---
title: "Pruna P Image — Prompting Guide"
description: "How to write effective prompts for the Pruna P Image text-to-image model."
---

Pruna P Image is a fast text-to-image model. It rewards concrete, visual prompts and largely ignores conversational filler.

## Structure

Lead with the subject, then describe it, then set the scene, then the style. A reliable order:

`[subject] [subject details], [composition / shot], [lighting], [environment], [style / medium], [quality tags]`

## What works

- **Concrete nouns and adjectives** over abstractions — "a weathered brass compass on an oak desk", not "a nice old object".
- **Comma-separated descriptors.** The model reads a prompt as a bag of visual cues; short comma-separated phrases beat long sentences.
- **Name the shot and lens** for photographic looks: "low-angle shot", "85mm portrait", "wide establishing shot".
- **Lighting is high-leverage:** "golden hour", "soft diffused light", "dramatic rim lighting", "studio softbox".
- **Style anchors:** name a medium ("oil painting", "3D render", "analog film photo") or an era — not a living artist.

## What to avoid

- Instructions to the model ("please make", "I want") — describe the image, not the request.
- Negations inside the prompt ("no people"). Put exclusions in the Negative Prompt field instead.
- Contradictions ("realistic cartoon") unless intentional.

## Aspect ratio

Match the aspect ratio to the composition: 16:9 for cinematic/landscape, 9:16 for mobile/portrait, 1:1 for icons and product shots. Composition cues in the prompt ("wide panorama", "tall waterfall") should agree with it.