PromptZone - AI Prompts, Guides and Tools for Builders: Allen Zhao

A prompt template for consistent product scenes

Allen Zhao — Sat, 31 Oct 2026 03:01:20 +0000

When an AI product image looks good but the product changes shape between shots, the prompt often mixes fixed facts with creative directions. Separate the two before generating.

A reusable template

Product facts: [object], [material], [color], [distinctive features]. Match the supplied reference and preserve its proportions and visible labels.

Scene: Place it on [surface] in [environment]. Use [soft side light / daylight / another specified light], viewed from [camera angle]. Leave [position] clear for later layout.

Constraints: Do not add text, logos, extra components, or claims. Keep the product fully visible unless the shot is explicitly a close-up.

Test one variable at a time

Begin with a plain background. Once the object is recognizable, vary the surface or lighting while holding the product facts and camera angle constant. This makes it easier to tell whether the changes improve the scene.

Inspect the result

Compare the silhouette, cap or handle, materials, colors, and any label against the authorized source. Small lettering and exact product geometry can be unreliable; add final text separately and review every image before using it commercially. A generated scene is an illustration, not proof of a feature.

Prepared by the Jynno team. Jynno is an AI creative studio for image and video workflows.

A small character brief that leaves room for surprises

Allen Zhao — Sat, 31 Oct 2026 00:24:50 +0000

When I start a character with only a mood or aesthetic, I find it useful to add a few story constraints before asking for visual ideas. Otherwise the result can look polished but tell me very little about who the character is.

Here is a brief you can adapt:

Design an original character for a quiet coastal mystery. They repair old radios and are searching for a missing friend. Give them one recognizable silhouette, a practical outfit suited to wet weather, and one small object that hints at their history. Suggest three visual directions that share these core details, and explain what changes between them. Avoid copying a named character or artist.

I would keep the fixed details short and change just one thing at a time: the silhouette, the palette, or the relationship between the character and their object. That makes it easier to compare variations instead of losing the original idea in a pile of unrelated images.

If you're working from a photo, use one you have permission to use and decide which traits should carry through before generating anything. A reference sheet can then document the choices you actually want to keep.

I'm building Zoivu around this kind of character development workflow. The brief above is an example to adapt, not a claim about a particular model's output. What constraints do you find most useful when a character concept starts feeling generic?