How Image-to-Video AI Actually Works

A visual guide to the technology behind AI photo animation

How Diffusion Models Generate Images

The surprisingly simple idea behind AI image generation: learning to remove noise. With diagrams.

The compression trick that makes AI image generation fast enough to run on your computer.

How image-to-video models extend diffusion into the time dimension to generate motion.

Temporal consistency explained: why AI-generated videos flicker, morph, and how models fix it.

Kling, Runway, Veo, Luma, Pika and Hailuo compared on price per clip and clip length, with the same photo run through each.

Seven models priced per five-second clip in dollars on one bill, with what resolution, sound and retries do to the number.

One sentence about what moves. The photo does the rest. Real prompts from the app and a pasted prompt trimmed to nine words.

Seven photos through twelve models with the same prompt, every clip side by side with its price. Watch them and vote for the one you would keep.

Want to see these models in action?

Or start with the image to video AI page: what you can animate and what it costs.