Simulating a Chainsaw (Attempt 1) - Engine Simulator Devlog 5

EDIT - To those suggesting extremely simplistic "fixes" to this simulation: I appreciate the advice but the simulation isn't as simple as you're making it out to be. Trust me, the first thing I tried is "randomizing" the combustion events. It really doesn't work. I was going to include a clip demonstrating this but I cut it due to time limitations. The probabilistic model that I'm using IS random already. The periodic firing is due to other resonances in the engine and periodic variations in AFR and exhaust gas dilution. This oscillation is very complex and not trivial to adjust manually as it's an emergent physical phenomenon and not directly controlled by RNG. Throwing more RNG at the problem does not fix it in the way that you'd expect. I do have functioning ears and as I stated in the video, the end result is not a perfect recreation of the reference. My goal is to get videos out to you guys consistently which means that every video will show this project in an incomplete state. Some have suggested that this more recent work is somehow "worse" than the original Engine Simulator. I suggest rewatching those videos and realistically assessing how "true to life" those sounds actually are. Single cylinder engines are also significantly more challenging to generate convincing audio for. If I were to reimplement the same simulation in this video using the significantly less sophisticated fluid simulation from ES2D, I think you'd change your mind pretty quickly about how much better that software supposedly was. ORIGINAL COMMENT: I would have wanted to get the idle sounding closer to the real thing but the combustion model just isn't advanced enough yet to deal with the complex behavior of 2-strokes at idle. This definitely isn't the end of the chainsaw simulation journey though and I'll revisit this soon. Hope you guys enjoyed the video anyway and I'll see you all next time!

I always think "Wow this sounds near perfect" and then Ange comes along with the "now this does not sound quite right", proceeding to make it even better.

This sounds really good. I think that in idle there are also vibrations of the entire chainsaw and its components that contribute to the overall sound. So tuning and improving the fluid sim might not be enough to achieve the perfect idle sound.

Sounding good! With model airplane two-strokes, we call the misfiring at low speed "four-stroking" because it basically starts to sound like a four-stroke engine, firing on every other stroke as the AFR gets into the right window. I also think part of your problem is that your simulated muffler isn't muffling very well because of the ideal nature of the quasi-1D simulation. You could probably get some milage out of a "minor loss" model, which adds some pressure drop for things like sudden corners or pipe elbows. Don't let the name "minor loss" fool you, for a very short runner like that box muffler, the minor losses can easily outweigh the major (wall friction) losses. An easy way to test the corner dump case would be to arbitrarily inflate the friction factor in a very short length of pipe that expands from the initial cross section to the final one.

This is certainly the most I have ever understood about how 2 stroke engines work. This is also the most I've ever understood about how 2 stroke engines never want to work.

Great job with two-strokes, you never disappoint. Super excited for carburators in the future

honestly, this sounds better than every other game developers made their chainsaw.

Not as insanely true-to-life accurate as some of your car simulations, but still very impressive and nothing I could even dream of being able to do.

A major PITA with carburetors is the fact fuel doesn't like to get aerosolized when gas speeds are low (such as when idling), this means that fuel droplets will unpredictably pool up on surfaces before evaporating, leading to a more unstable AF ratio, which also contributes to the unstable idling. These problems tend to go away at higher RPMs due to higher gas speeds and from heat soak during load which helps to evaporate the fuel. Aim for a 0.8-0.9 lambda!

The idle from the simulation is actually very similar to how an old single 250cc snowmobile would sound for a few seconds after returning to idle from a rev!

I feel like all your work is going to change video games forever when sound simulation starts getting procedural. Do guitar strings next :D

Always insane what you're able to achieve by yourself

Sounds almost perfect! I'd say it still has a "plastic quality" to it, like it's modelled from plastic. You might want to introduce some rattling and grinding from the saw blade and other mechanical components. Also you should emphasize the characteristic tinny sound of the two stroke engine. With a convolution reverb I'd think.

Carburetor simulator would be very valuable to a lot of people, me included

I think this is almost kind of like the uncanny valley, just in sound form instead of visual. It's definitely very recognizable as a 2 stroke engine, but hard to put your finger on exactly what if off about it. Generically speaking I guess you'd just call it "synthetic" sounding, which is exactly what it is, but it's hard to express exactly what that means. Regardless this is super cool and I've enjoyed following along and am eager to see where it ends up. I can't imagine the amount of time is has taken to get here for just a pet project, kudos!

Very nice! For two strokes it's also important to note that many of them wouldn't, or would only barely run without well tuned resonator exhaust. It actually uses the exhaust wave bouncing off the end of the exhaust, then back again off the cylinder to draw a vacuum right as the exhaust port opens, drawing the burnt combustion products out of the cylinder.

You officially revolutionized the saw oscillator!

Wow, this is really impressive. Sure, the idle is a little off but the rest is pretty much on point. Qhat got me the most was the reoccurring of combustion when the AFR gets into the right window after the engine revs down. Keep up the great work!

Your videos always start with "eh it sounds too perfect, don't know how he could improve it though" and end with "are you sure this isn't just a real sound sample?"

This would have been inconceivable a few years ago even if it was just this one engine, let alone being able to make literally any engine within a few hours! Every time you upload it completely blows me away!