Some visual update:


Changelog: http://nobodyswar.tumblr.com/post/112260795353/progress-update

Created a small tumblr at http://nobodyswar.tumblr.com/ to track dev progress.

Today was mostly spent on redoing the AI.

It's much better and smarter now.

The AI work takes around 200ms to decide what to do, so I sent the bulk of the decision (basically, traversing the tree of all possibilities) to a separate thread so the game doesn't get blocked it.

The search space for each move is: 6 X positions times 5 possible rotations (in the worst case) = 30 possibilites. With a lookahead of 3 moves, this means 30*30*30 = 27000 end nodes. Which is quite a bit for a real time (in frame) decision.

One trick that makes things much more manageable (and the way it's different from a Tetris AI, for example), is that we can try to maximize the combo score (unless we are trying to defend). This way, we don't need to traverse all the tree of possibilities because we can stop as soon as something gets removed from the board.

I've also spent a little time working on a "AI Arena", where I can put two AIs to fight.

There are still at least 2 or 3 obvious improvements to it, but it was good work for today.

Video of new AI kicking my ass twice:

Today's update.

Here is a full game end-to-end video of the current game:



Things done today:
- redone the "show next piece" boxes and now showing the 2 upcoming pieces.
- played a bit with the "send/save" neutral puyos animation.
- improved the pieces to be more like classic Puyos (3-pieces have always 2 in common, 4 pieces can either have 2 columns or 4 the same which you can switch, etc).
- added a basic game over animatino
- implemented a proper Puyo score system with chain power, margin time, etc. Makes for much better balanced matches.
- Added a small popup that shows what you have done.

I'm still very unhappy about the send/save animation but I have no good idea on how to change it now.
Tomorrow I'll probably work a bit more on the AI.

I've also created a private github with the code. Will open it soon.

Before the redesign, I was going for a glitchy effect when the groups disappear (so it was intentional! ). Not sure if it matches the new design though, which is much simpler.

Probably not the cup fo tea of TIGSource, but let's see.

I'm making a remake of / version of Puyo Puyo (originally released in 1991).
It's a bit of a mix between the old versions and some newer variants (if you are a fan: Fever, maybe Mission).
It supports up to 4 local players (The code supports multiplayer but I don't really enjoy writing server-side code, so who knows).

I use Haxe + OpenFL for it and plan to release it for Mac/Win/Linux.
I also want to make the whole source code available, and plan to set up a github repo very soon for it.

I work on a JS version of it a couple years ago and I just finished porting most of it.
Then I decided to work a bit on Photoshop and try a few layouts and UIs for the game.
I ended up with this:



You can check some gameplay video at: https://www.youtube.com/embed/6o_46L4Ro70

What you can see here is 4 players (human left, bots on the other 3) playing for a bit. Some of the animations are missing (specially like when you send neutral puyos to other boards). Also, the lower part of the screen is mostly a mock.

I'm still trying to come up with a visual style and have loads of things that I want to add.

Anyways, I needed a place to keep track of my progress, so here it is.

Let me know what you guys think.
Cheers

As others have said, I'd suggest not dropping everything on the floor midway through the project. Have you looked into HaXe/OpenFL (http://haxe.org). I've been using it for a while now, and it has multiple targeting platforms (Flash, native C++ win/mac/linux, android, IOS, etc..) and it has pretty good AS3/Flash.

I've been working for a couple weeks on a top view 2D fast racing game. Mostly inspired by MicroMachines 2.

It supports tmx maps, which makes it fairly easy to create new tracks.
The AI doesn't use waypoints, so it's track independent, no-cheating and super challenging.
It currently has lap/track logic done and support for 4 local players split screen (can't be seen on the shots). 

Still very much WIP, i.e. my todo list is huge, but I've decided to share some shots:



Rescaled to 50%, without the GIF dithering (game currently runs 30fps 1280x720):



It's done in Haxe, openFL, HaxePunk. I currently plan to release it (and have been testing) on PC, Mac, Linux and Ouya.

It's not a novel concept, but I have decided to give it a shot into making a racing game without falling into the usual design decisions.
My goal is to focus on the racing mechanics (repeating the same track, finding ways to minimize lap time, etc).
I'm trying to avoid things like catch up logic, events that are random on the lap (like powerups), and the "OMG drifting is so cool we need to do it all the time".

Ideas, suggestions and comments? 

awesome! Thanks for sharing this.