We show examples comparing the text-to-video generation results of our ProMAG-4× latent (4× temporal compression) with ProMAG-16× latent (16× temporal compression). The highly compressed 16× latent latent space can achieve similar video generation quality as the base 4× latent, but with huge boost in efficiency compared to 4× latent space. Generated videos for both are at 192×320 resolution and have 68 frames.
4× Latent
16× Latent
A slow cinematic push in on an ostrich standing in a 1980s kitchen.
4× Latent
16× Latent
Entering a Martian cave to reveal an alien colony hidden within, Cinematic FPV.
4× Latent
16× Latent
Yellow mold growing in a petri dish, moody and dim lighting, cool tones, cold color grade, dynamic motion.
4× Latent
16× Latent
Hyperspeed hand held camera. An irregular sphere shape ball dramatically undulates, warps and explodes as it transforms into a completely different man. Surreal.
4× Latent
16× Latent
A young woman with vibrant red hair, adorned with a whimsical leafy crown, gazes off-camera with an expression of soft awe. Her freckled face is bathed in warm...
4× Latent
16× Latent
Tranquil lakeside scene during autumn. Wide shot of the entire lake, with colorful trees reflecting in the water. Slowly move the camera over the lake's surface...
4× Latent
16× Latent
A person sitting on a bed looking at a spectacular night sky full of galaxies and stars, view from behind, fisheye perspective, vibrant...