
Loktak phumdi rings seen from Sendra Hill above the dawn mist
GPT Image
Photos, video, thumbnails, logos, work and everyday writing in one library. Each prompt is tested before it is published, and the picture beside it is what it produced.

GPT Image

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo
The newest YouTube and blog thumbnails and social media posts, each shown with the picture it produced.

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image

GPT Image
Works with your photo
AI video studio for musicians and visual artists that turns images, prompts and songs into music-synced video clips.
Platform for building AI phone and chat agents that answer, place and route calls with a chosen language model and voice.
AI headshot generator that turns a set of casual selfies into studio-style professional portraits in under an hour.
AI headshot service for individuals and companies, with one-time packages and team credits managed from an admin dashboard.
AI presentation maker that builds and edits slides directly inside Google Slides and PowerPoint instead of a separate app.
A mobile AI photo editor for selfies and social posts, with one-tap trend effects, an object eraser, enhancement and hair try-on.

CCTV AI photo prompt guide for ChatGPT and Gemini: high corner angle, fisheye bend, timestamp overlay, night vision and six copy-ready security camera looks.

Copy-ready AI hug photo prompts for ChatGPT and Gemini: hug your younger self, a parent abroad or a pet you miss in line art or chalk, each with its result.

Copy-ready birthday AI photo editing prompts for ChatGPT and Gemini: giant age numbers, midnight surprises, chai-glass candles and family tables, each with its result.
Prompts for still images, each one shown with the output it produced and the model it was tested on.
29 categoriesPrompts for video models, with the clip each one produced and the settings it needs.
12 categoriesPrompts for work and everyday life: social media posts, thumbnails, logos, writing, analysis and code, and the replies, messages, notes and plans that come up every week.
32 categoriesThumbnails, reel covers and pictures that stop the scroll.
Product shots, logos and mockups ready for a listing or an ad.
Headshots and work prompts that return something usable.
Festival, wedding, anniversary and family pictures worth sharing.
What visitors are copying most right now.

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo

GPT Image
Works with your photo
Every prompt shows the output it produced, so you are choosing a result rather than guessing from a description.
The model it was tested on, the ratio it expects, and whether it works with a photo of your own. Those three explain almost every case where a prompt does not reproduce.
The copy action hands over Markdown with each part under its own heading, so you can change the wardrobe without touching the lighting.
Strong at following a long instruction literally and at legible text in an image.
OpenAIBuilt for editing an existing image and for holding a face across edits.
GoogleGeneral image generation inside a general assistant, useful for quick iterations.
GoogleThe most art-directed look of the group, at the cost of literal instruction following.
MidjourneyThe picture beside a prompt is what that prompt produced. You are choosing a result, not guessing from a description of one.
The same text on two models is two different pictures. A prompt that does not say which one it was run on cannot be reproduced.
Aspect ratio changes what the model puts in frame, not just where it crops. It belongs where you can read it, not buried in prose.
Some prompts only work with a photograph of your own. That is marked on the card, before you copy anything.
Worth having ready about 6 weeks ahead, so the page is indexed before anyone starts looking.
October or NovemberWorth having ready about 5 weeks ahead, so the page is indexed before anyone starts looking.
October or NovemberWorth having ready about 4 weeks ahead, so the page is indexed before anyone starts looking.
DecemberWorth having ready about 5 weeks ahead, so the page is indexed before anyone starts looking.
Every entry says what the tool does, what it costs, what it is genuinely good at, and where it falls short. Written to one shape, so two tools can be compared without reading two different kinds of page.
Open the directoryTraditional wear, devotional settings and warm light, for the run-up to a festival rather than the morning of it.
Everything that takes a reference photograph and holds the face through it, with what a usable source photo looks like.
Product shots, brand imagery and the listing copy around them, for a shop page or a marketplace listing.
Every prompt here was run before it was published. The picture beside one is the output that prompt produced, not stock imagery chosen to look good next to the text, and a prompt that could not be made to work is not published at all.
Most prompt collections hand you a block of text and nothing else. You paste it, get something unlike the picture on the page, and have no way to tell whether the prompt is wrong, the model is different, the ratio is different, or it needed a photograph you never gave it.
Each of those is answerable before you copy anything, so each is answered on the card: the model, the ratio, and whether a reference photo is required. The image beside a prompt is the output that prompt produced, not stock imagery chosen to look good next to it.
Copy it, paste it into the model named on the page, and set the aspect ratio it asks for. Where a prompt needs a reference photo, upload that first, before pasting. Then change one clause at a time: rewriting the whole thing at once leaves you unable to tell which change did the work.
Prompt bodies are written in Markdown, with each part under its own heading, because a language model follows a structured document more reliably than a paragraph of run-on description.
Image prompts for portraits, festivals, traditional wear, products and editorial work. Video prompts for cinematic shots, social clips and product motion. Productivity prompts for social media posts, thumbnails, logos, search, advertising copy, analysis and code, and for the writing most people actually do: email, messages, meetings and planning.
An instruction you give a model to get a specific result. For images it describes the subject, the wardrobe, the setting, the light and the camera; for text it describes the task, the voice and the format. A good prompt is specific enough that two people running it get the same kind of output.
Usually, but not untouched. Every prompt here names the model it was tested on, because the same text on two models does not give the same picture. Moving one across normally means adjusting a clause or two rather than rewriting it.
Because they are built to keep a real face. Those prompts anchor identity to a photograph you upload, so the output is recognisably you rather than a generic person. Every prompt that needs one says so on its card, before you copy anything.
Yes. Copy any published prompt, change it, run it in any tool, and use what it produces in paid work. No attribution is needed. The one thing not permitted is copying the library wholesale and republishing it.
No. Every published prompt is free and complete, including the settings it needs and the output it produced. The site is supported by advertising rather than by putting the text behind a wall.