Been meaning to play more with puzzlescript, it looks fun

Nitpick: a little more idiomatic don't need the double loop

Code:

var hist = [];
for (var i = 0; i < width * ss; i++)
{
  hist[i] = [];
  for (var j = 0; j < height * ss; j++)
  {
    hist[i][j] = 0;
  }
}

Plotting with 3x3 points would work to prevent the darkening, you could also try "3x3 circle" or "diamond" shapes and see that effect. My understanding of supersampling is different though? Maybe I'll learn something.

Fwiw that's what the "commands" above facilitate, just in a script friendly way, and with multiple handlers able to run on a given command - eg an object could have scripts that open or close it, and change its team based on passed parameters for an "activate" command. Commands are also a simple way to implement message passing (and we actually extended this facility in kag to allow message passing to a  specific player rather than always broadcasting)

Whoops, looks like I wasn't looking closely enough last night :^) seems like your scale is ok actually, but you're right that you aren't seeing enough detail in there :/ would need to see your implementation to find the issue, probably.

Ashaman is on the money - you're just rendering it bigger and scaling it down, you need to sample your field more for each unit of area (therefore the "super" in supersampling).

Code:

c = a random point in [0, 1]
for (a number of iterations) {
for(i = 0 -> 3 wide bucket)
for(j = 0 -> 3 high bucket)
{
  (x, y) += Fi(x + i/3.0f, y + j/3.0f); //accumulate
}
  (x, y) /= 9; //average
 
  c = (c + ci) / 2
  if (canplot(x, y)) plot(x, y, c)
}

Although I'm sure people will disapprove, we've done a _lot_ of gameplay construction in the non-networked sense and then applied the necessary synchronisation after-the-fact. Our engine is heavily networked anyway so adding a command to "call such and such function on all machines" (bit of a simplification, but not much) is trivial, and a handful of things on each object are automatically synced by default (position information, what config the object was loaded from (ie what script run on it), health, team, owner player, etc).

Getting to that point has been tricky, but all of that's essentially engine side and allows the gameplay stuff to remain relatively separate from netcode.

The way of going about it I'd suggest is definitely in line with Garthy's

• Tinker with networking separately so that you can send stuff over the network quickly and ideally reliably (or have some way of dealing w/ packet loss) - solutions like ENet and RakNet can help a lot in getting up and running quickly here.
• Get simple game networking engine going with synced variables (continuous + one off synchronisation) and commands (one off synchronisation with parameters).
  Things like Object.AlwaysSync("health", minimum_change); Object.SyncNow("health"); Object.SendCommand("die", stream_serialised_parameters);
• Write gameplay stuff and make sure it works single player.
• Fix any sync errors as needed.

You need the networking there, but I disagree with all this "OH YOU HAVE TO REWRITE EVERYTHING TO NETWORK ANYTHING THAT YOU ALREADY CODED" dogma.

fwiw you can always make a simple gui that outputs lua scripts for conversations Good luck with it!

The Ludum Dare game I'm currently working on. It apparently involves a Snake with pseudo-physics now.

This is really cool Tobias! Wish I'd had more time for LD this weekend, I had a 21st party to go to and ended up just tinkering with a verlet physics toy while hung over :^(

Wish I could tag Stanley here, JLJac's advice above is very much in line with his philosophy.

Note that you can do stills that look good to get the design down, but you do need to start from a more or less blank canvas for each frame; the stuff you copy in should be detail level stuff, not whole limbs that you then butcher with the marquee tool.

Fwiw craft studio and probably a bunch of other softcore IDEs has a drag and drop building blocks mode where the blocks you're dragging around are bits of lua. You can also just type it if you know what you're doing. Don't feed the idea that programming is a bad and scary thing, try to turn your designers into semi competent programmers. If they begin to understand code they'll also begin to understand what reasonable game design looks like, and how tricky implementing ideas "just so" can be.

Finished a 4k word CS paper. Started to like it a lot more towards the end, made me reflect more on my programming philosophy than I was expecting. Always a nice feeling when published papers are in agreement with you as well

Just FYI particle splashes look approximately 20x better if the particles all have slightly different lifetimes; they dont all vanish at the same time that way.

drew some anime stuff too

haha its like a wang sword

your technique is keeping getting better man, keep it up

Aw man now I cant even escape #auspol on TIGS whole thing makes me super depressed.

As capitalist as it might sound, try to use that for motivation to start making a dime on your work. Money is freedom these days. Hugs for being in a scary fucking place

Delko, cheers, I think that's subconsciously sort of exactly where I'm aiming with this spec.

For the decals, it hit a part of my memory with streets of sim city; you could just edit texture maps to colour your car (if it wasn't in game there was some easy way to find the textures), consider just making a nice mapping of most of the plane and letting people go nuts, maybe with some drag and drop cliche decals for starters. In case I haven't said it before, the game is looking pretty hot, it's been fun watching it grow over time.

Normally REASONS in this circumstance would be NDA or some other previous contractual or verbal agreement that's more or less unimportant to the reader.


Cell that thing is looking good, kinda egyptian-wall-painting-y, though most of the forms are nonsense, haha

Rasse that's all kinds of hot

"Shiprekt" is out, simply join any server hosting the mod from your regular KAG install to play it! Really not sure if I should do an announce thread or not, I guess we'll see how excited people get

Keen to hear what TIGS folk think of it, a few weeks for a total conversion mod from sidescrolling brawler to top down dynamic ship driving seems good to me but maybe my standards are too low



Absolutely loving that dog endless runner btw. I have a friend with two king charles cavaliers that pull the same idiot expression the whole time. He got the version without blue dashing forcefield though.

Last set of rocks for a while now I think, finally comfortable with the workflow. I think this set of rocks would benefit greatly from some plants and things around them so I'll have to make an alpha blend compatible version of the shader ( ) - probably without vertex painted damage as the vertex colours would be better used for procedural wind and bending animation.

This is the first one using a fully generated base texture, no photo sourcing. I think it came up quite well.

560 tris per rock. ~5.4k tris for the scene, including the ground.
512 norm+diff+spec, 128 detail+damage+damage+environment, 64 gradients (3 texture lookups)

I'm still working on optimising and improving the shader, figuring out what works and what doesn't - this is still in free time around work though so I'm just feeling my way.

ZBrush, Blender, UVLayout, Photoshop, Unity.

I've heard of people using blender for this kind of stuff, fwiw. Map the sprites to a bunch of planes like you would in puppet2d, then bind them to bones and only animate the bones in one plane). Might be more technical but you'd get 100% control of the deformations and also be able to do shape key (vertex transformation) based animations if needed; you could also probably set up a ragdoll with a bit of work!