GTA 6 is almost here, and with it comes the art style everyone recognises: the hand-painted loading-screen posters. Thick black outlines, glossy skin, a sky that looks like it’s on fire, palm trees, fast cars and people who are clearly up to no good. The good news: you don’t need to be an illustrator to make art like that. With the right prompt, AI image and video models produce the GTA 6 style in seconds.

This guide gives you a free, copy-ready style prompt, shows what it looks like in three different AI models, and explains how to turn the stills into video and build a full 3-minute clip. Everything below was made on VirtuaVixen, but the prompt works in most other AI tools too.

What makes the GTA 6 loading-screen style?

Before you write a prompt, it helps to know what you’re asking for. The look comes down to a handful of ingredients:

• Bold, clean black outlines around every figure and object, like a comic book.
• Cel-shaded skin with glossy highlights: flat colour areas with a few sharp, shiny highlights instead of soft photo lighting.
• Hyper-saturated sunset colours: hot pink, orange and teal, with almost no grey.
• A flat, graphic poster composition: one or two characters posed confidently in the foreground, a recognisable setting behind them.
• Slightly exaggerated glamour: everyone looks a bit more confident, tanned and dangerous than in real life.
• The setting: a sunshine state full of art deco hotels, palm trees, beaches, swamps, neon motels, 80s convertibles and police helicopters.

The free GTA 6 style prompt

Put this text in front of your own scene description. It works for still images:

Hand-painted digital illustration in the style of a crime video game loading-screen poster: bold clean black outlines, cel-shaded skin with glossy painterly highlights, hyper-saturated sunset pinks, oranges and teals, flat graphic poster composition, slightly exaggerated glamour proportions, no text, no logos.

For video, use this version. The extra “not a photograph, not 3D” stops video models from drifting back to realism after a few seconds:

A 2D hand-painted illustrated animation, every frame looks like a moving crime video game loading-screen poster: bold clean black outlines, cel-shaded skin with glossy painterly highlights, hyper-saturated sunset pinks, oranges and teals, flat graphic comic-book look, not a photograph, not 3D, no text, no logos.

Notice that the prompt never says “GTA”. That’s on purpose: when you type the name of the game, most models paint the game’s logo into your picture. Describing the style gives you the look without the logo.

How to write the scene

After the style prompt, describe your scene in full sentences. A simple formula works best: who (always adult characters) + what they wear + where they are + what happens + the camera. A few examples:

• A glamorous woman in her late twenties with long dark hair, in a pink bikini and big sunglasses, leans against a white 1980s convertible on a neon-lit oceanfront boulevard at sunset, palm trees and art deco hotels behind her.
• A woman in denim shorts and a cropped tank top runs out of a gas station at night clutching a bag of cash, her partner waiting in a muscle car with the door open, a police helicopter searchlight above.
• An outlaw couple races along an ocean causeway at sunset in a white convertible while two police cars chase them far behind. Wide tracking shot from the side.

Option 1: Krea 2, the strongest look

Krea 2 gave us the most convincing loading-screen look of every model we tested: thick outlines, glossy skin and a real poster feel.

1. Open the VirtuaVixen generator and go to the Image tab.
2. Pick a Krea 2 workflow.
3. Paste the image style prompt, then your scene, and generate.

Two tips: Krea 2 can’t write readable text, so leave signs out or ask for “signs without text”. It also likes big curves, so describe the body type you want plainly.

Option 2: Seedream V5 Pro, when you need lettering

Seedream V5 Pro gives a slightly softer, more polished version of the style and follows colours and details very closely. Its big advantage: it can write short, readable text. Want a motel sign that says “VACANCY” or a title on your poster? Write the exact words in quotes. In the generator, pick Seedream V5 Pro in the Image tab and use the same image prompt.

Option 3: WAN 3.0 Text to Video, straight to video

You can also skip the stills and go straight to video. In the generator’s Video tab, pick WAN 3.0 Text to Video (or the sharper and faster Prime version). Under Style, choose Crime game poster (GTA 6 look), so you don’t even need to paste the prompt. Describe the scene, choose 16:9 landscape and 5 to 30 seconds. WAN 3.0 adds dialogue, voices and sound effects from your prompt. These two clips were made exactly like that:

Want the same characters in every scene? Make a Krea 2 still of each character first, then use WAN 3.0 Reference and refer to the still as @image1 in your prompt.

Using other AI tools

The prompt is plain English, so it works in most image and video models that follow natural-language prompts. Paste the style text first, then your scene. If your tool has a negative prompt field, add “photograph, 3D render, logo, text”. And again: don’t type the name of the game, or you’ll get its logo.

Common mistakes

• Typing “GTA” in the prompt. Most models paint the logo in. Describe the style instead.
• Too much in one clip. Video models handle one or two actions per clip. Five actions in 10 seconds turns into chaos.
• Drifting to realism. Keep “not a photograph, not 3D” in every video prompt.
• Signs and text. Only Seedream writes readable text. In other models, ask for signs without text.
• Young-looking characters. Always describe adults (“a woman in her late twenties”). Childlike characters are not allowed on VirtuaVixen.

How to make a 3-minute GTA 6-style video

Nobody renders 3 minutes in one go. Build it like a real film, from short clips:

1. Write a mini story of 8 to 12 scenes: the plan, the heist, the chase, the motel, the twist.
2. Lock your characters with one Krea 2 still each, or copy the same description word for word into every prompt.
3. Render each scene as a 10 to 30 second clip in 16:9.
4. Continue from the last frame: pause the finished clip on its last frame, save it and use it as the reference for the next clip.
5. Edit it together with music and sound, and export in 16:9.

Win with it: the Neon Outlaws contest

Made something you’re proud of? Enter it in our Neon Outlaws AI video contest. Make a video of at least 3 minutes in 16:9 in this style, upload it to VirtuaVixen with the tag Neon Outlaws before November 30, and the three best videos win up to double their December view earnings. Mixing VirtuaVixen with other tools is allowed.

See the contest rules and examples →

FAQ

Can I make GTA 6-style art with AI?

Yes. An art style itself isn’t owned by anyone, so you can describe and recreate the look. What you shouldn’t copy are the game’s logos, its characters or its brand names. Make your own characters in the style instead.

Does the prompt work outside VirtuaVixen?

Yes. It’s written in plain English for any model that understands natural-language prompts. Results differ per model; on VirtuaVixen, Krea 2 gave the strongest look.

Which model is best for GTA 6-style video?

WAN 3.0. Use Text to Video with the “Crime game poster” style for quick clips, or WAN 3.0 Reference with your own Krea 2 stills when you want the same characters in every scene.

Why does my image show a logo or strange text?

You probably typed the name of the game, or asked for a sign. Remove the game name, add “no text, no logos”, and ask for signs without text.

VirtuaVixen is not affiliated with, sponsored or endorsed by Rockstar Games or Take-Two Interactive. Grand Theft Auto and GTA are trademarks of Take-Two Interactive Software, Inc.