I recently thought I’d just test to see how advanced AI generated videos are getting. Rather than try to make something ‘real world’ I thought I would test to the limit a bit more, so settled on a full AI workflow. By this I mean first creating a static reference image with AI and then getting AI to generate a video from the AI generated static image.
This means the whole process is exculsively AI generated. Two of the videos are only 5 seconds long, but on the third I deceided to cut three tests together and add sound (the only bit not generated by AI) manually as well as playing each clip twice. All these videos were generated on the lowest quality setting of 540p. I imagine that the output quality would be significantly better if output at the highest resolution (but it is very expensive to do).
1) Three 5 second AI generated video experiments run back-to-back and again in reverse with added non-AI generated music:
2) A spaceship and astronaut floating in space:
3) A spaceship engaging reverse thrusters:
View my AI art galleries.
