What is an AI prompt?

The text that tells an AI tool what to make. For image tools you write the subject, light, camera and style; for video tools also the camera move, the length and the sound; for text tools the role, the task, the rules and the output shape. The more concrete the prompt, the more predictable the result.

Should I write prompts in English?

For image, video and sound tools, yes: the models were trained on English data and other languages give weaker or wrong results. Text tools such as ChatGPT, Claude and Gemini work well in any major language.

How do I find the prompt behind a picture or a video?

First recognize the tool: Midjourney's stylized light, Veo's 8-second clips with sound, ChatGPT's figure-in-a-box pictures are tell-tale. Then find the closest example in this library and change the prompt to your subject. A tool that guesses the prompt from an uploaded picture is coming for SEOLOK members.

Can I use the pictures these prompts make on my website?

Most tools allow commercial use on paid plans; free plans may be restricted, so read the tool's terms. Get permission when you use a real person's photo. Adobe Firefly was trained on licensed content only and carries the least risk on a business site.

AI prompts

AI prompts: how was that picture, that video made?

Every example carries the tool, the exact prompt, the parameter code and the steps. The first 6 are open to everyone; the whole library comes with a free account.

An AI prompt is the text that tells an AI tool what to make. The same tool gives a similar result for the same prompt, so knowing the prompt behind a picture or a video means being able to make it again for your own business.

The prompts in this library were chosen for businesses: product photos, menu pictures, intro videos, a blog outline built to rank on Google, replies to Google reviews. Image and video prompts are in English; the tools work best that way. Each prompt comes with its parameter code (for example --ar 16:9 in Midjourney) and the steps.

Members find the whole library in the panel under "AI prompts".

Tool

24 prompts

ImageNano Banana (Gemini)

Turn a phone product photo into a studio shot

Open the tool

The clean, softly lit product photos you see on Instagram and in online shops. Upload your own photo and type this: the product stays the same, the backdrop and light become a studio.

Example

Prompt

Turn this product photo into a professional studio shot. Keep the product exactly the same (shape, label, colors). Place it on a seamless light grey backdrop, soft diffused key light from the upper left, gentle reflection under the product, shallow depth of field, 85mm lens look, no text, no extra objects.

How to do it

Open Gemini and attach the product photo (flat surface, in focus).
Paste the prompt; keep the sentence about label and colors if they matter.
Refine in the same chat: "make the backdrop white", "less shadow".
Resize to 1200 px and save as WebP before putting it on your site.

#product photo#e-commerce#studio

ImageChatGPT (GPT Image)

Show yourself as a boxed toy figure

Open the tool

The "action figure" trend everyone posts: the person inside a plastic blister pack with small accessories of their job. One photo and this prompt.

Example

Prompt

Create a realistic collectible action figure of the person in this photo, in a clear plastic blister pack with a cardboard backing. Keep the face recognizable. Add 3 small accessories next to the figure that match their job: [a laptop, a coffee cup, a notebook]. Packaging title text: "[NAME]", subtitle: "[JOB TITLE]". Studio lighting, product photography style, front view.

How to do it

Open image creation in ChatGPT and upload a clear half- or full-body photo.
Fill in the brackets: accessories, name, job title.
If the face is off, ask again with "keep the face closer to the photo".

#figure#social media#trend

ImageMidjourney

Portrait that looks like a film still (Midjourney)

Open the tool

The cinematic portraits all over Pinterest: dim light, neon reflections, film grain. The codes at the end set the ratio, the version and how realistic it looks.

Example

Prompt

Cinematic portrait of a woman in her 30s, rain-soaked city street at night, teal and orange neon reflections on wet asphalt, shallow depth of field, Kodak Portra 400 film grain, soft rim light on hair, looking slightly off camera, photographed on a 50mm lens

Parameter code

--ar 2:3 --v 7 --style raw --s 150

How to do it

Type /imagine in Discord or on midjourney.com and paste the prompt.
Append the parameter codes: --ar 2:3 for portrait, --style raw for a less "AI" look, --s 150 medium stylization.
Upscale one of the four with U; V gives variations.

#portrait#cinematic#parameters

ImageNano Banana (Gemini)

Repair and colorize an old family photo

Open the tool

Those videos where a scratched, faded black-and-white photo is renewed and colored in a second. Upload the photo and type this; faces are not changed, only repaired.

Prompt

Restore this old photograph. Remove scratches, dust, creases and fading. Add natural, period-accurate colors (skin tones, clothing, background). Do not change faces, expressions, poses or the composition. Keep the original grain subtle. Output at the highest resolution.

How to do it

Scan the photo at the highest resolution you can, or shoot it in flat light.
Upload it to Gemini and paste the prompt.
If a face changed, add "do not alter facial features" and ask again.

#restoration#colorize#family

ImageMidjourney

Isometric miniature room / shop

Open the tool

Game-like tiny rooms and shops shown in cutaway. Café, barber, office: swap the contents and draw your own business.

Example

Prompt

Isometric miniature diorama of a cozy coffee shop interior, cutaway view, tiny espresso machine, pastry display, two small tables, warm pendant lights, pastel color palette, soft studio lighting, clean white background, highly detailed 3D render, tilt-shift look

Parameter code

--ar 1:1 --v 7 --s 250

How to do it

Replace "coffee shop" with your business (barber shop, dental clinic, flower shop).
Change the 3–4 objects inside to match your shop.
--ar 1:1 is square, right for Instagram; use --ar 16:9 for a website cover.

#isometric#3D#business

ImageIdeogram

Minimal logo draft with correct lettering

Open the tool

Most tools garble letters in a logo; Ideogram renders text properly. Change the brand name and the sector, get 4 drafts. Have a designer vectorize the final one.

Prompt

Minimal flat logo for a dental clinic named "[BRAND]", a simple tooth icon merged with a smile curve, geometric, two colors (deep teal and white), the word "[BRAND]" in a clean geometric sans-serif below the icon, centered on a plain white background, vector style, no gradients, no 3D

How to do it

Put your brand name in place of [BRAND]; change the sector and the icon idea.
Pick the "Design" style in Ideogram; turn Magic Prompt off so the text stays as typed.
Do not use a draft before it is vectorized (SVG); it breaks in print.

#logo#brand#lettering

ImageNano Banana (Gemini)

Move the product to another place (change the background)

Members only

ImageMidjourney

Flat illustration for a website

Members only

ImageNano Banana (Gemini)

Appetizing food photo for a menu

Members only

ImageAdobe Firefly

Blog cover image (copyright-safe)

Members only

VideoGoogle Veo

Beach video that looks shot by a drone (with sound)

Members only

VideoSora

10-second product commercial

Members only

VideoKling

Turn one photo into a moving video

Members only

VideoRunway

Venue tour: camera gliding through the door

Members only

VideoGoogle Veo

Glass fruit cutting ASMR video

Members only

VideoHailuo (MiniMax)

Talking intro avatar from a photo

Members only

SoundSuno

30-second ad jingle for a business

Members only

SoundElevenLabs

Natural voice-over for an intro video

Members only

TextChatGPT

Blog post outline built to rank on Google

Members only

TextGemini

Meta description that gets the click (155 characters)

Members only

TextChatGPT

Human replies to Google reviews

Members only

TextClaude

Turn AI-sounding text into a human voice

Members only

CodeClaude Code

Audit a WordPress theme's speed

Members only

CodeCursor

Generate structured data (schema) for a page

Members only

Frequently asked

What is an AI prompt?: The text that tells an AI tool what to make. For image tools you write the subject, light, camera and style; for video tools also the camera move, the length and the sound; for text tools the role, the task, the rules and the output shape. The more concrete the prompt, the more predictable the result.
Should I write prompts in English?: For image, video and sound tools, yes: the models were trained on English data and other languages give weaker or wrong results. Text tools such as ChatGPT, Claude and Gemini work well in any major language.
How do I find the prompt behind a picture or a video?: First recognize the tool: Midjourney's stylized light, Veo's 8-second clips with sound, ChatGPT's figure-in-a-box pictures are tell-tale. Then find the closest example in this library and change the prompt to your subject. A tool that guesses the prompt from an uploaded picture is coming for SEOLOK members.
Can I use the pictures these prompts make on my website?: Most tools allow commercial use on paid plans; free plans may be restricted, so read the tool's terms. Get permission when you use a real person's photo. Adobe Firefly was trained on licensed content only and carries the least risk on a business site.

Let's take your site to the top together

Start with your free report; use the platform yourself or leave the SEO + GEO work to our team.

Go to my dashboard