Update October 2023

In early October I finished the previous month's task of optimizing loading in the game, including making levels load in the background, as well as changing the way levels are reloaded in order to speed up restarting levels. It was the last big programming task. I realized I could have explained more last time, actually, so I will do it now. As usual the 4D makes things a bit harder than normal.

First, some background: levels in Miegakure are lua files. For those unfamiliar, Lua is a scripting language. This allows using more procedural methods to generate levels in addition to placing objects the regular way. For example, when I want to place a bunch of rocks in a region, I don't have to place each rock by hand, I can just specify the region and the number of rocks. I can try different variations until I find one I like. This is good because placing things by hand in 4D is often way too much work as there is much more space to fill, and it is hard to see all of 4D space at once 😁.

Ok, so a common way to program restarting a level in a game is to reload it from scratch but keep a copy of every mesh or texture you have loaded previously, and when you try to load a mesh or texture, look up in your "database" using the file name whether you can just use the copy you already have. This makes sure you don't forget to clear everything you need to clear (like the position and states of the objects) while still being fast. However, because in this game the levels are filled with procedurally made or placed meshes and textures there's not really a file name to use as a lookup for each mesh/texture. You could come up with a name for every mesh/texture anyway, but then you're basically saying to not execute part of the code when the level loads a second time, to not regenerate meshes.  So I just did that instead: I just mark certain portion of the lua code that generates a level as only running once.  You could also save out everything to a big file, but it feels incorrect to save out things that could just be generated. Anyway, before this change I would sometimes have to recreate very expensive meshes every time the player restarted a level, which would take a while, but now reloading is instantaneous.

For the background loading, because I use OpenGL I unfortunately cannot run graphics command on any thread but the main thread. So I have to load everything in the background thread, then pass that data to the main thread to be sent to the graphics card.

On consoles you are not allowed to hang the game while it is loading, so if we're going to ship on them, this is required work.

The main thread also displays the loading screen, which by the way is inspired by the tesserakt engine logo of the trailers. I wanted something elegant and understated, and obviously something you don't get bored staring at.

Using OpenGL is not the best sometimes and some people prefer DirectX because they say the API is better designed, such as in this case where it would not have this problem. But using OpenGL allows me to run the same code on iOS and Windows in an easy way... and in general it is well supported everywhere. Now that Apple has just announced a way to run DirectX games on Apple hardware by converting its graphics commands to Metal, it might be worth thinking about switching, but for now I will ship with it, mostly to save development time.

After I got done with loading I finally worked on making lots of cool 4D objects to place in the levels, which has been on my list for years, so it is very exciting. Time spent on this is directly translated into cool 4D stuff, which feels good.

I also worked on ways to get the finishing funds that will allow us to finish the game.

I also started redoing the wall trailer with new footage. I will post it here soon. I had to dig up the original recordings from the archives, ahah.

I do want to say, it is hard for me to decide what to show of the game and in what order, because I have worked on this game for a long time without showing anything, so I am very far removed from the average "customer." I want a good ramp up in terms of visual complexity in order to not overwhelm, but now tons of levels are visually interesting in different ways... For example, even in an early level like the Wall, the tree is crazy, even if the level and wall itself are fairly simple...