LarpGPT · Guides
AI image prompt tools, compared
Prompt libraries, prompt builders, in-generator assistants and writing it yourself: what each saves you, what each costs you, and which one teaches you anything.
In this guide5
Four ways to arrive at a prompt
Between an idea and an image there is a sentence, and there are four ways to get to it.
A prompt library gives you texts written by someone else, ready to copy. A prompt builder assembles one from choices you make. An in-generator assistant rewrites your words inside the tool that makes the image. Writing it yourself is the fourth, and it is the only one that makes the next prompt easier.
These differ mostly in where the thinking happens, and therefore in what you keep afterwards.
What actually changes an image
Prompts are not spells. The words that reliably move a result are the concrete ones: subject, framing, lighting, lens, material, and what you do not want. Style names borrowed from living artists change less than people expect and raise questions this page will not pretend do not exist.
A tool that hands you a long prompt full of adjectives has not made you better at this. A tool that shows you which part of the sentence did the work has.
Criteria worth checking
The practical criteria are whether the text is portable, whether you can read and edit it, and whether the tool works with the generator you already pay for.
Portability matters more than it sounds. A prompt you can copy into any generator survives a change of subscription; a prompt locked inside one product does not.
| Criterion | Prompt library | Prompt builder | In-generator assistant | Writing it yourself |
|---|---|---|---|---|
| Text is portable between generators | Yes | Usually | No | Yes |
| You can see and edit the words | Yes | Partly | Sometimes | Yes |
| Teaches you for next time | A little | A little | Rarely | Yes |
| Needs a connection | Not necessarily | Not necessarily | Yes | No |
| Tied to one paid product | No | No | Yes | No |
Where LarpGPT sits
LarpGPT is a library and a starting point: photo prompts to explore, a scene to choose, and text you copy into the image generator you already use.
It deliberately does not generate images. The App Store listing states that no account is required.
What LarpGPT does not do
It does not create images, does not connect to a generator, and does not guarantee any particular result: the same prompt gives different images in different tools, and different images in the same tool.
It also does not decide for you what is appropriate to generate. Rights, likeness and disclosure remain the responsibility of the person who presses the button, whichever tool produced the sentence.
Questions
Do I need a prompt app to get good images?
No. The words that change a result are concrete ones — subject, framing, lighting, lens, material, and what you do not want. A library saves time; it does not hold a secret.
Why does the same prompt give different images?
Image generators are not deterministic, and each one interprets a sentence differently. A prompt is a starting point, not an instruction that guarantees an outcome.
Is it fine to name an artist in a prompt?
Naming living artists raises rights and attribution questions that differ by tool and by country. The responsibility sits with the person generating and publishing the image.
Guides
How to structure an AI photo prompt from scratch
A repeatable order for writing image prompts, what to put in each part, the common mistakes that flatten a result, and how to iterate so that each attempt teaches you something.
Six worked photo prompt examples, and why each line is there
Complete prompts for an architectural interior, a city moment, an object, a landscape, a food shot and a workspace, each broken down sentence by sentence so you can adapt them.
The anatomy of a photo-realistic image prompt
What each part of a prompt does inside an image generator, why one sentence per decision works better than a wall of adjectives, and how to adapt a prompt without breaking the thing that made it work.
Midjourney photorealistic prompts: a practical method
How to write Midjourney prompts that read as photographs: a camera-first structure, the documented parameters that help, and how to iterate.
ChatGPT image prompts for luxury photos that look real
How to prompt ChatGPT for believable luxury and lifestyle images: materials, light, restraint, worked examples, and care with brands and disclosure.
How to write AI prompts for realistic people
Describe believable people in AI images: posture, clothing, skin, hands and light, plus the consent, likeness and disclosure rules that come with it.
Lighting keywords for AI photos: a practical glossary
Golden hour, blue hour, window light, backlight, low key and more: what each lighting term looks like in a photo and how to write it in an AI prompt.
Negative prompts explained: when to use them, and how
What a negative prompt is, which image tools document one, how Midjourney's --no and Stable Diffusion's negative_prompt work, and why to use them sparingly.
Aspect ratios for AI images: how to choose and set them
Square, 3:2, 16:9 or vertical 9:16: how to pick an aspect ratio for an AI image, how composition changes with it, and how Midjourney and OpenAI set it.
How to write AI photo prompts that stay readable
A repeatable prompt structure for subject, light, lens language and constraints, with examples that are easier to refine.