Explore what's new in AI

Gary Oldman’s 50-millisecond eye timing makes close-up villains feel predatory

The acting breakdown says letting the head move before the eyes reverses normal gaze behavior, making a look read as deliberate threat before dialogue begins.

MiniMax H3 Max stages a cinematic chef's-revenge clip

A pan-flame gag and plated reveal show H3 Max handling action, timing, and a clean payoff, while the creator reports a roughly one-minute turnaround.

Grok animates four visual studies into looping live wallpapers

The demo cycles through neon cities, astronauts, forests, and robotic heads as subtle animated loops, suggesting a fast way to turn still artwork into phone-ready motion.

Tank Fu turns martial-arts combat into a tank-based action game

Knight Owl Games' pre-alpha lets tanks dash, grapple, parry shells, and chain air and ground moves, with an arena and playtest access planned.

AI Artworkgen maps anamorphic framing to environmental contrast

The framing guide links anamorphic lenses with deliberate environmental contrast and controlled light, giving image makers concrete tokens for cinematic composition.

Inspector Lui combines Midjourney and Seedance for a place-based short

Painterly streets, temples, and markets turn lived memories into a moving city portrait, showing how image-to-video can carry place, atmosphere, and nostalgia.

Magnific Originals turns Don Quixote into The Sorrowful Figure

The animated short uses battered heroism and textured dark fantasy to reinterpret Don Quixote, offering a finished reference for AI-led literary adaptation.

ElevenLabs inline audio tags give Seedance performances finer control

Inline cues for whispers, breaths, pauses, and coughs shape delivery before Seedance animates the image, giving dialogue scenes more expressive control.

Dreamina keeps one speaker consistent across five visual styles

One frontal shot moves from realism through manga, comics, watercolor, and clay while preserving identity, dialogue, lip-sync, and a realistic background.

Seedance 2.5 swaps performers by separating identity from motion

Using a reference video for movement and two images at a time for identity, the workflow preserves timing, expressions, cuts, wardrobe, and selected positions.

Bilawal Sidhu builds a walkable memory palace from seasonal 3D scans

Bilawal Sidhu aligns winter footage with a summer scan so viewers can step through the same place across time, turning memory into navigable space.

iamneubert turns a CyberSub joke into a detailed CGI submarine clip

The clip turns a throwaway CyberSub gag into a neon-lit, cybernetic underwater vessel, showing how a focused prop and environment can make a VFX idea feel production-ready.

Lloydcreates explores a chrome Harley Quinn character system

Four linked studies vary armor, masks, props, and poses while preserving a red, silver, and transparent-metal palette, a useful reference for consistent character design.