Two Minute Tips: Vidu Q3 Grids & Durations
- SORAY-AI
- Mar 1
- 6 min read
Medium Version:
This is written as a companion article to my video:
Here, you will find the prompts for each piece used in the video for easy copy pasting! Feel free to use them as a reference. AI is great with writing these prompts (I think I had AI write all of these since I was pressed for time, and I just wanted to show how these tricks work.)
First we’ll be starting with Specifying Durations in Your Prompt
semirealism, world of warcraft cinematic style
follow the prompts to show each scene for the duration requested.
[0–2 seconds] Extreme close-up of thick, bubbling liquid gold with high-fidelity metallic reflections
[2–4 seconds] Sudden transition to a massive silver dragon’s eye snapping open in a snowy blizzard
[4–5 seconds] A deep blue cobalt crystal violently shattering into thousands of glowing azure shards
[6–8 seconds] A white porcelain surface cracking as glowing liquid kintsugi gold veins rapidly spread across the frame. Cinematic 8k, high-speed transitions, realistic physics for liquid, scales, and stone.
Semirealism, world of warcraft cinematic style.
Follow the prompts to show each scene for the duration requested.
[0–2 seconds] A majestic stone owl statue in a moonlit forest suddenly turning its head 180 degrees;
[2–4 seconds] The stone statue instantly melting into a swirling pool of glowing violet ink and arcane runes;
[4–6 seconds] From the violet ink, a swarm of glowing mechanical butterflies erupts and flies toward the camera;
[6–8 seconds] The butterflies freeze mid-air and shatter into sparkling diamond dust that settles on a dark velvet floor.
High-fidelity textures, dramatic lighting, magical particle physics, seamless temporal transitions.
[0–4s] Cinematic wide shot of the volcanic forge entrance, orange magma flowing slowly down jagged black rocks
[5–8s] Camera pushes deep into the forge interior where a metallic hammer strikes a white-hot metal core, sending golden sparks flying
[9–12s] The glowing metal core is carried by a mechanical arm toward a submerged stone tunnel
[13–16s] The core is plunged into deep teal ocean water, creating massive clouds of hissing white steam and violent bubbles.
8k resolution, realistic fluid and thermodynamic physics, epic orchestral lighting. semirealistic style, world of warcraft cinematic style
High-fidelity anime animation, vibrant sakuga style. Follow the prompts to show each scene for the duration requested.
[0–2 seconds] A fierce warrior with long flowing hair draws a glowing blue katana from a sheath; intense electrical sparks and white light erupt from the blade
[2–4 seconds] The warrior performs a high-speed dash through a dark forest, slicing through a giant shadowy monster in a blur of motion
[4–6 seconds] The warrior unleashes a massive, swirling vortex of blue fire energy from the blade that explodes across the entire screen
[6–8 seconds] The battle is over; the warrior slowly and calmly sheathes the katana while white cherry blossom petals fall gently in the golden light of a sunrise.
4k resolution, dynamic camera angles, fluid character motion, seamless transitions.
That covers the precise timing/duration section of the video.
Note that the clips above were all text 2 video but this trick can work with image 2 video as well.
Keep in mind that if you’re using I2V with a character, they may lose the consistency of details/features when using Vidu Q3 and other video models.
The next section we’ll be covering is Grids/Storyboards which starts at roughly 52 seconds into the tutorial video.
The first step is to generate a grid of the scene you are envisioning.
generate a 2x2 grid of an anime anime fight scene between a knight and a wizard, friendly duel. magic effects, epic. in the 4th panel the two men shake hands after a fun duel.
Do not include text, borders or numbers.
the image should be separated into quadrants, with each quadrant illustrating part of the duel, and the last panel is the resolution with the handshake.
You can do 2x2 or 3x3 grids, or even more! I personally prefer 2x2 grids because the images are bigger and show more detail.
Now, here’s how to adjust your prompt for making use of these grids.
FIRST WAY — STRAIGHTFORWARD PROMPTING
Treat the reference image as a 4-panel sequential storyboard. Animate the friendly duel between the wizard and the knight, starting at panel 1 in the top left and ending with panel 4 on the bottom right.
Completely ignore and remove the white grid borders. The final video must be a single, seamless, continuous animation following this sequence. High-resolution textures, fluid character motion, vibrant colors
Transition each frame with a blur or fade effect so the scenes do not merge or mix
SECOND WAY — USE DURATIONS
[0–2s] Focus on the top-left panel (Panel 1): The wizard blocking th e knight’s sword
[2–4s] The scene transitions to the top-right panel (Panel 2): The wizard unleashes a barrage of spells at the knight
[4–6s] The scene transitions to the bottom-left panel (Panel 3): The climax of the battle, a big explosion
[6–8s] The scene transitions to the bottom-right panel (Panel 4): The knight and wizard shake hands after their friendly duel
Instructions: Seamlessly animate and transition between these panels using fade transitions. Maintain high-fidelity 2d anime style throughout.
For Vidu Q3 in particular, I’ve found that grids seem to work a bit better when combined with the duration method. However, as seen in my tutorial video, the results for the grid method are not always accurate. It varies from model to model, but can still be a useful tool for brainstorming scenes without making 4 separate videos.
Now, I will share the other grid used in the video, which features Vidudu, Vidu’s mascot character. :)
IMAGE PROMPT
generate a 2x2 grid of this character going to an amusement park and having fun. each quadrant should follow a chronological order, top left being the first image and bottom right being the last.
the small grid/storyboard should contain 4 separate pictures to illustrate the character’s time at the amusement park, from getting there in the morning to riding the ferris wheel at night during fireworks. make it adorable and fun
VERSION 1
Treat the reference image as a 4-panel sequential storyboard. Animate the blue mascot’s journey by starting in panel 1 at the sunny park entrance, then fading to panel 2 on the rollercoaster, then transitioning to panel 3 with the cotton candy, and concluding with panel 4 at the night-time Ferris wheel.
Completely ignore and remove the white grid borders. The final video must be a single, seamless, continuous animation following this sequence. High-resolution textures, fluid character motion, vibrant colors
Transition each frame with a blur or fade effect so the scenes do not merge or mix
VERSION 2
[0–2s] Focus on the top-left panel (Panel 1): The blue mascot stands at the sunny ‘Funland’ entrance, waving excitedly as the camera pans slightly.
[2–4s] The scene transitions to the top-right panel (Panel 2): The mascot is now on a rollercoaster, rushing down a steep track with its ears flapping wildly in the wind.
[4–6s] The scene transitions to the bottom-left panel (Panel 3): The mascot enjoys pink cotton candy in front of a spinning carousel.
[6–8s] The scene transitions to the bottom-right panel (Panel 4): It is now night; the mascot waves from a Ferris wheel gondola as fireworks explode in the dark sky.
Instructions: Seamlessly animate and transition between these panels using fade transitions. Completely ignore and remove all white borders. Maintain high-fidelity 2D cartoon style throughout.
And finally, the prompt for the very last clip celebrating Chinese New Year!
Semirealism, world of warcraft cinematic style.
Follow the prompts to show each scene for the duration requested.
[0–2 seconds] A small, adorable snake made of polished green jade with golden kintsugi cracks is curled up on a pile of shiny gold coins and red silk
[2–4 seconds] The jade snake wakes up and its eyes glow with warm golden light while red paper lanterns overhead begin to glow and sway
[4–6 seconds] The snake nudges a red lucky envelope which bursts into a shower of glowing golden sparkles and flower petals
[6–8 seconds] Vibrant red and gold fireworks explode in the starry night sky as the jade snake looks at the camera with a happy, smiling expression.
8k, high-fidelity textures, magical particle physics, seamless temporal transitions.
That covers everything from the video. I hope this gives beginners some new ideas to test out in their video creation journey.
Have fun, and stay tuned for more!
