I've been kind of dragging my butt writing a 2018 retrospective. I looked at the 2017 one and it turns out I'm still working on #1 of my goals for 2018 and those goals have not changed for 2019. My unofficial goal was to have all my PC's desktops be screenshots of Miranda, and although it took some time, Miranda did finally take over a few months back.

I watched "Every Upcoming MMO & MMORPG 2019 And Beyond!" yesterday, and it was a little disheartening. I've been playing Guild Wars 2 for years, I've racked up hundreds of hours, it is spectacularly awesome and beautiful, but I'm sort of running out of things to do in it. I'm looking for something new to play and the current PC MMO market isn't looking very promising.

You're probably going to smirk at this comment from someone who's making a "games as a service" game, but GAAS isn't really working out for me as a player. Games are living far longer than they should. I'd be excited by a new Star Wars MMO, but there's already a Star Wars MMO, it came out in 2011 and shows no signs of being replaced. The Old Republic was good, I quite liked it, I played it for a few months, but at some point you're done as a player and want a new game. Publishers on the other hand, don't want to compete with themselves. "You want to play Star Wars? Here's the Star Wars game. It was really expensive to make." Similarly, The Lord of the Rings MMO was fun back in 2007 but it's still going! Yes, there is apparently a new LOTR game in the works at last.

I think part of the problem is that in the past, AAA MMO developers have just aimed too big. Every game doesn't need to be on the scale of World of Warcraft. I'd be quite happy with an experience that topped out at a hundred hours or so. That's still a big game for most people. Given that making content is a huge part of the cost of developing an MMO, maybe if developers didn't make so much of it, the games might not be so overwhelmingly expensive to make. Sure, if you have a massive hit then build an add-on, but maybe don't start with 12,000 hours of content. Amortize the tech cost over multiple titles.

Since I'm making a wish list, I'd also like some more options that aren't set in the middle ages. I loved The Secret World with its twisted take on the modern age. I was so disappointed when after I watched the amazing Cyberpunk 2077 Gameplay Reveal, I found out it was a single player game. I'm sure it will still be great, but I really like MMOs.

For a while I had been thinking that a Fallout style game would be really interesting as an MMO. It would have a futuristic open world, vaults to explore, loot to pore over, mutant beasts to flee or hunt. So I was quite excited when I heard Bethesda was making Fallout 76 because their production values are generally pretty good, and it sounded like the scope was a little smaller, but apparently it's just broken.

The original plan for Secret Lair Games was to make more, smaller MMOs. Maybe it'll get there.

I'm making a blog post of this in case anyone else ever runs into this problem so that the web won't let them down.

I was having a problem where wrong, or maybe old vertex data was being rendered after the program had been running for a while. I checked everything, and as it turns out everything was in fact working exactly as designed. To try and narrow down the problem, I spent a day building a frame debugger so I could isolate the vertex/index data that was rendering incorrectly. Once I had this working I noticed that the VBO ID's of the meshes that were rendering incorrectly were ones that had been recycled by OpenGL.

Vermillion Hardship is the eighth and final biome for Bold New World. I finished up most of the work for it last Monday, the rock formations are looking great, texturing looks really sharp, and I made several new microprops. I'm finally starting to feel comfortable making my own art in Blender. The last thing left to do is to find some new larger decorations since I removed all of the existing ones.

While I was thinking about that, I started on my QA pass. It didn't take long to find a big problem. The ability to move around Miranda smoothly has always been a huge priority for me, and at the start of the week, movement was very jerky due to an inconsistent frame rate. Luckily, optimization is just about my favourite thing to do as a programmer. You don't have to figure out how to make something new, you just have to find out what's slow, and then puzzle out how to make it faster. Oftentimes, much, much faster. So I turned on Miranda's frame profiler and found several functions which were sometimes taking a very long time to execute (as long as 250ms - a whole frame usually takes just 16ms.)

The cause of the 250ms delay was Miranda's model loader. Miranda loads 3D models on a worker thread so that doesn't affect framerate, but once they are loaded they have to be put into a tree for rendering which is done on the render thread. It turns out that occasionally a whole bunch of models finished loading at the exact same time which caused that frame to be very, very late. The fix was to simply put a timer on the render thread model loading code so that it can't run for more than 5ms. Anything that doesn't get done immediately when it comes in, gets done over the next few frames until all the models are processed.

The next delay was 180ms spent compiling the material script for each block of microprops. It turns out that too was running on the render thread, so I moved it to the already-existing worker thread that builds all the renderables for the props, easy peasy.

Next up was a 175ms spike processing compiled material scripts. Every 3D model in the game has a list of materials associated with it. Materials control the renderer to tell it how to render each and every triangle that makes up a 3D model. Each material has a set of passes, each pass has a set of shaders. Like the model loader, it happened that occasionally a lot of new material lists were queued up for processing on a single frame. The first step was to limit the number of those processed each frame, and that got the time down to about 75ms - still far too much. Miranda's renderer processes a full material list at a go and one of them was simply too long. When Miranda loads a new set of shaders for a pass, it checks to see if it has already loaded that particular set of shaders before. With all of Miranda's new art, there are far more shaders than there used to be, and the code to check for an existing shader got called a lot of times which was where the problem lay. The shader comparison function itself is pretty short, so first I removed a couple of memory allocations and simplified some string handling, but that only got me a few milliseconds. Miranda was doing a full source code comparison of the shaders, so next up was to change that to a comparision of a cryptographic hash of the source code. That got it down to about 35ms at the end of Friday.

So there's still work to do on that one, plus probably another half dozen to look at in the 30-40ms range which I'm hoping won't be quite so difficult. But even with just those improvements, by Friday afternoon movement was looking a whole lot better. Pretty slick in fact.

Maybe you read my tweet that I'm building a dynamic weather system, so you're probably wondering "Is he EVER going to finish this thing? How long can this go on?" Well, it's super-cool-looking, so there.

The whole thing came about because I was playing with GIMP to do the color correction for the biomes when I accidentally created an awesome looking screenshot that I just couldn't not put in the game. All I needed to make it work was a scheduler and a map to control where and when the effect occurs. Since I was in there anyway, I made it support all of the weather effects that were already in the game but were turned off because I didn't have anything to control them. Miranda now has a five hour day. The dynamic weather system also controls the sky.

Effects include a dust storm, an ion storm, super-hot weather and a fancier cloud shadow system. If you're looking for them, right now they only occur in the white area except for the cloud shadows which work everywhere. I'll add weather effects to the other areas in some happy future where I have tons of free time.

I meant to finish QA-ing this tonight, but instead I watched Deadpool 2. The world needs more R rated superhero movies.

Today is my 9th anniversary working on The Imperial Realm::Miranda so I decided to take the day, have a little party, eat cake and play the game I made.



I had 110 kills at the end of the afternoon which is pretty good.

And since you've all been so very patient and supportive all these months, here's a little peek at what's coming in Bold New World.



You can see the Silverswords and the volcanic vents at the bottom as well as one of the new rock formations and the sky. Like before, everywhere you can see, you can set up a base or drive your tanks.

The first biome, recently renamed The Perilous Precipices is complete and I am thrilled with how it has turned out. Making an area that is mostly black is quite challenging from an art perspective because black on black doesn't really look like much. I wanted to have some unusual plantlife in the area and I remembered the Silversword plant I saw at the top of the Haleakala volcano. I struck out at finding a pre-made model (although I did find a lot of lovely silver swords,) so I took one of my photos from my trip and created my own low-poly take on that plant.

I started the week with a last bit of work on the sky. There was a visible seam that needed fixing and I figured out a better way to position the sky. I also decided that it might be nice if the sky dome rotated every 5 hours so that the sky changes when you log in at different times of the day. That was probably the easiest feature I did this week.

The rest of the week I worked on finishing the first biome, what I'm currently calling The Breathtaking Precipices. This is known in the industry as a vertical slice. You get one part of the game complete so you work out any remaining bugs, evaluate it, and once it's perfect, you repeat the same process for all the other slices (in my case, 3 more biomes.)

This week I finished the last planned new feature for Bold New World. All that is left now is finalizing the art and bugfixing. It's been a while since I've blogged about the game. That is at least in part due to the fact that things were not really working out.

Atmospherics

A feature I really wanted to include is Bold New World is atmospheric effects: embers and ash falling after explosions, sparkles when harvesting, seeds drifting in the air. Atmospherics don't look like much in screenshots (although you can see them in this screenshot) but when they're moving, they bring life to the world.

I've been working on one of my stretch goals for the Bold New World update: a new sky. It might seem a little weird to work on the sky when RTS games focus primarily on the ground, but occasionally players do see the sky with the game camera, and I'm hoping the sky helps upgrade the marketing screenshots I make using the free-look camera.