Це відео не доступне.
Перепрошуємо.
Burning Down an Entire Virtual Forest! 🌲🔥
Вставка
- Опубліковано 21 чер 2021
- ❤️ Check out the Gradient Dissent podcast by Weights & Biases: wandb.me/gd
📝 The paper "Fire in Paradise: Mesoscale Simulation of Wildfires" is available here:
computationalsc...
🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Aleksandr Mashrabov, Alex Haro, Andrew Melnychuk, Angelos Evripiotis, Benji Rabhan, Bryan Learn, Christian Ahlin, Eric Haddad, Eric Martel, Gordon Child, Ivo Galic, Jace O'Brien, Javier Bustamante, John Le, Jonas, Kenneth Davis, Lorin Atzberger, Lukas Biewald, Matthew Allen Fisher, Mark Oates, Michael Albrecht, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Ramsey Elbasheer, Steef, Taras Bobrovytsky, Thomas Krcmar, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi.
If you wish to appear here or pick up other perks, click here: / twominutepapers
Károly Zsolnai-Fehér's links:
Instagram: / twominutepapers
Twitter: / twominutepapers
Web: cg.tuwien.ac.a...
#gamedev
"Now to compare how accurate the simulation is we burned down a real forrest"
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA!
Am I funny?
easy, we just need to wait for the next gender reveal party.
@@liamhansolo1464 you were very funnny
@@liamhansolo1464 it mostly depends if you would still think it is funny or not when you grow at least one or more years into the future and look back at this comment you made i guess /nm /lh
First some trees, now an entire forest? He’s gone mad I tell you!
He is going to partner with Kurtzgestat and destroy the world in some weird way just to see what happens
Next, the entire world!
If he keeps going like this, there’ll be nothing left for his “papers” 😂
Brazil is already doing that in real life.
1 minute in
congrats
KEKW
Lol
in real time with no delay
😲😳🤣😭😲
Next paper be like: let's simulate Ragnarok! What a time to be dead!
Thor will smite you with 120k hammers in real time
@@notatakennick What a time to be flattened!
Finally, a good enough excuse to make a sequel to Far Cry 2 !!!
lmao
Farcry 3?
@@cadenhood Not really. The fire mechanics, fire spreading system, and vegetation physics are all worse in Far Cry 3. See comparision: ua-cam.com/video/88PHHEziQtU/v-deo.html
I'm a firefighter. I don't think I can get my papers near this. That would be dangerous... 😂
I don't know how the team that made this has their papers still in tact.
2:04 Flammagenitus is my new favorite spell.
It would be a lot more impressive if the ground vegetation was taken into account as well, maybe next paper
None of this would have happened if they had raked the forest.
a model like this is supposedly so advanced and yet they have not accounted for raking parameters? amateurs.
Forests are supposed to burn just like nagualdesign said. It may seem weird, but that's reality because eventually for life to continue it has to reset the crowded areas.
@TomsterBG you do be sounding like Thanos ngl
@@nagualdesign @tomsterBG it's a joke referencing how Trump blamed California's wildfires on their forests "not being raked"
@@nagualdesign As far as i know, too much debris on the ground causes the fire to burn hotter and burns the trees to nothing. as opposed to letting nature fires happen to get rid of the underbrush, or for areas near homes, doing controlled burns to do it proactively, i suppose 'raking' is a way to do it without fire. but when people prevent all fires and only let it build up, it just burns so hot that it burns everything. As per usual there is never any 'facts' just BS from the media so they become misinformed and entrenched in whatever views their side has.
programmer: *starts up the burning simulation*
*sky behind him turns orange*
programmer: *hol up*
"What do you mean, it wasn't a simulation?!"
@@jierueichang he's just mean
"Stargate Atlantis flashbacks"
Ender's Game lol
@@jierueichang it always has been
And just like that, this went from eye candy to something with serious utility in forest management.
As cool as this would be in a game, I'm more interested in seeing its potential to simulate wildfires to better combat them.
🔥 Possibly the only environment where setting fire to trees is a good idea. 🔥
In order for life not to get too crowded there's regular fires all over every day and they're left untouched on purpose. They're only beign fought when a village or city is in danger.
@@tomsterbg8130 sometimes fires are even started on purpose to burn down dead, dry wood carefully, before a wild fire starts and causes chaos and so on
natural fires are good, and sometimes controlled fires are too
Your channel is very close to the 1 million subs goal, I'm waiting for that day to come.
Does it include sparks that may travel further than the flames, helping spread the fire?
@Ezequiel Ciamparella Hmm. That doesn’t sound very detailed. They should consider simulating solar flares and asteroids in the next paper, too.
@Ezequiel Ciamparella He probably meant embers lol
Lmfao these replies
Can't wait for the ground truth experiment to check the accuracy of the model.
I can’t wait for this to be used in games and cgi!
This would be very cool in games like Cities Skylines
@@Fisch0 agreed
Dude I can barely understand what is going on sometimes, but these videos are really addicting :D I love seeing how new ideas and techs have come to fruition and your wonderful examples actually makes it so much easier to digest :D So much cool and awesome stuff around here I'm glad you showed up in the feed :D
There is one flaw in this simulation in that it doesn’t take into account spot fires (separate fires caused by burning material being flung into the air by the thermals caused by the original fire and travelling large distances through the air, setting fire to new areas when they land). Growing up in Australia they were one of the main things we were told were dangerous about bush fires.
I love your videos! I'm so excited to see this technology and others implemented today!
Finally! A safe way to quench my pyromaniac desires
Thats amazing work and lets hope it can help stem the flow of wild fires in future when coupled with up to the minute weather data!
95...
Me : 95 seconds for 12k trees? That's neat.
*...miliseconds*
Me : (Drop the jaw)
This is so fascinating! I hope one day we can accurately simulate most aspects of reality.
Everybody gangsta until he says „hold onto your papers“
that poor forest never even stood a chance
So cool! I must show this to my mom. She studies ponderosa pine forests after fires and as even _published a few papers_
What a coincidence! My city got the first lighting fire I've seen above the artic circle yesterday. This would be the second time in 20 years I've heard of lighting at all where I'm from, and the first wild fire I've ever seen. Thankfully the fire crew got aid from a massive rain cloud passing by.
Really held on to my papers the entire time for this one. Truly amazing!
Pq esse video ta na playlist do jornal não rir
I was a firefighter for 17 years and spent quite a bit of time on wildfires and they are really missing out on a few things. One, while they somewhat addressed ladder fuels in terms of vertical spread, the distances matter, fire can travel across the canopies of trees without ever burning down to ground cover and the inverse can happen as well, a fire moves through the floor but doesnt climb. They are not simulating runs with gusting winds, the way their fire spreads does not look entirely correct, nor does it look like the saddle in the hill causes the blowtorching effect we see in actual fires. While this is headed in the right direction, it needs a lot of work to be useful to create real world in silico demonstrations of concept.
That can come in handy in a lot of cases. Video games, wild fire predictions maybe?
Now. Lets simulate. Entire forest. With this paper. - Interesting usage of comma
Virtual tree: Exists.
Simulated fire: "And I took that personally."
Seems like we’re really close to this tech being possible in real time! There would be a lot of games this kind of interaction would be awesome to see in, from Sim City to Breath of the Wild.
This simulation took everything into account, and still at at 95Ms refresh speed. If you reduce the accuracy of the trees, say, by only using a few volumes for leaves rather then each one individually, and reduced the accuracy of the airflow simulation it should be possible to get a really good game interpretation to close enough to real time.
This will be a great asset to Forest Service and Land Management all across the world if commercialized successfully.
Work like this shows the potential scope for machine learning is far greater than we've seen so far. Exciting!
Narrator: What a time to be alive!
Burned tree: yes
Flamegenitous clouds sound like something I don't want to experience
I wouldn't want the clouds to experience it either.
Another next step to take would be the implementation of ash and embers. Embers can travel huge distances and start secondary spot fires, which is always important to be aware of when trying to predict the movement of a wildfire.
Idea: A neural net that does convergence on any neural net. Like for example start small with one tree burning, extend to a group of trees and train a neural net on it, scale up to a forrest, to multiple forrests, to a country and so on. For each step a new neural net.
4:42 That virtual forest isn't the only thing on fire, I can't believe you did us like that...
I live on the west coast of the United States, last year obviously we had some devastating wildfires. I hope that technology like this will allow governments to test out solutions to totally prevent these from happening in the future.
But are they devastating, or are they supposed to happen naturally from time to time.. A forest fire has its positive sides too, among the most important is wiping out/keepin in check parasites. It also wipes out flora that is not native to the region, and returns nutrients to the ground. Usually the fire passes so quickly that the strongest trees survive, and the forest quickly recovers to its former glory.
@@franticsledder I totally agree that fires do need to happen, I'm not saying we should just let debris and stuff pile up, but instead that we can find out ways to allow nature to take its course without having such a severe impact on those living in the area. I've driven through a few areas that burned, that I've been to beforehand. It's absolutely heartbreaking to see just the amount of widespread destruction of people's homes and such. Not to mention all the people that are killed by the fires.
Would love to see this tech in a city building game.
All i could think was city skyline 2
Still not fast enough to be in a game, probably by an order of magnitude.
There are many things you need to calculate and all, added together, need to be processed in real time. The problem is worse with city building games, when you usually can speed up time several times (You'll probably need to change the model to a simpler version for this, adding to complexity). While this is nice, the results are mostly esthetical and would take too many resources from other processes that affect gameplay directly.
@@bronzehd6212 I think the number of trees would be higher in a game. 95 ms per physics frame would give you 10.52 fps, which is very low for a game
This guy sounds like he's constantly whispering... Love it keep up the great work :)
Now to add Fire hawks! They are notorious for intentionally spreading fires by snagging burning twigs and dropping them in places to force prey out of hiding.
Neat. I've been toying around with the fire simulation in Teardown and a forest fire has been my ultimate goal to simulate. The problem I run into is the fire not spreading fast enough or through embers. This should be pretty neat.
this could be used in a game as a teaching aid
from 1 tree to 120k is insane. That's a lot of trees and amazing simulation.
I was assuming seconds to render...but milliseconds, wow!
You are about to reach 1 mln subs 👍👍👍
Time to put that simulation into practice!
pq q isso tá na playlist do jornal não pode rir?
TBM N ENTENDIKKKKK
they should map out popular forests and calculate which are the least fire resistant, and inspire change? idk where im going with this but this can be really useful. maybe provide the software to governmental agencies, which can distribute per state and city, where the mapping and calculation will be faster and more wide spread? idk
This is insane , my papers almost caught on fire but i was able to stop them , the real world applications of this are amazing
Incredibly relevant paper, given that there are already dozens of fires in western USA right now. Would be interesting to select any of the fires, and see if you can predict how it will spread.
Virtual forest? Haha, im 4 paralel universes ahead of you
When this gets implemented in games it'd be so awesome
Its all coming together
ive been waiting for something like this i want to simulate fire on bonsai tree models for a video
I live in a region that is very susceptible to forest fires. Last year was particularly bad for us. Hopefully, simulations like these can assist our firefighters to improve their strategies to combat and curtail the spread of these dangerous fires!
This would be incredibly useful if you could have a model of a location that was highly accurate and used real time weather information to simulate current fires, and provided that info to wild land firefighters.
Just in time for fire season!
I actually dropped my papers.... miliseconds! Fire dynamics in video games would be quite a thing and it looks like that's within reach.
that's amazing
Is it actually comparable to the single tree sim though? Like, if you burned a single tree, will it work as well with this technique as with the old one? I'd think they used more approximations to make this work so well
Researcher one: how should we index our trees?
Researcher two: R-trees
Researcher one: yes, our trees
What a time to be alive!
This could be used to prevent disasters like the Camp Fire by running simulations and showing key evactuation hurdles
why is nobody talking about this this is amazing
This is going to sound weird, but I think this would make a great game. Could also bring awareness to how easily forest fires spread.
I feel like these kinds of simulations could be extremely useful to help fight forest fires in Australia or the US, and in general it could give some great insight for firefighters everywhere
Finally, a way to do gender reveal parties over the internet!
I forgot to hold onto my papers and now they're engulfed in flames
To make a more complete sim they need to add embers and ash which can also spread fires. while they mention moisture content there is a lot more they still have to add not only trees but smaller plants and the leaves on the ground. There are many places that fire are part of the ecosystem like some forest the branches are really high up for this reason. In australia eucalyptus trees their leaves are really flammable and burn quick which leaves the trees safe. Also when the fires get big they create their own wind/weather. These guys are doing an amazing job not discounting the work they have done.
side note:We are causing huge fires by putting out small ones that naturally happen and burn up leaf litter. with out thoses fires clearing it out then big fires happen and can get hot enough and high enough to burn many trees that survive small fires
WHAT A TIME TO BE ALIVE!!!!!!!!!!!!
~ 8k subscribers away from ONE MILLION subs !!!
What a time to be alive ;)
Cant wait for these papers to be used in procedural fire in games. Seeing as we dont need to simulate all the variables listed, and we are like almost 20 years since far cry 2, this should be commonplace.
Forest firefighting simulation game from this would be so cool.
Game devs please take note and deliver us a game based on this paper
Very good. So, the more sensors we can put on the ground we would be able to just from machine learning algorithms, operating on Google Earth maps, be able to feed this system in real time, during wild fires, and thus increasingly accurate predict the spreading of any fire, even under a heavy smoke screen, as well as in darkness, and continously feedback the system with real time data from overviewing drones using infra red cameras with exakt known position?
Makes me want to make a realistic fire fighting coop game
This is such important work. I feel a lot safer knowing that simulations like this are being worked on.
We call those clouds pyrocumulus clouds over in Australia. often the rainfall from them is evaporated and boiled before reaching the ground which causes a large downdraft of fresh air for the fire. Very scary to see
Looks like Portland
We need some kind of commercial product to use this kind of stuff, to prevent and fight wild fires
So does it use the same method per tree or not? If not, what are the trade-offs when compared to the single three one?
4:15 Did all 120,000 trees have the same particle-based breakable physics like the tree in 0:30?
If so that's incredible in real time.
"95 what?"
yeah yeah, 95 seconds, it's really amazi-
"milliseconds"
MILLI WHAT 👀👀👀 *papers flying everywhere*
a paper with rather profound ramifications (please pardon the pun). apart from modelling the increasing likelihood of forest fires, it looks like this is getting closer to being usable for movie forest fire simulation - something that is otherwise either impossible, stupid, or very dangerous to achieve.
So, we can finally have a spiritual successor to Far Cry 2!
How about the grass though? Is it not a large contributor for spreading the fire where there aren't so many trees?
jornal n pode rir ta meio estranho
What is the software used to make these simulations?
amazing but have they taken into account sparks embers from the fires carried on the wind, that are able to ignite further fuel sources and also take into account wether it is a wet or dry season? even without these this is an amazing simulation!
Ok, I was not expecting you to say "milliseconds" when looking at the computation time, that's incredible.
Hey everyone! I just found something brilliant regarding interpreting networks, and I am so excited that I want to tell someone! :D
Over at Steve Bruntons channel there is a new video called "Interpretable Deep Learning for New Physics Discovery" and it is basically about converting neural nets into actual readable math formulas! It is really well explained, even for people that aren't good in math or programming. While I didn't understand everything, I was still able to follow.
And as icing on the cake he shows examples of the network "re-discovering" known physics and then even applies it to uncharted territory like dark matter.
I love it! (As if you couldn't tell)
Is this the beginning of the machines teaching us?
First they said neural nets couldn't be creative ... and we got stuff like original music, style transfer for everything, contextual content generation and faces of humans that never existed.
They said (and I thought): "Fine, but AI will never do the real scientific work!"
But oh boy, this guys work makes me question that believe.
Or am I getting this all wrong?
Thanks!
0:41, I just realized what was weird here. I think it's missing air resistance which would prevent a tree from rebounding that fast.
1 to 120,000 thats awesome crazy
the next paper down the line is probably going to simulate how the world would end because of an asteroid
This is so crazy