next »

[Chis]

I'm picking up this project again because I don't want to be a quitter!!!
After some long hard thinking, I'm going to try this time to pick one thing and do it really well.

So the plot is that you're stuck inside a world of worlds. You look for coins and feed them to an arcade machine, which will unlock more worlds and eventually the exit.
The catch is, you're doing this with a partner, aka bird girl. Bird girl will have her own mindset and so more than following you around, she may choose to explore off on her own or help you out or play around or even boycott based on your relationship. The game will therefore revolve around building a friendship with her.

You won't be able to communicate verbally with her because that would explode the interaction space. Instead you'll have to do things like jumping or pushing.
The immediate goal is to have just moving around be interesting of itself. Then maybe there will be puzzles. I don't want to think about combat.

As with last time, it will be a platformer. Right now it's an HTML5 game - that was a really bad idea. I'm heavily considering moving to other languages not Javascript.

[ｂ∀ ｋｋｕｓａ]

So glad to see this project revival. good luck this time ^^

[Mr. Levich]

I really like the idea of an AI "companion" - not a stupid bot, but something more complex, like you said - with a mindset of her own. I think "Majesty" was one of my favorite examples of AI mindsets - would really like something like that implemented again

Judging by bakkusa's comment, this is a second take on this project? Do you have a devlog of a previous one, where I could read about your idea and problems with HTML5 and JavaScript? 'Cause I've had a bunch of others developers tell me that HTML5 for games is "The Future", and it's "oh so powerful" and so on and so on, but I don't really buy it. Would love to read about your experiences in that field

[Chis]

@bakkusa thanks so much! I'm going to need it lol

@Mr. Levich I've never played Majesty but the game mechanics sound really interesting! Yeah, this'll have the influence-but-not-control aspect to it it'll also be more exploratory and intimate.
My old devlog is here: http://forums.tigsource.com/index.php?topic=34190
HTML5 for games is a scam The biggest problem for me is Javascript. Once you have a sizeable project it becomes more difficult to work on, because on top of implementation errors, you don't get basic things like type and variable and class checking, so if you're not meticulous about it it can easily spiral into mush.
Also, it's slooow.



Let's talk about pathfinding!

To find reachable platforms from one platform, I simulate a range of jumps; every next step I solve some kinematic equations to determine the range of movement as well as any blocking tiles. If the space becomes smaller than the body size, I discard it.
At the end I get a list of platforms, as well as the boundaries of the jumps.



So now that we have paths between platforms, we can run A* on this search space.



Unfortunately this only works with a limited scenario - no variable jump speed, running acceleration, multiple jumps, or slopes All things I am willing to sacrifice though.

[SolarLune]

Huh, this entire thing sounds interesting, but this in particular:

The catch is, you're doing this with a partner, aka bird girl. Bird girl will have her own mindset and so more than following you around, she may choose to explore off on her own or help you out or play around or even boycott based on your relationship. The game will therefore revolve around building a friendship with her.

is intriguing. So what exactly dictates what she'll do? And from the sound of the game, it seems like if it mainly focuses on you, the immediately useful thing would be to have Bird Girl follow you and help you. If that's the case, then it seems like the other options (exploring, playing, being upset) are all bad...? Would there be places for her not to follow you that would advance the game? Do you have any immediate influence over her actions?

Sounds cool so far!

[Mr. Levich]

Thank you for posting the link - I've read through your previous devlog and the progress looked amazing. Really great that you've decided to come back to finish it Keep spinning those ideas 'till they work!

As for HTML5/JavaScript - I know it might be too late, but have you tried CoffeeScript or TypeScript? Never tried any of it myself, 'cause I don't do any major JavaScript development, but I've read enough to think that TS might have the features you were longing for. As for performance - maybe asm.js can help?..

[jctwood]

This is an incredible intriguing concept. The fact that combat is not a priority alone makes me happy, look forward to seeing more development!

[marcgfx]

javascript is not that bad really. it allows you to write messy code, but in the end its your own responsibility speed has been improving a lot too, but it's not as fast as other languages. with webGL and webAudio there are some pretty powerful features that can help you make state of the art games. If you want your game to work on mobile, things do get very tricky though. I've noticed the main issue to be power-consumption.

your concept sounds interesting, making believable intelligence is always a really tough challenge. good luck!

[jctwood]

Having read through the entire Carousel devlog and playing the game I am infinitely excited for this! May I ask what the job you got around graduation was? I feel like I have seen similar pixels somewhere...

[Raku]

Well! This is interesting for sure! Dont give up!

[botwoj]

YES!! was really looking forward to your game, glad that you're continuing on it All the best!

[Chis]

@SolarLune: The girl will start off with her own personality, but she'll change her behavior over time depending on your actions. For example, if you're always on the go, she'll probably just follow you around a lot. If you act playfully, she'll think it's okay to play around too. If you stand around too much, she might get impatient and go off on her own. You might even switch roles with her leading and you following.

You don't have immediate influence other than pushing her hahaha. But I think for the most part she'll stay by your side while occasionally deviating. When that happens, exploration can be done in parallel. If she separates from you and finds something interesting, she can come back to you and signal for you to come over, and vice versa. I guess I never thought it would be a problem because your interaction with her would ideally be as interesting as progressing the game. But you're absolutely right about not going overboard with it.

Thanks for the support!!

@Mr. Levich: Oh, I think you just solved my problems especially Typescript. If I get completely fed up with my code I'll definitely give them a go! though at that point I might just drop Javascript altogether.

@marcgfx: Aye, you're totally right about that. I forget how easy it is to prototype with Javascript. Do you have any idea how it compares to C++ or C# in terms of performance right now?

@JctWood: It's not that I wouldn't do combat, it's that I couldn't ... hahaha!
but yeah, I don't think violence would work well with this game anyway.
Aww, much thanks for updating my devlog! I doubt I'm the person you're thinking about, but I graduated a year ago, working as a mobile engineer now.

@Rakugaki-Otoko: I remember you! Thanks for your continued support  

@botwoj: Awww thanks!! Glad to know people are still interested!

[Chis]

I've been working on observation! Specifically, when bird girl enters a new level, she doesn't have any knowledge about it. The platforms and reachable links are pregenerated, but she doesn't actually know about them until they enter her field of vision. Likewise for items/npcs.
and when I say field of vision I am going a bit meta here, because instead of only seeing what her onscreen character would see, she has a viewing screen just like the player. So if you and her are sticking together, you can generally assume she's seen everything you've seen.

[Ouren]

Don't get off your carousel



 

[Raku]

This is looking so exciting! I miss games with AI partners in them, and this one looks like it's going to be very indepth!

[Chis]

Unbelievably I'm still working on this!!!

[midoredomi]

oh my god yes 

[Atnas]

<3 __ <3

[Pixel Noise]

Good to hear - like people have said, it's a really cool concept.

[Chis]

Thanks for the kind support, everyone!

Good news - I got the pathfinding to work in between rooms! (the medium sized blobs are doors lol)

Bad news - I am going to scale back my original concept of a human-like AI After attempting a behavior model, I am convinced that my original vision is beyond my technical expertise. I'll keep what behavior I've implemented so far, but probably not much else.
To compensate, I'm making the game more story / character focused. Still hashing out the details, so stay tuned!