Honestly, after months of fighting floaty falls in Unity, his illustration of how scale affects the perception of gravity got me to change my player from 5 units tall (Yikes! The ignorance) down to 1, scale the world down accordingly, fix the mass, and (alongside a simple fall speed multiplier script) my whole game's playing a lot snappier. Thanks, indie darling Bennett Foddy!
19:07 if you want to get friction right you need to understand that there is a fundemental difference between friction with the air/drag, and solid/solid friction. the latter does NOT depend on the magnitude of the velocity, but rather on the product of the friction coefficient and the magnitude of the force that it's beeing pushhed into the groud with. this means that this type of friction is only really feseblle to calculate with by the physics engine. It also means that objects that mave more mass(and therfore higher foce of gravity) are going to have more friction.
I've moved a prototype from Unity to Unreal, and while both use PhysX, I've found that I couldn't duplicate the same physics in Unreal as I had them in Unity. It was pretty interesting to see, and I'd rank Unity better in that regard. Things like increasing the frequency of physics calculations is handled differently in both engines, and perhaps this impacts things. How scaling affect collisions (which should result in some resulting physical forces) seems to be also handled differently as well (well, ignored in Unreal). This talk, factored in with another GDC talk that spoke about Rocket League using a simplified physics system, shows the importance of shaping the physics to suit your gameplay. Eliminating chaos, reducing complexity, or even making your physics a little less sensical, may be considerations for making your physics feel better.
i know that we have the technology for physics engines that are more complex than what is described here. for example collision resolution that considers all the contects at once instead of iterativly resolving each pair. I am doudbting weather mister foddy is oversimplifing or that game engines are lagging behind, or where laging behind at the time. Honnestly more confused after whatching this than I was before...
why don't you use a collision point instead that use sin cos to determine collision angle by giving each object a x and y each that sums to a single number that produce that pixel that two objects can colide with each other at any angle. just let each x and y vector sum track another at 360 degree rotation and collision will always be at angle in respect between the pixels. just decide distance from the center of the objects on how far apart you want the colission to happend. in 3d space you need a z vector to get the pixels to collide at any angle between each other in 3d space. let say the sum of vector of object one is let say x 20 y 20 sum = 40 if you do this on the other object as well, you already know they are in the same spot. if you want the collision to happend with more distance, you need to recalculate the x and y of each object and then sum them and that will give you the distance between them. then you say somthing like if number is in range between some number and other at some aproximity then collide thell each object to stop at direction of motion.
In all sorts of fun and terribly ways. Especially when you don't know what you're doing and somehow make every polygon of the character model react differently... The things I've seen that cube become horrify me, and I desperately want to recreate it so I can throw it into an eldritch horror game.
Reducing the physics timestep is not a solution and should NEVER be done. This causes the CPU to do A LOT more work and will cause the project to crash on all mobile devices and lower-end computers.
Honestly, after months of fighting floaty falls in Unity, his illustration of how scale affects the perception of gravity got me to change my player from 5 units tall (Yikes! The ignorance) down to 1, scale the world down accordingly, fix the mass, and (alongside a simple fall speed multiplier script) my whole game's playing a lot snappier. Thanks, indie darling Bennett Foddy!
19:07 if you want to get friction right you need to understand that there is a fundemental difference between friction with the air/drag, and solid/solid friction. the latter does NOT depend on the magnitude of the velocity, but rather on the product of the friction coefficient and the magnitude of the force that it's beeing pushhed into the groud with. this means that this type of friction is only really feseblle to calculate with by the physics engine. It also means that objects that mave more mass(and therfore higher foce of gravity) are going to have more friction.
One of the best talks here on GDC
I've moved a prototype from Unity to Unreal, and while both use PhysX, I've found that I couldn't duplicate the same physics in Unreal as I had them in Unity. It was pretty interesting to see, and I'd rank Unity better in that regard.
Things like increasing the frequency of physics calculations is handled differently in both engines, and perhaps this impacts things. How scaling affect collisions (which should result in some resulting physical forces) seems to be also handled differently as well (well, ignored in Unreal).
This talk, factored in with another GDC talk that spoke about Rocket League using a simplified physics system, shows the importance of shaping the physics to suit your gameplay. Eliminating chaos, reducing complexity, or even making your physics a little less sensical, may be considerations for making your physics feel better.
I'm going to need a reminder to read this reply in full...
the first GDC talk that I finished watching (definitely not because it's short). Really great work!
Just found that he is the creator of "Getting Over it". Nice one
I LOVE YOU BENNET FODDY! THANK YOU! This video helped me fix a gamebreaking bug in my game.
That was exactly what i needed. Gonna save this and rewatch when i get another physics problems
That horse portion looks like a machine learning problem. Let it find clusters of stable parameters, then you choose the one you like
So lost on a large majority of what he said, but still found it interesting enough to finish the video!
Really GOOD.
Thanks, Bennett
Great talk
16:07 Wouldn't the ball get absolutely crushed by the mass on top of it, assuming it wasn't indestructible?
Awesome, thank you!
Wonderful and useful.
Awesome talk! Loved it all!!!
Really cool talk, thanks!
i know that we have the technology for physics engines that are more complex than what is described here. for example collision resolution that considers all the contects at once instead of iterativly resolving each pair. I am doudbting weather mister foddy is oversimplifing or that game engines are lagging behind, or where laging behind at the time. Honnestly more confused after whatching this than I was before...
Thank you! Really helpful
This vid was really good and educational
Excellent talk!
why don't you use a collision point instead that use sin cos to determine collision angle by giving each object a x and y each that sums to a single number that produce that pixel that two objects can colide with each other at any angle. just let each x and y vector sum track another at 360 degree rotation and collision will always be at angle in respect between the pixels. just decide distance from the center of the objects on how far apart you want the colission to happend. in 3d space you need a z vector to get the pixels to collide at any angle between each other in 3d space. let say the sum of vector of object one is let say x 20 y 20 sum = 40 if you do this on the other object as well, you already know they are in the same spot. if you want the collision to happend with more distance, you need to recalculate the x and y of each object and then sum them and that will give you the distance between them. then you say somthing like if number is in range between some number and other at some aproximity then collide thell each object to stop at direction of motion.
I'm going to need a reminder to read this reply in full...
The controller is excellent idea, could save countless hours trying to tweak the physics.
I assume that the last platform and ball glitch are resolvable with soft-body physic.
great thanks!
Does anybody know what is the name of the game at 3:10?
Very useful talk! nice work
Why does Unity use linear drag and not quadratic drag? It doesn't seem like a hug performance impact.
Very handy, thank you!
Holy shit that's a great talk. Very insightful.
Thank you!
My GUT tells me I should learn more about dragging :)
9:21
TLDR: Physics Parameters can be subject to Animation Theory.
But you should watch the video and take notes as to how they directly relate.
thanks bean it
Mafia 3
How can you screw up 9.81m/s^2
In all sorts of fun and terribly ways.
Especially when you don't know what you're doing and somehow make every polygon of the character model react differently...
The things I've seen that cube become horrify me, and I desperately want to recreate it so I can throw it into an eldritch horror game.
Pp
Reducing the physics timestep is not a solution and should NEVER be done. This causes the CPU to do A LOT more work and will cause the project to crash on all mobile devices and lower-end computers.
It's _a_ solution, but often not a good one