The early simulation without the rendering reminds me of the mantle and tectonic plate movements...there are so many possibilities with even a physics engine like this. Amazing work!
That phenomenon is called Rayleigh-Benard convection. It happens in the Earth's mantle, it happens with fire, and it happens with oil in a hot pan. Really interesting to see this emerging from the simple sphere collision model. With my lattice Boltzmann fluid solver, I can do this in 3D in real time: th-cam.com/video/lDsz2maaZEo/w-d-xo.html
I studied fire simulations at university and this is really cool! You're effectively doing a particle simulation of convection cells, and I'm betting if you scaled it up to 3D you could get some Rayleigh-Bernard convection cells going too! I'm really curious how long it takes to run and render these. Awesome work!
@@PezzzasWork I love it! Are you using SIMD? If not and you don't feel like dealing with C SIMD intrincis, you could a) use OpenCL to run it on the GPU (SIMT) and CPU (SIMD). I don't know how good the OpenCL compiler is at translating a SIMT program (kernel) to SIMD but I expect it to do rather well. b) There is also Intel's ISPC language to more comfortably write SIMD and SIMT programs. At least more comfortably than in C. Really, I think the way forward would be OpenCL or ISPC. There's of course also compute shaders in OpenGL and so on. And CUDA. Personally, I'd take the flexibility of OpenCL over CUDA any day but ymmv. Maybe I'm just burnt on it because I did CUDA for multiple months for my thesis.
Looks quite nice! However what I don't like is how the flame positions are essentially self-reinforcing with that model. This was especially visible with the ring as the flame positions just did not change. I think it needs some random variation with the heat source to combat that
@@PezzzasWork you can add viscosity and collision. you could make a tornado of fire, you could create the sun, you could make a flamethrower you definitely have interesting things to share
@@PezzzasWork Have you considered making a second channel for uploading said "less interesting" things (basically cool things with not enough depth to produce a video)
Great work! I think adding wind as a noise generated Vector field would make the fire Look even more realistic. Your simulation forms many straight columns in which the hot particles move upwards. These Columns don't are that clean and straight in real fires. I think that is because of the wind and air turbulence.
Really impressive. I've been a firefighter and dabbled in programming for a few years now, have always wondered how possible it is to recreate fire behaviour in code. You're taking early steps to something I don't think that has been accurately done before! Really curious to see more.
@@PezzzasWork One of the best examples I've seen of reasonably accurate fire propagation.. Add smoke as fuel, and some interaction with oxygen and then it would interact in very realistic ways. It's great mate! Better than you realize
The "trick" you used to get flying particles is the key to how the body of the fire as a whole works on a fundamental level. The chance that a particle is drawn on any particular frame is proportional squared to its temperature, and a particle only cools on a frame where it renders, otherwise it stays the same temperature and is invisible. You don't need to pick out a threshold of 3% or a particular range. The same rule applies to all of them. The only thing missing is that collisions between particles should convert some amount of their momentum into temperature, so it's not a straight transfer of heat, the sum of heat increases some from the lost velocity.
I discovered your channel today, I've been watching your older videos, and now there's a new upload too. Today is a good day. Thank you for your content, it's amazing! Both impressively interesting and beautiful.
I'm so glad I subbed to this channel recently. You make some very clever, innovative stuff! Those ring simulations remind me of the ending of Annihilation.
This is insanely cool, dude. I was so hyped by this and the evolutionary sim that I got called a nerd by my friend because of how much I was talking about it lol
Very cool! I don't think i've ever seen a Verlet fluid sim like this before. I was surprised to see those nice vortices show up around 1:00 in! I've messed around with stable fluids, PIC/FLIP, etc. but it has never occurred to me to use actual particle collisions like this. Inspiring work!
not a programmer but here for the aesthetics. your rendering and breakdown of fire taught me so much about how flame and convection work. a piece of art! thanks for sharing.
Now add a surface nearby, then have the heat transfer build up until the property 'flashpoint' is reached then the fire can spread? It is way more complicated than that but, it gets the point across I think. Awesome video, love this stuff, keep up the great work 👍🏻
This looks very cool but the fire seems to separate out into single strands and lacks general volume, I wonder if playing with masses of the objects based on temperature could help it fill up. Keep up the great work!
This is so freaking awesome. I'd love to see this in 3D, simulating an oil slick or something, but what would be really awesome, and probably require a supercomputer, would be simulating something like a small star (3D fireball, essentially). I'd love to see what it looks like at each order of magnitude more particles.
The collision response force is here depending on heat but I didn’t explored much the relation between collision interaction and objects state (temperature, pressure, speed etc ) there probably a lot to improve
you have no idea how much I love these videos. They are so inspiring to me as I really love simulation programming, but am hopeless at it myself. Please keep making more of these
Wow, just wow! For as awesome as the animation is, I want to say that your choice of music could not have been better! I'm a fan of any background music that fits its video, and boy did this one do that! You could have gone the more enthusiastic/energetic route, but the slow, almost pensive nature of this track really served the animation so, so well! Awesome track, awesome animation, awesome video!
The music has been made specifically for this video by a friend, if you like it you should really check his channel th-cam.com/channels/UWRUMnMP50v4x0CI2WMVHw.html !
In the case of adding velocity, it would've been nice to see a more realistic interpretation, that being that heat rises, not just increasing in velocity. When looking at the circular forms then, this would result in a fire which wraps around the form to reach the upper limits of the container, rather than spewing out in all directions.
so nice ! for the sparks, i would have randomized a percentage of hot particles with more speed. But if you can do it in 3D, you have a better fire than i've seen in every games.
i think something you can do you the fireball is instead of hot particles go straight away from the center of mass, have hot particles fly perpendicular or at an angle to the direction toward the center of mass, making a spiral effect and make it more like a fireball
This reminds me of the fiery backgrounds of videos for "reverb" or "party" versions of songs. I imagine that you could change a parameter real time in sync with a characteristic of a song to make a dancing fire! For example, the color changing with time, and the upward force of the heated particles being in proportion to the songs volume. Maybe an entire flaming DJ-frequency-bar-chart-thingy (IDK what they're called) is possible.
you could really do a lot with this, such as making some sort of object like a house you can burn, or just wood or trees, hell maybe even an iron beam or something than you can melt.
One notable thing missing is the smoke. Fortunately you don't need to do anything different to simulate the smoke since the bright flame and the smoke are the same thing: the burned out gasses rising up because of heat. Just add cold particles all the way up and make sure the heat transfer is much slower the colder the particle get so the smoke keeps rising up because the temperature difference remains much longer.
Cool! Kudos on your channel. I like how you show the simple rules that let fun complex behavior emerge. For this you should look into non-monotonic coloring. What we see as the "white hot" regions of a typical hydrocarbon fire are actually radiating soot particles that are only present in an intermediate temperature window, not the hottest parts of the actual flame itself.
This must have been a hoot to tinker with. I just finished with boids simulations and just got into 3d physics and it's a whole world to rabbit hole into with so many experiments, I love it! 👍👍👍👍
I wanted to say that you can add resistance to the particle to create surface fire that sticks and spreads to it's neighbours. The resistance is high for metals but low for wood, lowest for things like cotton or hay. Love the simulation result for the ring. It looked like an eclipse! Can you also render fire in the middle and give it a gravity constant so the flames from the ring don't go that far? You'd be making the sun then probably. Be careful not to turn it into a blackhole!
Oh my God this is fantastic! The insane realism that the bloom adds is really amazing and surprisingly simple. Really great work!!! I wonder if you could a 3d cube version 🤔
Beautiful! I would have used some kind of fluid sim for something like this, but this is simpler and gets better results than I could. Incredible work.
I think I'd kill off invisible particles and instead spawn new ones from the sides of a fire "source" that also heats. It's pretty obvious that the falling cool particles get reused and cool the bottom of the fire as they're falling, if realistic fire was the goal that is.
This is unreal. I always wanted to dabble in the dark (cool) arts and this is a perfect way to start! Thanks you well earned a subscription I hope you earn many more! (Sorry for bad English not my first language)
That's amazing! Especially medium and campfire just looked so realistic The one thing that I found odd: You used gravity to hold the fireball and rings together, so they were basically in 0g. I don't actually know if fire with more force than a candle would behave the way yours did, so it may be accurate, nobody wanted to test that aboard the ISS, I guess^^ But I would love to see the fireball and rings under gravity (:
As always, this is absolutely beautiful. The work you do is stunning, and not just visually. What language do you code in? Have you ever considered doing coding tutorials? I'd love to make something like this myself, but I have no idea where to start.
Amazing work, It looks incredible! Have you made this code open? Also, I think it might be interesting changing the container in which the convection cells are placed (like in a convex or concave bottom)
That is so cool. Can you please try to make a simulation of a volcano. At the 50 Second mark in the video the way the particles moved remind me of a volcano. I apologize about spelling and grammar. I have one recommendation can you please add a voice over or Meritor to your videos. Not everybody can see the words very well. I’m partially blind. Hope you have a Great day. Keep on learning.
Hello, I tried to make the collision response from your previous tutorial multithreaded and had no luck; is yours multithreaded, and can you share any implementation details for that please? Love your videos, this is amazing!! Oldschool demoscene fx fire rendering is what got me into graphics programming, this is the next level of that :D
This is sweet! Incredible work. If you do not mind me suggesting, I think it would be cool if based on the temperature the screen would warp slightly. Seriously impressive work.
The early simulation without the rendering reminds me of the mantle and tectonic plate movements...there are so many possibilities with even a physics engine like this. Amazing work!
Exactly and the air convection currents and marine water movement around magma it’s so interesting
@@sergioibarra3461 I guess I never thought about how "hot air rises" applies to heat in general. Incredible generalization!
That phenomenon is called Rayleigh-Benard convection. It happens in the Earth's mantle, it happens with fire, and it happens with oil in a hot pan. Really interesting to see this emerging from the simple sphere collision model.
With my lattice Boltzmann fluid solver, I can do this in 3D in real time: th-cam.com/video/lDsz2maaZEo/w-d-xo.html
The circle ones remind me of solar eclipses
same
I studied fire simulations at university and this is really cool! You're effectively doing a particle simulation of convection cells, and I'm betting if you scaled it up to 3D you could get some Rayleigh-Bernard convection cells going too! I'm really curious how long it takes to run and render these. Awesome work!
The title does say real-time, so I'd imagine that it's actually written as a shader to run on the GPU
The simulation is running on the CPU, but it could certainly be ported to a shader for massive speed up
@@znefas Lol good point, I got so distracted with the visuals I didn't read the title. I'm used to more science-focused simulations that take days
My OpenCL lattice Boltzmann CFD software can simulate real time Rayleigh-Benard convection in 3D: th-cam.com/video/lDsz2maaZEo/w-d-xo.html
@@PezzzasWork I love it! Are you using SIMD? If not and you don't feel like dealing with C SIMD intrincis, you could a) use OpenCL to run it on the GPU (SIMT) and CPU (SIMD). I don't know how good the OpenCL compiler is at translating a SIMT program (kernel) to SIMD but I expect it to do rather well. b) There is also Intel's ISPC language to more comfortably write SIMD and SIMT programs. At least more comfortably than in C. Really, I think the way forward would be OpenCL or ISPC. There's of course also compute shaders in OpenGL and so on. And CUDA. Personally, I'd take the flexibility of OpenCL over CUDA any day but ymmv. Maybe I'm just burnt on it because I did CUDA for multiple months for my thesis.
Looks quite nice! However what I don't like is how the flame positions are essentially self-reinforcing with that model. This was especially visible with the ring as the flame positions just did not change. I think it needs some random variation with the heat source to combat that
Yes I didn’t experiment heat source variation, this is a first shot there is plenty to improve
@@PezzzasWork Does this mean we can expect more fire stimulations? 🔥
@@jfk_the_second maybe! it depends if I have interesting things to share :)
@@PezzzasWork you can add viscosity and collision. you could make a tornado of fire, you could create the sun, you could make a flamethrower
you definitely have interesting things to share
@@PezzzasWork Have you considered making a second channel for uploading said "less interesting" things (basically cool things with not enough depth to produce a video)
Great work! I think adding wind as a noise generated Vector field would make the fire Look even more realistic. Your simulation forms many straight columns in which the hot particles move upwards. These Columns don't are that clean and straight in real fires. I think that is because of the wind and air turbulence.
Really impressive. I've been a firefighter and dabbled in programming for a few years now, have always wondered how possible it is to recreate fire behaviour in code. You're taking early steps to something I don't think that has been accurately done before! Really curious to see more.
Thank you! I don’t know if my approach could be used for realistic simulations since it is very simple
@@PezzzasWork One of the best examples I've seen of reasonably accurate fire propagation.. Add smoke as fuel, and some interaction with oxygen and then it would interact in very realistic ways. It's great mate! Better than you realize
Wow that's really cool to hear from an expert in the field..! Thank you for your service Gaz. Big respect
@@Shrooblord xd
The "trick" you used to get flying particles is the key to how the body of the fire as a whole works on a fundamental level. The chance that a particle is drawn on any particular frame is proportional squared to its temperature, and a particle only cools on a frame where it renders, otherwise it stays the same temperature and is invisible. You don't need to pick out a threshold of 3% or a particular range. The same rule applies to all of them. The only thing missing is that collisions between particles should convert some amount of their momentum into temperature, so it's not a straight transfer of heat, the sum of heat increases some from the lost velocity.
Higher temperature shouldn’t explicitly cause an upward force, but a reduction in particle mass
I love how you iterate on the emergent properties and behaviour of the simulation and see where it takes you.
I discovered your channel today, I've been watching your older videos, and now there's a new upload too. Today is a good day.
Thank you for your content, it's amazing! Both impressively interesting and beautiful.
I DONT UNDERSTAND HOW THIS TH-cam CHANNEL ONLY HAS 43K SUBS!! this is so underated
I have no idea how this is working and where to start understanding it, but it looks amazing!!
So beautiful seeing it all come together step by step
I swear, the music on these videos + the mesmerizing visuals is second to none... I've watched these like several times while implementing along with.
at some points in the beginning it looked like magma - with tweaking this could become a really cool volcano simulation
It’s funny you mention this because I was trying to do a volcano simulation when I saw this and decided to turn it into fire
Been following you for a couple years and every video you amaze me with some new computational technique. Awesome stuff, keep it up!
Man you are genius! I am doing warehouse programming at work and now trying to reproduce your results at home as a pet project. Thank you!
I'm so glad I subbed to this channel recently. You make some very clever, innovative stuff! Those ring simulations remind me of the ending of Annihilation.
There's something very enchanting about this video, the combination of music and visuals is amazing
I was waiting for the Eye of Sauron shape
I love your outros so much.
This is insanely cool, dude. I was so hyped by this and the evolutionary sim that I got called a nerd by my friend because of how much I was talking about it lol
shut up
nobody asked
nobody cares that you have friends
@@jazzling You're right, nobody asked you.
wow, it doesn't get more cooler than this, lads.
Very cool! I don't think i've ever seen a Verlet fluid sim like this before. I was surprised to see those nice vortices show up around 1:00 in! I've messed around with stable fluids, PIC/FLIP, etc. but it has never occurred to me to use actual particle collisions like this. Inspiring work!
very realistic! i love it! :)
not a programmer but here for the aesthetics. your rendering and breakdown of fire taught me so much about how flame and convection work. a piece of art! thanks for sharing.
I just found your channel and I think this stuff is super interesting. Thanks for sharing it!
Now add a surface nearby, then have the heat transfer build up until the property 'flashpoint' is reached then the fire can spread? It is way more complicated than that but, it gets the point across I think. Awesome video, love this stuff, keep up the great work 👍🏻
i really do love how your outro is to somehow put your pfp into the simulation, that's amazing
I cant look away from the rings and fireballs. Feels like a dance
This looks very cool but the fire seems to separate out into single strands and lacks general volume, I wonder if playing with masses of the objects based on temperature could help it fill up. Keep up the great work!
This is so freaking awesome. I'd love to see this in 3D, simulating an oil slick or something, but what would be really awesome, and probably require a supercomputer, would be simulating something like a small star (3D fireball, essentially). I'd love to see what it looks like at each order of magnitude more particles.
Did you consider making the size and/or restitution/rejection force of collisions change based on heat too, since heat increases pressure?
The collision response force is here depending on heat but I didn’t explored much the relation between collision interaction and objects state (temperature, pressure, speed etc ) there probably a lot to improve
you have no idea how much I love these videos. They are so inspiring to me as I really love simulation programming, but am hopeless at it myself. Please keep making more of these
I thought this was going to be a ‘Stable Fluids’ paper like implementation, very impressed with your pure particle approach !
Wow, just wow! For as awesome as the animation is, I want to say that your choice of music could not have been better! I'm a fan of any background music that fits its video, and boy did this one do that! You could have gone the more enthusiastic/energetic route, but the slow, almost pensive nature of this track really served the animation so, so well! Awesome track, awesome animation, awesome video!
The music has been made specifically for this video by a friend, if you like it you should really check his channel th-cam.com/channels/UWRUMnMP50v4x0CI2WMVHw.html !
!
@@PezzzasWork omg i was absolutely mesmerized when I watched it first, went to bed straight afterwards. Such calming music
You should combine this simulation with the "star" simulation but more pressure = more heat
When I saw the rendering of this, instead of thinking of fire I thought of magma eruptions.
In the case of adding velocity, it would've been nice to see a more realistic interpretation, that being that heat rises, not just increasing in velocity. When looking at the circular forms then, this would result in a fire which wraps around the form to reach the upper limits of the container, rather than spewing out in all directions.
Some looks likes lava, the lava breaks through creating cool visuals.
This is oddly therapeutic and relaxing
looks great! the way you made the sparks was very smart.
I really gotta study this someday
It's very satisfying seeing how this is made step by step
Always love the quality of content from this channel
so nice ! for the sparks, i would have randomized a percentage of hot particles with more speed. But if you can do it in 3D, you have a better fire than i've seen in every games.
i think something you can do you the fireball is instead of hot particles go straight away from the center of mass, have hot particles fly perpendicular or at an angle to the direction toward the center of mass, making a spiral effect and make it more like a fireball
This reminds me of the fiery backgrounds of videos for "reverb" or "party" versions of songs. I imagine that you could change a parameter real time in sync with a characteristic of a song to make a dancing fire! For example, the color changing with time, and the upward force of the heated particles being in proportion to the songs volume. Maybe an entire flaming DJ-frequency-bar-chart-thingy (IDK what they're called) is possible.
Love your work, very inspiring!
Thank you!
Found your channel through youtube reccomendation to your previous video. Hope algorithm keep working in your favor. Great work
you could really do a lot with this, such as making some sort of object like a house you can burn, or just wood or trees, hell maybe even an iron beam or something than you can melt.
Your videos never cease to amaze me, great job as always
One notable thing missing is the smoke. Fortunately you don't need to do anything different to simulate the smoke since the bright flame and the smoke are the same thing: the burned out gasses rising up because of heat.
Just add cold particles all the way up and make sure the heat transfer is much slower the colder the particle get so the smoke keeps rising up because the temperature difference remains much longer.
This video was *fire*. I would have liked to see an eye of sauron in the shape variations though. That would have been awesome.
I love the ending most of all. Thanks for making this.
this channel deserves so much more subscribers, amazing content.
This looks very nice. I wonder if creating heat distortion as well would make it look even better to
And it's so smooth! I absolutely love this! Thank you for sharing! I'm learning programming and want to get into physics simulation. This is awesome!
Really cool!! I want to get into stuff like this, are you using C++? I’ve been learning it and it’s so much better than python.
You deserve mkre subscribers, all your projects are so well made and it really encaurages you to learn C++
It looked like you simulated magma for a while there. Outstanding work.
This both looks really cool, and like it would be fun to play around with the rules and see what you can get out of it.
Mans doing god's work
Best simulation I've ever seen, I can even smell the fire, wait..
Cool! Kudos on your channel. I like how you show the simple rules that let fun complex behavior emerge. For this you should look into non-monotonic coloring. What we see as the "white hot" regions of a typical hydrocarbon fire are actually radiating soot particles that are only present in an intermediate temperature window, not the hottest parts of the actual flame itself.
Excellent work, Pezzza. Your videos are always delightful
This must have been a hoot to tinker with. I just finished with boids simulations and just got into 3d physics and it's a whole world to rabbit hole into with so many experiments, I love it! 👍👍👍👍
Amazing work, also, the music brought me back to my EvE Online days
I wanted to say that you can add resistance to the particle to create surface fire that sticks and spreads to it's neighbours. The resistance is high for metals but low for wood, lowest for things like cotton or hay.
Love the simulation result for the ring. It looked like an eclipse!
Can you also render fire in the middle and give it a gravity constant so the flames from the ring don't go that far? You'd be making the sun then probably. Be careful not to turn it into a blackhole!
Oh my God this is fantastic! The insane realism that the bloom adds is really amazing and surprisingly simple. Really great work!!! I wonder if you could a 3d cube version 🤔
Extremely well structured video! i loved the way you did everything in easy-to-understand simple steps.
Always a pleasure to watch your stuff man.
Black SUN is magnificent!
Awesome video as always! Keep up the good work
Beautiful! I would have used some kind of fluid sim for something like this, but this is simpler and gets better results than I could. Incredible work.
I think I'd kill off invisible particles and instead spawn new ones from the sides of a fire "source" that also heats. It's pretty obvious that the falling cool particles get reused and cool the bottom of the fire as they're falling, if realistic fire was the goal that is.
Pezza's work on fire 🔥
This is unreal. I always wanted to dabble in the dark (cool) arts and this is a perfect way to start! Thanks you well earned a subscription I hope you earn many more! (Sorry for bad English not my first language)
3:50 That would be a cool cinematic shot for a survival indie game.
That's amazing! Especially medium and campfire just looked so realistic
The one thing that I found odd:
You used gravity to hold the fireball and rings together, so they were basically in 0g.
I don't actually know if fire with more force than a candle would behave the way yours did, so it may be accurate, nobody wanted to test that aboard the ISS, I guess^^
But I would love to see the fireball and rings under gravity (:
Thats amazing!
You motivate me to do more simulations. You are a great inspiration.
That is so cool man, must have taken forever to render that, good job!
TH-cam recommended showing me more cool random videos. Interesting topic!
One thing that might make this look cooler:
Decrease the heat required to rise, and make hotter particles rise faster.
I love how you often add your channel picture/ logo to the end.
Also, your timestamps are perfectly round (multiple of 5) numbers.
Absolutely beautiful
Wow I love it! Your simulations are always so inspiring for me to code stuff like this as well
this channel really is something else! really inspiring
This guy deserves more subscribers. Here take me 🎣
0:58 that's Rayleigh-Benard instability!
As always, this is absolutely beautiful. The work you do is stunning, and not just visually. What language do you code in? Have you ever considered doing coding tutorials? I'd love to make something like this myself, but I have no idea where to start.
I believe he used C++, but the language isn't important. You've gotta know the physics behind this.
@@verified_tinker1818 Well, yeah- but you also gotta know how to express the physics in code to be able to actually use it, y'know?
This is incredibly beautiful. Well done
Amazing work, It looks incredible! Have you made this code open? Also, I think it might be interesting changing the container in which the convection cells are placed (like in a convex or concave bottom)
That is so cool.
Can you please try to make a simulation of a volcano.
At the 50 Second mark in the video the way the particles moved remind me of a volcano.
I apologize about spelling and grammar.
I have one recommendation can you please add a voice over or Meritor to your videos. Not everybody can see the words very well. I’m partially blind.
Hope you have a Great day. Keep on learning.
This is incredible.
Hello, I tried to make the collision response from your previous tutorial multithreaded and had no luck; is yours multithreaded, and can you share any implementation details for that please?
Love your videos, this is amazing!! Oldschool demoscene fx fire rendering is what got me into graphics programming, this is the next level of that :D
Incredible. Where are all the views and subscribers? You deserve many more.
This is sweet! Incredible work.
If you do not mind me suggesting, I think it would be cool if based on the temperature the screen would warp slightly.
Seriously impressive work.
Allof these would make a great wallpaper for wallpaper engine
Would be a nice interactive wallpaper. Fire centered on your mouse.
Breathtaking 🤩🤩
i like how on occasions medium ring becomes pentagon