AI video generation is still under intensive development and I feel that for that moment and the technology and the technics to use it are not yet mature.
The technics are still progressing and some new tools are being released:
like Stable Video Diffusion (test it here while you can)
And websites are being created like Pixverse.
Also the technics and the tools used for video generation are not mature enough yet...
So one person called me a noob but I think he is right.
I am doing this website also to teach myself how to do the things, because this helps me to put the things straight in my head.
So with AI and this website, I can have 2 approaches:
- One consists to teach myself how to do things and this gives kind of video.
I showcase this, but it is very hard to do a human animation that looks decent without controlnet.
I still have to work on the tutorial that I promised, just because it takes a while to test... and I am probably gonna come with an obvious conclusion: don't do like me, use controlnet, it works better...
Still... if I can find the good settings, I can make things without having to rely on the good will of someone else.
And as I said, the kind of teaching is useful to fix the issues that you may have... even if you use controlnets...
- The other approach is to use the software for what it can already do well. I think the clip that I post here today falls into the second category.
There are some effects that AnimateDiff-Cli-prompt-travel can do, that are actually usable to produce content.