We have a new champion 🏆 🏆 🏆
New #1 just dropped ! ! !
https://t.co/xczCHJAfYl has taken the top spot on https://t.co/y4zAKKAlEd.
Congrats, champ. 🎉 🎉 🎉
Who’s coming for the throne next? 👑
Here’s the prompt for another one:
Overall Requirements
Use the provided screenshot of the X profile page as the sole reference for the background, layout, and character identity. Place the complete screenshot proportionally in the center of a vertical canvas, adding blank space above and below. Do not crop the screenshot or alter its internal aspect ratio.
Preserve the original gray-blue header background, profile picture position, white UI, buttons, icons, screen layout, spacing, colors, and proportions. Keep the camera completely static throughout—no zooming, panning, rotating, or rearranging the layout. Except for the specified character animation, do not alter the UI or background.
Main Action
The character in the profile picture climbs out of the circular avatar frame.
0–1 seconds: The character looks to the left.
1–2 seconds: The character looks to the right.
2–3 seconds: The character rolls its eyes in a full circle, then breaks into a mischievous smile.
3–4 seconds: The character reaches out with both hands and grabs the edge of the circular avatar frame.
4–6 seconds: The character pulls out one leg, then the other, and lands in the empty white area of the profile page.
Only the character should emerge from the avatar. The page must not distort or deform. The circular avatar frame must remain fixed in its original position. Do not create or show a second animated character.
Here’s the prompt
you can use this platform to generate it: https://t.co/s0jo2C7e0N
15 seconds, vertical format. Use the provided screenshot of the X profile screen as the sole background and layout reference.
Maintain the header image, profile icon position, white UI background, each button, icon, screen composition, margins, colors, and ratios from the original image.
The camera is completely fixed. Do not zoom, pan, rotate, or reconfigure the screen itself.
Do not change the UI or background except for the specified effects.
Use the girl inside the profile icon—wearing a white cap, gray clothes, and brown hair—as the sole character reference.
For 15 seconds, completely maintain her facial features, hairstyle, hair color, white cap, outfit, line art, coloring, and the soft anime illustration texture.
No photorealization, 3D-ization, changing to another person, or outfit changes allowed.
The girl starts holding a red apple, just like in the original image.
[0.0–2.4 seconds]
The entire screen remains completely fixed as in the original image.
Only the circle inside the profile icon begins to move.
The girl takes a small, natural bite from the red apple in her hand and cutely munches on it.
Do not open her mouth unnaturally wide. Do not puff out her cheeks or lips excessively.
The apple naturally shrinks bite by bite until she finishes the last bite.
After finishing, she smiles a little mischievously.
[2.4–4.0 seconds]
The girl places both hands on the round frame of the profile icon and climbs out from it, using the circular icon like a small entrance.
She steps out naturally onto the screen side, first one foot, then the other, and lands directly on the white margin area of the X profile screen.
Only the character emerges from the icon.
The profile screen itself does not deform.
Leave the original round profile icon in its place as-is.
[4.0–8.2 seconds]
The girl who has landed spots the text written on the screen and gets an idea for some "mischief."
Target all readable text/letters/numbers existing on the screen, such as profile name, user ID, bio text, link text, join date, following count, follower count, subscription label, text inside buttons, text in posts, timestamps, or battery numbers.
The girl peels off the letters one by one with her fingertips, as if they were paper stickers.
She crumples each peeled letter on the spot into a small wad like paper scraps.
She tosses the wadded-up letter scraps toward the viewer with a flick.
The tossed scraps fly toward the foreground of the camera and disappear off-screen in the front.
Repeat this "peel → crumple into a wad → toss toward here" action several times.
Continue at a brisk tempo: first time, second time, third time, and so on.
Each time she tosses, the letter properly disappears from its original spot, turning into white space.
Peel only the letters.
Leave non-text visual elements untouched, such as the header illustration, profile icon frame, character images, UI graphic icons, button shapes, divider lines, etc.
Ensure the tossed paper scraps aren't so large that they obscure the entire screen.
[8.2–10.4 seconds]
After tossing several letter scraps, the girl looks around with an expression like "There's still some left."
Then, she pulls out a simple broom from somewhere.
She lightly grips the broom and sweeps the remaining letters briskly from top to bottom, left to right.
Only the letters touched by the broom gather like paper scraps in sequence and are swept off-screen in a flash.
Letters do not vanish suddenly; they always disappear in sync with the broom's movement.
As she sweeps, the screen's text information gradually vanishes, eventually creating a large white space.
[10.4–11.2 seconds]
Finish removing all text.
The girl satisfiedly sets the broom aside horizontally.
She rummages in her pocket and pulls out a thick black marker pen.
She pops off the cap with a "pop" and turns back toward the now-white space.
[11.2–13.7 seconds]
The girl uses the black marker to handwrite boldly in the large white space in the center to lower part of the whitened profile screen:
The prank was a success!
Display the string exactly as
The prank was a success!
No spelling changes allowed. No omitting words. No garbled text. Do not add extra sentences.
Thick black handwritten marker letters.
Each character is naturally written from the pen tip one by one.
Large enough and clear enough to read easily.
Write boldly using the screen's white space.
[13.7–15.0 seconds]
After finishing writing, the girl caps the marker.
She gazes at her own doodle with satisfaction for a moment.
Then, she turns to look over her shoulder at the viewer, narrowing her eyes slightly and flashing a mischievous grin with a "Gotcha!" vibe.
Not a wide-mouthed laugh, but a sly, delighted, cute smile.
For the finish, freeze for about 0.7 seconds in a composition where both the large black text "The prank was a success!" and the girl turning back with her mischievous smile are clearly visible simultaneously.
[Motion/Effects]
A brisk, comical animation.
Add light bounciness to the character's movements, but keep the background X profile screen always fixed.
Use natural 2D character animation only when the girl moves within the screen.
When "peeling letters," clearly show the letters curling up and peeling away like paper.
When "crumpling into a wad," make the motion visible as she lightly squeezes it with her fingertips into a small paper ball.
When "tossing toward here," add just a bit of depth to make the paper ball fly into the foreground.
Pull out the broom only after the tossing is done.
The broom's movements are lively but not overly intense.
Do not obscure the screen with smoke or overly flashy effects.
Prioritize visibility above all, making it easy to understand what's happening.
[Important Fixed Elements]
・Use the provided image as the sole visual reference
・Camera completely fixed
・Do not change the original screenshot's composition, header, or UI layout
・Do not arbitrarily change or delete text until specified
・In the first half, peel letters one by one, crumple them into wads, and toss them toward here
・Pull out the broom from somewhere only after tossing is finished
・In the second half, sweep away the remaining letters with the broom
・At the end, leave none of the original text
・Remove only text; leave all non-text UI graphic elements and images
・The only new text remaining at the end is "The prank was a success!"
・Maintain the girl's face, hair, white cap, gray outfit, line art, coloring, and illustration texture
・Do not photorealize the character
・Do not 3D-ize
・Do not change to another person
・Do not add extra fingers, arms, or legs
・Do not duplicate the broom or marker
・Do not add unnecessary characters
・Do not generate new logos or text
・No garbled text
・No dialogue, narration, or speech bubbles
GPT-6 Astra + Three.js = an incredibly smooth, interactive experience with stunning results. Maybe you don’t even need Blender anymore.
https://t.co/thbE8VpCMi
GPT-6 Astra + Three.js = Smoothly INTERACTIVE and AWESOME. Might not even need Blender, the stuff you can build on the web now is wild. Prompt⬇️
It cut my workflow from a couple of hours down to just 30 minutes!
GPT-6 Astra has only been out for 1 day, and after scrolling through X
I’ve already seen people doing some absolutely wild things with it.
Here are a few of the most impressive examples 👇
dude GPT-6 Astra is some kind of turbo-AGI machine god for 3D games. It one-shot this in 45 minutes for hardly a couple % of my quota.
I figured out how to get great graphics out of it. The trick is image gen. I'll share the process below.
GPT-6 Astra modeló una casa completa en Blender y la convirtió en un recorrido en Unreal Engine 5. Sola.
Alberca, jardín, sala, cocina. Así vas a ver tu casa antes de que exista.
Eso hoy lo cobra un estudio de visualización con arquitectos y artistas 3D.
OpenAI dice que Astra ya maneja Blender, FreeCAD y KiCad como lo haría una persona: abre el programa, modela y exporta.
En BenchCAD, el examen de diseño CAD, sacó 95.9%.
El modelo anterior: 83.3%.
Y con Sites, dentro de ChatGPT, arma y publica sitios, apps y juegos desde un prompt.
Los números y el video son de OpenAI.
¿Le encargas tu casa a un arquitecto o se la pides a ChatGPT?
So how good really is GPT-6 Astra at 3D modeling?
I took an old drawing of a steam train, gave it to Astra to reconstruct it in Blender. After few minutes it crafted 3,295 fully editable detailed objects with beautiful geometry. You can obviously tell it how detailed or simplified you want it. Go build your own transport tycoon 😎
GPT-6 Astra built this Manhattan world in Unreal Engine over the course of a week.
It was literally able to go street by street to make each one perfect.
Built this browser-playable kart racing game featuring racers, tracks, drifting mechanics, power-ups, realistic physics, and polished visuals.
https://t.co/Iqd510imYc
GPT-6 Astra hizo este juego de karts. Completo.
Corre en el navegador y tú lo puedes jugar ahora mismo.
8 corredores, 3 vueltas, derrapes, ítems, física y gráficos. Todo lo armó la IA.
Lo hizo Pietro Schirano con Sites, dentro de ChatGPT.
Le describes el juego y Astra lo construye, lo aloja y lo publica.
OpenAI lo puso como demo oficial del lanzamiento.
Y dice que cualquiera, sin saber programar, puede crear y jugar su propio juego en minutos.
Hace un año esto era un equipo de desarrolladores y meses de trabajo.
Se llama Tidal Rush y está en Sites de ChatGPT.
¿Fin de los estudios chicos o tu momento para hacer un juego?
Astra vs Fable 5 (not fable 5.1)
build a 3d aquarium game using single prompt with reference image from gpt-6-astra
astra >> fable 5 confirmed, now time to test against fable 5.1