please add to this :0
Log in with itch.io to leave a comment.
Awesome project and thanks for sharing it! I only played for about five minutes, but I couldn't see anything to interact with (nothing happens with the bushes and houses). I tried walking in circles around the starting nest to see if the other birds would mimic me, but didn't notice any change in their aimless wandering.
Do you have exaples of actual behaviors the birds can mimic at this stage? I'm sure I wasn't playing long enough to see results, but is would be fun to know what I might try to see some changes in behavior.
Hi Clickityclak! Thank you so much! I'm glad you think so highly of it already.
To answer your question in-depth — In Alpha 0.1.2, the only gameplay that exists is movement and a symbolic, useless punch animation.
Birds will mimic your movement if you are within their line of sight, and you can get them to punch as well by punching while they are looking at you. The more you train this behaviour the more likely they are to mimic it.
The actual parameters of their learning are super rough right now. I'm currently tuning the birds to be less instantaneous with their mimicry, because I've found that they're over-adapting to the immediate moment, and this means they are overwriting long-term behaviours in favour of the short-term.
As for what is planned...
As I type this, I'm working on a huge number of upcoming features. I've already coded the basics for the following:
- The collection, carrying, and storing of objects, including an weight-based encumbering system. I just need to spend a day or two animating in Blender.
- Punching. It just requires some sort of death system. (I already track hit-points.)
- Sustenance. Birds will need to eat or they will die. This will add an amount of natural selection which should help improve the early gene pool a lot. Super important for early-game.
- Squawking. Birds can squawk to temporarily draw attention of other birds to themselves. You will be able to do this too. Helpful for teaching other birds
- Ranking of birds along a 'success metric'. In machine learning terms this is referred to as a cost function. This will enable prioritization of learning for birds so they don't learn from the shittiest of their peers.
- Passing of genes to a new generation, and saving birds' unique neural networks between sessions. (Super super important for deep learning, since the amount of training that needs to happen in a deep network before complex behaviour is seen is substantial.)
- Randomized names. (This one is actually finished, but there's no way to display the names.)
- 'Class archetypes' (think Soldier, Ascetic, Maniac, Farmer). This affects their dispositions toward categories of objects, both within their self-ideal and with their perception of others. This is also finished, but won't become noticeable until we have a few types of interactions happening in gameplay.
- Hunting other animals. There is code for this, but I
need to create a totally new set of models for the actual mob creatures. This will take a while and is fairly low priority compared to gathering/bird-to-bird combat.
Hello there! Really nice upcoming features your have here! I can't wait to see them added to the game!
Have you considered participating in our GDWC #gamedev contest?
Check out our itch.io page for more info :)
This is absolutely beautiful, in the ironic and unironic sense. I love the concept of it; Very simple but also very creative with its execution. It looks promising, even if its a joke lol.
- Thank you so much! 💖
- I take my birds very seriously, and they WILL deliver on their promise.
If anyone would like a Mac or Linux build of Alpha 0.1.2, let me know!