next »

[Gregg Williams]

Oh yeah; unit deployment. I guess I should add support for that in my own game, huh. I keep forgetting about that for some reason.

Haha yeah I miss your updates here, I need to remember to peek over at your forums sometimes.

[Gregg Williams]

Getting back to this I promise just life is being dumb at the moment, much like the bloody jeep that seems like it has u-joints that are starting to head towards the path of blowing out.

[Conker534]

Good, I want this to be a thing!

[Gregg Williams]

Good, I want this to be a thing!

Cheers so do I though today has been bad again. Had to fix a power window and now am at an ER. Stupid complications of life lately.

[Conker534]

Ah man no rush! The time will come when it comes!

[Gregg Williams]

Alright finally getting back into this, after far to many delays that ended up killing all momentum and a healthy round or three of debuffs to the motivation pool.

Anyways started looking at the pathfinding requirements again. What I ultimately want to do is to pre-allocate a fixed size array of nodes that can be correctly indexed by integers. A node should hold a flag indicating if its on the open list, closed list, or hasn't been visited yet, in addition to the various other common bits like cost estimate, parent, location and so forth.

Then I want a priority queue for an open list that just holds integers that are just the indexes into the node array.

With all of this, I should be able to arrive at something that has zero dynamic memory allocations, no longer needs a closed list, and has a quick open list thanks to the priority queue backing.

All of this is likely way more optimized than it needs to be for the client side, but I want something very solid, and performance oriented for the server side. Thankfully a better implementation never hurt the client either.

The main trick now is to just implement the above bits, with the most challenging part likely being the priority queue, as I'm not familiar with standard back ends for such a data structure like binary heaps. There seems to be a huge lack of open source c++ implementations around as well. The other possible route is to use the STL heap functions on a vector to implement my own priority queue. The STL has its own priority queue, but it doesn't allow finding of elements in the queue, or updating an existing element in the queue and re-sorting, both of which are needed to implement A*.

Kawe should also be returning to work on M.I.N.T now or quite soon, thanks to some side work being wrapped up, so hopefully more art assets will start to flow.

[Gregg Williams]

So I'm just having more universe problems, allergies kicked into overdrive causing me endless eye problems, and I also need to rip apart my jeep's axle and replace the u-joints before they blow and likely take the shafts with them. I expect to fight the 20 years of rust this weekend, wonder if I can sell tickets?

On the game front it's been decided that we really need more planning to enable steady progress. So I'll be spending more time on project management and production soon. Putting together some longer term milestones, and more importantly some vertical slices that are probably around a 2 week effort each. This way we can better plan what assets are needed when, and get something that can be executed upon, evaluated, and a new 2 week slice planned.  

There is also a chance that we might switch to ever beautiful and classic hex well see what the future holds on that front though.

[Gregg Williams]

The *#)#@!&*! 20 years of rust is still winning after 2 days of hell. On the plus side I've also managed to get a bit more of the production side of MINT coming together.

[Gregg Williams]

So a 1 day job that should of cost 50 bucks, took more than a week and around 10x the cost, but its done. Which is more than I can say about the 8 month ongoing game with the IRS. Still I should be climbing back into the dev saddle tomorrow which is exciting. Need to get momentum going again badly.

[Gregg Williams]

Alright finally making progress again.. *whew* also no idea why that gif speed is on crack but oh well.

[nikolail]

Gif speed is compensading your lack of dev speed ?

[Gregg Williams]

Gif speed is compensading your lack of dev speed ?

Haha ouch

[Gregg Williams]

Thinking about animation a ton today. Specifically how to handle movement animation and sync it to actual movement in the game. Isometric tiles mean to stay straight you need to move in a 2:1 step distance. So far to date, I've just had in-game movement take 5 frames, which divides perfectly into our 100x50 base ground tile size. However, this was without actual unit animations. Looking at our current animations, there are units with 4 frames, 8, 10, and so forth. Plus some of those need looped multiple times to be correct. For instance a heavy mech taking a step is 10 frames I think, but really should take 2 or 3 steps to cross a tile which is 20-30 animation frames.

20-30 frames does not divide at all evenly into a 100x50 tile. This isn't even considering animation frames that are longer than 1 frame in duration. I think ultimately the solution here is likely to give up on nice linear movement distances, and instead assign distance to be moved on the x/y axises to each frame in a given movement animation, along with a number of loops needed. This way the total frame duration to move from one tile to another is the unit's movement animation frame time total times the number of loops needed. This should also lead to tighter animation and movement in-game.

Of course it also likely means getting Kawe to give me move distance per frame of an animation.. that might be a tough sell

[Gregg Williams]

Actually I guess thats all probably semi wrong, and just me being obsessive when it comes to perfect synchronization, which pretty much no one does.. Obviously failed an insanity save roll.

[Belimoth]

I think ultimately the solution here is likely to give up on nice linear movement distances, and instead assign distance to be moved on the x/y axises to each frame in a given movement animation, along with a number of loops needed. This way the total frame duration to move from one tile to another is the unit's movement animation frame time total times the number of loops needed. This should also lead to tighter animation and movement in-game.

You're not that crazy, I went through an eerily similar process when I was coding my current animation library.
I had it so you would construct animations with an array of displacements, one for each frame.

[Gregg Williams]

Don't feed my insanity but uh yeah.. well see.

[Gregg Williams]

Alright, got movement queues working properly, and I can now slide a unit around via the pathfinding as needed. Next steps will be actual unit animation, and then actually doing all of this movement via server requests and replies. (Right now just been testing the client side.)

Unfortunately its gotten hot here again, and I'm pretty sure my PROCHOT bit is tripped, and i'm dropping clock cycles to prevent thermal shutdown..

Yeah heat really kills my productivity something fierce. =/

[Gregg Williams]

M.I.N.T development is on halt again, for a few days while I do branding and API integration for our flash game Gravitum.

After nearly 4 months of trying to work out a flash sponsorship deal, I'm really excited to just be able to wrap up and finally release the game into the wild

[Gregg Williams]

Just a quick update. Almost have finished with the Gravitum work, things are sadly going slower than I anticipated due to a complex API and a lack of example code. Support is thankfully decent, but tends to only be able to reply once a day, so things are slow going.

I had been hoping to get back to M.I.N.T early next week at the latest, but another business deal popped out of nowhere and is going to have me busy doing a slightly customized version of A.R.C.S for a few days. Which reminds me, anyone interested in Indie documentary stuff check out the Game Loading kickstarter that ends in a few days, and is close to being funded.

http://www.kickstarter.com/projects/studiobento/gameloading-rise-of-the-indies?ref=live

You can get your very own free copy of A.R.C.S as well

On the M.I.N.T development front, Kawe is hopefully creating some very much needed interface and GUI bits for in-game, though he has refused to show me them yet.. so your guess is as good as mine

Overall the vertical slice for the last 2 weeks has gone fantastic, other than this sudden Gravitum interrupt which has left the actual network side of movement unfinished which I expected to spend a day or two doing. The next slice will be delayed some, but I'm looking forward to it, as it should conclude with action point/movement point use, and the start of COMBAT.

[nikolail]

Thanks lad for keeping us in the loop, hope it brings money on the table