A Tale of Two Image-to-Video Generators
It was the best of times it was the worst of times.
I've been a big fan of RunwayML for almost an entire year now - they have a series of tools, which are very advanced.
Here is one of my first viral generations with RunwayML:
BUT there are more competitors in the space and they are doing very well. One is Haiper.AI. I use them exclusively for the anthropomorphic flower redo that I posted the other day. But here are two quick examples of how it's performance for the layman’s toolk - uploading an image and telling it to animate i - is really giving the elbow to Runway.
First, we generated an image. First, we generated an image about "wǔxiá" — a genre of Chinese historical fantasy featuring the adventures of martial artists in ancient China. Midjourney, of course.
Then we ran it through HaiperAI for a 4 second animation - and the results are stunning.
Check out how the engine recognized each of the individual faces their motions, the wind elements, and even the background fellows it animated appropriately.
I ran the same thing on Runway, and the results were well… wonting. Wait till the end.