You know it's like porn to game devs :C.

But, did you notice the creeper face?

Disappear from BYOND for a week or so due to working a ton, and come back to that video? Best thing to return to since Animate() etc. Just... bravo, Bravo. Lol. I am getting pretty hyped for Lux.

Nope. That would be my love for platformers, his past work with them (Gravity), and now Lux. It stems from playing a whole hell of a lot of Super Mario Bros/World and Donkey Kong Country 1-3 as a child.

Got a bit more glow for the robots eye. It's using BLEND_ADD on it, so it only appears to glow when the robot's white parts are darker than normal, otherwise it just looks normal.

I also added some standing lights which kind of slowly fluctuate over time.

The area I'm in is a new room I worked on today. About the size of the very first room in terms of width, but about double the height. So this room would be a 3x2 compared to stuff like the save/upgrade rooms which are 1x1 on the "map grid".

I worked to kind of figure out how these things work. Looks like if you set the alpha to 200+ you start to see a weird lit edge around the light, which looks terrible.

Here I've set the two lights at 150 each, which together end up being about as powerful as something set to alpha of 1.

Stress testing my 47-state autojoin system with a really peculiar autotile. The tile basis is 16x16, but the autotile covers a 16x20 area, causing overlap of the walls on certain configurations.

This test is showing a stress test generating 20,000 turfs per second using runtime modification.

Load-time autojoining is significantly faster than runtime modification. Load-time autojoining only searches 4 neighbors per tile (NORTHWEST, NORTH, NORTHEAST, and EAST). Runtime autojoining is slower than load-time autojoining because it has to search all 8 neighbors of the modified tile, which then each themselves perform a single update.

The matching algorithm is very flexible, and allows each tile to have its own autotile group, which links by id to a global list that contains a series of tile_id records that allow tiling. The lookup is associative, so it's quite fast.

This is a demonstration that entirely different tile types can create a join match by configuring the rules properly:



Allowing different tiles to have their own join_ids and join_groups allows for autotiles to treat their neighbors as part of their group, while their neighbor doesn't treat that same autotile as part of its group. This allows for join inclusions unlike other algorithms which only match tiles with their same types.

The algorithm stores its neighbor matches in at256 format, but it uses a reduction algorithm to generate at47 format. I've left code hooks in to allow other formats to admix naturally with at47.

This is the distilled format showing the at256 codes when applied to an at47 set:



This is the base image that contains all of the art that is used to generate each at47 variant:



So far, I'm pretty pleased with this approach. It's pretty sprightly. It could be optimized further, but at this point I'm satisfied with it enough to move on to another thing that needs doing.

:O. Would make for some really good diablo dungeons.

New enemy.

- Like the stretchy coord effect. :)
- On way back up, might swivel head to stair down the bot in spite. Creepy effect for cheap.
- Wondering what it looks like if mouth is slightly open on way back up.

Updated. It now moves as if it has physics on the mouth rather than just sticking partway into the ground when closed.

I also made it open/close it's mouth a bit on the way back up in order to express the "chewing" animation a lot more.

All I have to do is animate how it will look to die and I should be done with it. I may also add an effect that kicks up some dirt when it hits the ground.

One a side note, some insight into how I came up with it: I originally wanted to have it swing like a wrecking ball at it's target. I wanted to do that because there's already games where enemies drop down like this and I didn't want to just be ripping them off.

Unfortunately, getting that to work properly would've been much more difficult, not to mention the fact that it's not really practical for a creature to capture prey by swinging when it can just drop down.

Also, I'm lazy.