Chat GTT AI Prompt: Copy-Paste Image-to-Video Prompts for You and Your Dog

Chat GTT AI Prompt for a Person (Copy and Paste)

Attach one clear, front-facing photo of a single person, then paste the text below. Words in square brackets are yours to change. This version casts the person in your photo as the one who says "I heard you the first time, bro"; swap the two roles if you would rather play the cashier.

Vertical 9:16 video, realistic, handheld phone-camera look, one continuous shot.

Setting: the checkout counter of a small streetwear shop. A wall of sneakers and a rack of caps behind the counter, a card reader on the counter, bright overhead lights.

Customer: the person from the attached photo. Keep the same face, skin tone, hairstyle and build in every frame. Outfit: [oversized hoodie], [fitted cap], a thin gold chain worn over the hoodie, a phone in one hand.

Cashier: a fictional store clerk behind the counter in a plain black T-shirt, calm and expressionless, not based on any real person.

Camera: medium shot from the customer's side of the counter, chest up, both faces fully in frame, slight natural shake, no cuts and no zoom.

1. The customer taps a bank card on the reader, card face turned away. The reader beeps.
2. The cashier looks at the screen and says in a flat, robotic monotone: "The card declined."
3. The customer taps the card again. The cashier repeats it word for word, same flat tone, same blank face: "The card declined."
4. The customer lowers the phone, tilts their head and says to the cashier, mouth moving with the words: "I heard you the first time, bro. Why you keep sayin' it like a robot?"
5. A short pause. The customer turns to the camera, deadpan: "Chat GTT lookin' ahh."

Sound: quiet shop noise, card-reader beeps and the two voices only. No music, no on-screen text, no captions, no logos, no brand names on clothes or signs.

Your tool cuts off before the last line? Run beats 1–3 as one clip, then start a second clip from its final frame for beats 4–5, and join them in any editor.

Chat GTT Dog Prompt: Gold Chain, Phone in Paw, Same Comeback

Attach one photo of one dog with its face and most of its body visible, in good light. Fill in the breed and coat in the brackets: describing them in words helps the tool keep your dog looking like your dog.

Vertical 9:16 video, realistic, handheld phone-camera look, one continuous shot.

Setting: the checkout counter of a small streetwear shop, a card reader on the counter, shelves of caps behind, bright overhead lights.

Dog: the [breed] from the attached photo, with the same [coat colour], markings, ear shape and eye colour in every frame. It stands up at the counter with one front paw resting on it and a smartphone held against its chest in the other paw. It wears a chunky gold chain that stays clearly visible. Paws stay dog paws: no fingers, no human hands.

Cashier: a fictional human clerk behind the counter in a plain uniform, calm and expressionless, not based on any real person.

Camera: eye level with the dog, medium shot, dog on the left and cashier on the right, both fully in frame, slight natural shake, no cuts.

1. A bank card lies on the reader. The reader beeps.
2. The cashier says in a flat, robotic monotone: "The card declined."
3. The dog nudges the card. The cashier repeats it word for word, same flat tone, same blank face: "The card declined."
4. The dog tilts its head, looks at the cashier and speaks in a relaxed, deep voice, mouth moving with the words: "I heard you the first time, bro. Why you keep sayin' it like a robot?"
5. A short pause. The dog turns to the camera, deadpan: "Chat GTT lookin' ahh."

Sound: quiet shop noise, card-reader beeps and the two voices only. No music, no on-screen text, no logos, no other animals, no extra people.

A cat, a rabbit or a bird may work if you swap the animal words, but expect more retries: the counter pose is easiest for a medium or large dog.

What Each Line of the Prompt Controls

Both Chat GTT prompts are built from the same nine parts. When a clip comes out wrong, find the part that failed and change only that line.

Change one line per retry. If you rewrite three lines at once and the clip improves, you will not know which edit helped.

Keeping the Face, or the Dog, the Same From Start to Finish

A Chat GTT clip only lands if viewers recognise you, or your dog, in every second. These habits help most.

• Start from a good photo. One subject, facing the camera, even light, no sunglasses, no hand over the face. For a dog, pick a shot where the eyes and both ears show.
• Describe what the photo shows. Video tools lean on words as well as pixels. "Short black curly hair, round glasses" or "tan and white corgi, white blaze down the nose" gives the model a second anchor.
• Keep the camera still. Big camera moves force the tool to invent angles your photo never showed. A near-static medium shot keeps the face close to the source.
• Avoid turns away from the lens. Profile shots and looking down are where faces drift. The prompts keep both characters facing forward or toward each other for that reason.
• Generate a few takes. Even with the same prompt, results vary. Make several and keep the one where the face holds and the lines land on the right speaker.

Avoid List for Negative-Prompt Fields

Some tools have a separate box for things you do not want. Paste this there. If yours has no such box, the "no" sentences at the end of each prompt above already do the same job.

extra fingers, human hands on the dog, merged paws, morphing face, changing hairstyle, changing coat colour, second customer, people walking through, crowd, text, captions, subtitles, watermark, logos, brand names, readable card number, smiling cashier, music, slow motion, jump cuts, sudden zoom, horizontal frame, cartoon style, blurry face

Keep the list short and specific. A very long avoid list can make some tools ignore part of the main prompt.

Common Mistakes and How to Fix Them

Most failed Chat GTT videos go wrong in one of these six ways.

• The cashier rewords the line. "Your card was declined" kills the robot joke. Fix: put the exact words in quotes both times and add "word for word".
• The wrong character says the comeback. Fix: name the speaker at the start of every spoken line, as the prompts do, and never leave a quote without a speaker.
• The dog grows hands. Holding a phone tempts the model to draw fingers. Fix: keep "paws stay dog paws: no fingers" and let the phone rest against the chest instead of in a grip.
• A busy or dark photo. Group shots, sunglasses, side profiles and dim rooms all give the tool too little to copy. Fix: retake the photo before you rewrite the prompt.
• Too many ideas in one clip. Extra gags, a crowd or a dramatic camera move crowd out the dialogue. Fix: one counter, two characters, five beats.
• A real card on screen. A readable number is a privacy risk once the clip is posted. Fix: "card face turned away" is already in the person prompt; keep it.

How to Run the Prompt, Step by Step

The same five steps work in most image-to-video tools.

1. 1

   Pick one photo

   Choose a sharp, front-facing photo of one person, or one dog with its face and body in view, in good light.

   Watch out: Use your own face or a photo you have permission to use, never a stranger's.

2. 2

   Attach it and paste the prompt

   Upload the photo as the start image, paste the person or dog prompt, fill in the square brackets and choose a vertical frame if the tool asks.

   Watch out: Delete the brackets themselves; some tools read them as literal text.

3. 3

   Generate and check the beats

   Watch the result for three things: the repeat is identical, the comeback comes from the right character, and the face or dog matches your photo.

   Watch out: If the clip stops early, split it into two clips as described under the person prompt.

4. 4

   Add the dialogue

   Keep the voices the tool made, or add captions or a voice-over in your editor (see the next section).

   Watch out: Generated speech does not always match the mouth. Captions cover the gap.

5. 5

   Label it as AI and post

   Export the vertical clip, switch on the AI-content label in the app you post to, and say in the caption that it is AI-made.

   Watch out: Realistic AI video usually needs a label under platform rules; check the app's own guidance.

Adding the Dialogue: Captions or Voice

The words make the meme, so plan how viewers will hear or read them. There are three routes, and you can mix them.

Voices from the tool. If your image-to-video tool makes sound, the quoted lines in the prompt tell it what each character says. Quality varies: words get swapped, voices land on the wrong mouth, or the second "the card declined" comes out with a different tone. Check before you post.

Your own voice-over. Generate the clip silently, then record the lines in an editor and line them up with each beat. For the cashier, a text-to-speech voice is a neat fit: the whole joke is that he already sounds like a machine.

Captions only. Burn in short on-screen captions, one per beat: "The card declined." / "The card declined." / "I heard you the first time, bro…" / "Chat GTT lookin' ahh." Many people scroll with the sound off, so captions help even when the audio works. Some creators spell it chatgtt or write "chat gtt looking ahh"; any of these is fine, as long as it is not corrected to the real chatbot's name, which would lose the joke.

Label the Clip as AI Before You Post

A clip made from one photo is AI-generated, even when it looks like a phone video from a real shop. TikTok, Instagram and YouTube each give you a way to mark a post as AI-made; switch it on and add a line such as "AI remake of the Chat GTT card-declined meme" to the caption.

Two more rules keep the trend friendly. Only use faces that are yours or that you have permission to use. And leave the people from the original clip out of it: our prompts use a fictional cashier and never ask for the original customer's look, name or voice.