Recently I had the opportunity to test the new Google AI-Video generator powered by Veo 3(.1). The demo was truley impressive and scary at the same time! And then we were able to test it ourselves …
While others created cool videos, I tried something like “Create a Scene from the Witcher with Geralt riding on a horse towards Kaer Morhen with Ciri by the side“.
And: A true in game scene! But with Console controls (or health indicator) visible on top and in the bottom left.
Next, same prompt – with s,th like “make it look like the movie“
“Sorry, can’t do that” – okay. We don’t want to replicate the movie obviously.
Next: “… make the camera on ground level. No Game Controls or Health indicators must be visible“
… The camera was definitely not ground level and even after a couple of attempts I wasn’t able to get a video without the console stuff. Well It seems pretty obvious what the training material was. Maybe a ways smarter prompt would have done the trick with removing the control and really have a perspective on ground level. But after a couple of tries I gave up and decied to try something else.
Next try: “make a sunset scene of the brauneck mountain range in Lenggries, photorealistic”. (A mountain range nearby – and I’m sure there are thousands of photos)
I had a clear result in mind. Maybe that was my mistake.
And there came a video of a sunset! Someone behind was like “wooaaah impressive! Does it look like the Brauneck?!” — ” … no”.
It was impressive and beautiful — but not close to the real location. *Sigh* another edge case maybe. Maybe I’m just bad at prompting or I hit a couple of edge cases.
The orthers created cool and impressive clips of Ice Hockey players cheering etc. The seemed to be more happy with the results than I was. I’m pretty sure Veo is very cool — but I had a clear result in mind — and didn’t manage to get it. Maybe that was my mistake.
Leave a Reply