Physics-Based Destruction! news

Welcome to this week's devlog for Assembly RTS, the physics-based real-time strategy game!

Hey strategy game enthusiasts! You may be wondering what I've been up to for the past two weeks?
I had just finished implementing the ability to 'train' units from factories when everything took a small turn...

perafilazof contacted me about needing some footage from Assembly RTS to feature in his next showcase video. A showcase video for upcoming RTS games releasing in the year 2024!

Update: Click here for the finished video!

It's not every day that you get an opportunity to have your game shown off by an established YouTuber.
I've been watching perafilazof's videos for many months myself.

While I wasn't expecting perafilazof to contact me for footage any time soon, it couldn't have come at a much better time. I've been working on visually appealing features for a while now, to get a Steam page for Assembly RTS up and running as soon as possible (as you may already well know)

There was just one key difference in this new situation I found myself in: The deadline for the footage was quite close. Naturally I sprung into high gear to produce some great footage by then.

I was still missing important features to show off Assembly RTS in a sufficient capacity.
Such important missing features included the following:

• Weapon firing and targeting systems
• Vehicle damage and destruction using physics
• Projectiles and more visual effects
• Modular vehicles (weaponry)
• At least one additional vehicle weapon
• An impressive-looking demo environment

Including animated mechs, sound effects or even infantry and airships would be an added bonus!

I was cutting corners wherever possible to make all this happen in such a short timeframe.
Cutting corners is the complete opposite approach to what I'm used to. I implement features to last.

But this time around, I knew very well that I'd have to make the needed features a reality quickly and haphazardly.
I reckon that I'll only keep around half of what I've completed during the past two weeks as it is now.

To make the physics-based destruction of vehicles possible, I make a seamless transition between the vehicle physics based on a 2D plane and 3D physics. This worked out surprisingly well.

You'll have to wait for the showcase video to see it in action. Spoiler alert, the transition is completely seamless and unnoticeable. It's like nothing happened (except for the fact that a vehicle just exploded).

This is made possible by simply copying the current velocity information from the 2D simulation over into the 3D simulation.

Unlike the physics-based destruction logic, the new projectiles and visual effects are currently very unoptimized, using ''Companion GameObjects".
Down the line I'll have to completely change how VFX work in Assembly RTS.

In theory, millions of particles shouldn't be problem for any capable GPU. Even the Steam Deck should have no issue... It's just a little tricky to get it right.

I got everything that I needed for the showcase done, and even managed to include the old mech!
The replay system I implemented a few weeks ago came in very handy while recording the footage.

And with slow-motion enabled, everything looks more epic (this has been objectively proven by science).

On the day of the deadline I recorded a replay of a match playing against myself. I quickly integrated a free look camera behavior into the game and captured the action from up-close. It went pretty swell.

A few weeks ago I ran Assembly RTS on the Steam Deck for the first time and it ran with over 30 FPS!

That's really surprising! Consider, the Steam Deck uses a Linux operating system and a small APU.
Thanks to Proton doing some crazy magic, Assembly RTS already has promising performance on SteamOS even though it was written for Windows.

I bet that the Steam Deck, paired with a small monitor, a mouse and keyboard are going to make a great setup to play Assembly RTS on LAN parties for example!

The only major optimization I'd need to do for the Steam Deck to reaching over 60 FPS is finding a better solution for the terrain shader. The terrain shader in my game is very complex and by far the most taxing on the GPU.

To drastically reduce the GPU requirements of Assembly RTS I would have to split the terrain into multiple meshes with standard physically-based shaders at runtime. Then I'd have to bake the complex shader output PBR channels into textures to apply to those meshes.

I expect Assembly RTS to garner a lot more attention soon, thanks to perafilazof's upcoming showcase video on YouTube.

Now would be a good time to introduce the Assembly RTS Discord Server!
I invite anyone interested in Assembly RTS to be one of the very first in our new gathering place.

Next up, I'm going to add infantry units to the game utilizing a highly optimized GPU animation system.
I also going to modify my old mech to use the same technology.