It's because you should not be able to have an action chooser on empty action besides the empty state and the "Remove Action" button does not do anything when that's the case. I'm thinking maybe there's a bug with the "new action X" numbering scheme because I'm using a rather simple generation mechanism as I did not think anybody would store the scene with default action names, but maybe I need to reconsider ...
I'm a armchair coder too... and the first thing i go in and do is click everything, over and over until it breaks
and I thank you for these words it really means a ton to me.I'm a armchair coder too... and the first thing i go in and do is click everything, over and over until it breaks
I'm still playing around and trying to figure out where in the return data it's generating actions, i think i got a few to work, still playing with it, but def the model you choose to use makes a huge difference too, sometimes it's so off the wall. But this is exactly what i wished for years ago when vam first started and i had my oculus DK 2. lol. Thank you for all the work!!!
HI guys,
Just to let you know what Alita and me have been up to ... there's a huge UI overhaul heading your way. It was A LOT of work but we've survived. I just need to iron out a few rough edges and you'd be able to play with it yourself.
Just a little sneak peek
View attachment 312420
Regards.
There are models like the one you ask for but they are still slow, imperfect and surely cannot be run on a consumer level hardware in any reasonable time. It's still sci-fi unfortunately but we are getting there slowly.I'm still imagining that if one day, AI can train mocap data very well and then apply it to VAM, whether there will be a completely random, coherent, natural, real and unpredictable character animation. She can pose continuously, sit down and stand up, and also can twist and dance simply. I don't understand why AI has not trained mocap data and combined it with VAM. In theory, mocap data is controllable variables, and its regularity and cleanliness are very suitable for AI training requirements. I don't know why no one has made it applied to the VAM characters.
Yes. Most probably you've created 2 actions but they are not linked anywhere. You need an entry point for your AI. Sort of a bootstrap. The easiest way for you to achieve what you want is to create a third action called "Start". Link two actions you've created and click "Test Execute" on the Start action. This will feed the LLM with options to stand up and sit down. If you want we can discuss it more on Discord. You can easily find me there if you have any questions.
The error you're getting is something that will disappear after the next release. I'm really working hard to release the next version with user actions but the tutorial takes forever to make ...
Thanks, I will give that a try. I appreciate it. Voxta definitely needs this kind of plugin to move to the next level of awesomeness. I will find you on discord.
