Decided to make a blog post, but interactive! As I've had some issues understanding the depth penetration algorithms of MPR I decided to make a series of blog posts about it, starting with the basics: Minkowski algebra & support function. Enjoy!

https://www.gameloomstudio.com/blog8.html

Playing around with GJK algorithm and managed to get visualization to the debug UI.

The 2 simplex case is without any conditions, 3 simplex with 4 and 4 simplex has 4 as well; where one condition is shared between the cases. All in SIMD so it's checking 4 pairwise collisions, but table lookup is sadly not SIMD.

Black line segment from middle: Search direction
Gray dot: Origin.
Black dot: Newly added point.
Red, green and blue dots: Previously added points.

The (almost) three month crunch is over, it's released!

The old (not so good) prototype has come from left image to right image.

Improvements include:

• Level count from a measly 20 to 60 (63 if you count costume levels).
• Major new gameplay features; including boxes.
• A saga map.
• Improved music and sfx.
• A demo version.
• Halloween costumes.
• Candies.
• A secret level (sorry, does not feature cows).
• Steam achievements.
• A native Linux build.
• A bunch of bug fixes regarding the asset compiler (especially with the custom Spine2D runtime and serializer).
  ... and probably a bunch more that I've forgotten about.

Get it here if you're into action puzzle games: https://store.steampowered.com/app/3137550/Halloween_Picket_Maze/

Not showing off coding super powers today, but level design and marketing - and today got confirmation something is working.

Our game level 8 was too hard, but instead of changing it I replaced level 7 with a tutorial.

The problem players had was "can I do this movement?", where the answer is NO. With a bit of tutorializing, and a bit clearer UI (red areas linger on longer) it still wasn't obvious, but it is clearer than before.

I've been contacting YouTubers to cover the game for several weeks now and I'm looking at about 11 % "success rate". I must say that YouTube mostly showing me content creators with over 100 K subscribers is really hurting me as there's no way I can reach those with this game (and lack of previous success stories).

But with the latest YT contact I got confirmation that the level is passable, and also that demo playtime is around 30 minutes (which is about what I was aiming for) https://www.youtube.com/watch?v=DsHc7umek9w

Of course, game is 100 % coded by me, assets by better half. Only external stuff is the music, sound effects (excluding ghost sounds, that's me speaking to the mic) and font.

It's been almost 10 years ago I started on my journey of refusing to use other ppl's bad code in favor of using my bad code. Many software releases have been done since then; 6 games, Bedit, an Android logger (because logcat is lagware), png-to-ico-converter, a better Windows aero version, a timer app for jams, and... even a program that changes your background every day - a lot of stuffs.

Also a lot of tech has been teched. My codebase runs on Windows (>=7), Linux, Playdate (to some extent), Android and web (with horrible emscripten integration). OpenGL, Vulkan, Steam, X11, WASAPI, Spine2d, Blender, ... there are integrations - by me alone (except 7zip library and whatever stb has to offer).

Today, we have our latest (hopefully not last) project announcement for the world; a Halloween game. Code by me and graphics & animations by my better half https://store.steampowered.com/app/3137550/Halloween_Picket_Maze/

I was going to improve the performance of our asset compiler by adding more threads, but instead I ended up writing a blog post regarding how I do CPU-based task management for our current project.

Have a read here: https://www.gameloomstudio.com/blog7.html