Add "three additional objects visible at varying distances from camera: a ceramic mortar and pestle, a stack of three linen coasters, an empty glass tea vessel" to the subject description. Multiple objects at different focal distances test the depth-of-field spec directly — at f/2.0, nearest objects should be sharp and background objects progressively blur. If Gemini defaults to rendering everything in focus, add "consistent shallow depth of field, only foreground object in sharp focus."

Variation 5 — Remove the camera spec entirely

Drop the Canon/85mm/ISO clause. Run the prompt with only material and lighting descriptions. This control test isolates exactly what the camera spec is contributing — you'll see a shift toward slightly flatter rendering and less depth compression. If you don't want camera-specific color science, remove the spec; if you want it, keep it.

Model Compatibility

Gemini Imagen 3: Excellent. Best handling of material specificity. The camera spec lands consistently — Canon color science renders warmly, the 85mm compression is visible. Ash rim marks render correctly on first or second run.

Midjourney v6.1: Very Good. Add --style raw to prevent the model from adding unsolicited atmospheric haze. Replace "Canon EOS R5" with just "85mm equivalent, f/2.0" — Midjourney doesn't need brand-specific camera references.

DALL-E 3: Good. Material descriptions land well. The camera spec contributes less to color rendering here — DALL-E 3 color science is more neutral. The eucalyptus leaning position may require a re-run if the first result places it flat on the surface.

Stable Diffusion XL / Flux.1: Moderate. Add negative: "saturated, high contrast, Instagram filter, vignette." Flux.1 Dev handles material vocabulary better than base SDXL. The camera spec at this complexity level needs a LoRA to actually influence rendering.

Failure Modes

1. The generic Pinterest still life

Despite no aesthetic labels in the prompt, Gemini defaults to its understanding of "artisan product photography aesthetic" — warm, soft, slightly oversaturated, with a staged quality. Fix: add "natural color rendition, no added warmth or saturation, technically accurate exposure" after the camera spec.

2. The floating eucalyptus

The branch renders floating in space next to the bowl rather than physically leaning against it. Fix: replace "leaning against the left side of the bowl" with "eucalyptus branch resting at 30-degree angle against the left side of the bowl, bottom of branch in contact with the concrete surface."

3. The opposite problem: hyperrealism

On some runs, the no-aesthetic-label approach works so well that Gemini produces something that looks like a commercial product photograph — perfectly lit, technically correct, and completely soulless. The material specificity is there, but none of the handmade irregularity you were trying to invoke. If this happens, the concrete and glaze descriptions weren't specific enough. Replace "rough hand-cast concrete slab surface" with "hand-cast concrete surface, visible aggregate and air-bubble pockets near the surface, slight color variation across the pour." The irregularity needs to be named, not implied.

Your Turn

Run this with the ceramic bowl, then swap the material description for something else — stoneware, cast iron, tarnished brass, hand-thrown porcelain with celadon glaze. Watch how the surrounding composition adapts to the new material. The concrete and the light don't change; the object does.

If the eucalyptus starts growing from the concrete on your first run, you got the typical failure mode. The fix is in the failure modes section above. Try it.

Drop your results in the comments or tag @chickenpie.co.

Share
MORE FROM THIS LITTLE DESK

A little worth keeping.

Stories, experiments, and things I find along the way. Occasionally, in your inbox.

Get the letter ↗
/feed/gemini-material-still-life-zero-adjective-prompt

Have a project in mind?

Tell us what you are making and what you need. An unfinished idea is welcome too.

Start a project →

Another paper for you.

Chickenpie illustration of Mr. Chicken, small and in the background, glancing at a glowing glass snow-globe dome on a desk containing a miniature jeepney terminal diorama

The Snow-Globe Trick That Makes Gemini Build a Whole Tiny World

A five-clause Gemini prompt that turns any city street into a miniature diorama inside a glass dome, decomposed clause by clause so you can reproduce it.

Open paper
Chickenpie illustration of Mr. Chicken arranging objects into a knolling grid on a desk

The Knolling Flat-Lay Prompt That Makes Gemini Actually Follow a Grid

A knolling flat-lay prompt for Gemini 3 Pro, decomposed clause by clause, with the geometry instructions that actually stop the grid from drifting.

Open paper
Mr. Chicken at a desk pointing at a laptop screen showing a colorful AI-generated image, with reference photos stacked nearby

6 AI Art Tool Tips That Stop Your Marketing Images From Screaming "AI"

6 quick tips for using AI art tools like Nano Banana and Gemini 3 to make marketing visuals that don't scream "AI" — prompting, texture, consistency, and a pre-publish check.

Open paper
Chickenpie mascot Mr. Chicken pointing at a glowing screen showing chaotic color swatches organizing into labeled design system tokens

Stop Hardcoding Colors: Build a Real Design System with Figma Variables

A hands-on 9-step tutorial: wire a Primitives, Semantic, and Brand variable system in Figma so your whole file updates from one place, light mode and dark mode included.

Open paper
Mr. Chicken at a cluttered home studio desk, working on an AI image generation interface with an uncanny AI-slop draft on one side and a clean composited image on the other

How to Art-Direct AI Images So They Stop Looking Like AI

Nine steps to art-direct AI image generation like a photographer, not a wishlist — light, texture, junk tokens, and the two spots (hands, signage) that always give it away.

Open paper
Chickenpie-style illustration of a crafting desk with a tiny hand-cut paper diorama of a sari-sari store, Mr. Chicken watching from the edge of the frame

The Shadow Gap Trick That Turns Flat Paper Into a Diorama

Flat paper collage vs. a real diorama comes down to one four-word phrase. Here's the exact prompt, decomposed clause by clause.

Open paper

You might also like