Skip to content
View original post on X: MagnificOfficial
Published
16/100

Magnific shows a multiverse panel technique for AI video with masked reveals

AIMagnific describes a method that generates six synchronized video clips of one rider galloping in different settings, then stacks them and animates overlapping rectangular masks to reveal each one. The prompt keeps the horse, rider posture, timing and camera static so the subject reads as one continuous figure across the boundaries.
MagnificVerified on XOfficial
@magnific
View on X

Part of a thread · earlier post

3. Multiverse panels

One continuous movement across six different worlds

Match the action, layer the clips and animate rectangular masks to reveal each reality

Prompt:

Create six synchronized side-profile video sequences of the same woman galloping on the same horse: a sunny beach, green forest, cloudy countryside, dusty sunset road, misty mountains and rainy urban street. Give each version a different outfit, lighting and atmosphere, with a natural smartphone-video aesthetic.

Use exactly the same galloping cycle, rider posture, timing, scale and screen position in every clip. Keep the camera static.

Stack the six clips and reveal different portions through overlapping rectangular masks of varying sizes. Animate only the masks: slide them horizontally and vertically, expand and contract them, and switch between the video layers.

Keep the horse and rider perfectly aligned across every boundary, so they read as one continuous subject assembled from different worlds. Generate distinct synchronized clips rather than simply replacing one background. No duplicated subjects, disconnected limbs, camera movement or floating video screens.

Source details

Magnific · x.comPublished · added here

Original title: 3. Multiverse panels