Amazing results on speed have been posted on the new smoke simulation by Daniel Genrich.
Blender Development – by Daniel Genrich
If you would like to support Daniel's effort there is a PayPal donation button on the upper right side of the website
Amazing results on speed have been posted on the new smoke simulation by Daniel Genrich.
Blender Development – by Daniel Genrich
If you would like to support Daniel's effort there is a PayPal donation button on the upper right side of the website
69 Comments
EXCELLENT !!!
THANK GOD!
It's very awsome!
ok any on tell me how he did that? not the smoke the explosion, did he use explode modifier which only turns mesh into scraps of 2d triangles for me or did he "glue" together those individual objects? and use the trajecytory script on each one?
This is amazing! I can hardly wait to see more. Thanks.
WOW!
Awsomness!
Wow!!! Next to the new architecture, additions like these will make Blender truly compete with its commercial alternatives. I'm really curious how Blender's future from 2010 and onward will look like! Keep 'em coming!
Finally. The blender is getting better and better. The smoke is really realistic. I love it.
Wow! Blender is getting better and better...
No freakin way... nuh uh, that can't be my beloved Blender, can it? [drooling all over the place]
what if Roswell alien technology DNA was added to bender code development.Wow! ---------------------Nah..
Pps. smoke only work on better graphic card instead of cheap graphic card.
Great work guys. Looks like it might give some SideFX coders something to compare against!!
Looks awesome.
i just keep replaying the video...playing over and over again.... speechless....
Wowwowowowow that's nice.
I know what you mean, we've been looking at how to scale up houdini's simulation resolutions over our farm and blender just 'does' it!!! Wow. :D
just testing the script
ok thanks guys, for not answering my question.
nice. even very nice! but it is demanding? how long it takes to render this?
Can't wait 'till 1st Blender 2.5 final release
@random
you can make such an object explosion (not for the smoke) with a mix of explode modifier + particle modifier with some option. This is probably what he did. Lookup the Internet for a tutorial on that.
WOW
Fantastic work. A milestone for open source rendering.
That is a beautiful thing to see in Blender, thank you!
myselfhimself THANKS
even though i got to do a little research that's the best answer i've received. appreciated :)
great stuff
omigosh.
WOW, so is this for sure going to be implemented in 2.5 or is this just one of those crazy custom builds put up on the SVN? Please do something like this for fire!!!!
I agree with Acolyte. Please do the same for fire. Ben Dansie showed us it's possible :)
ooh smoke simulation :-D this is going to make 2.5 absolutely amazing
how can you bake the smoke simulation to play it in real time like that????
The blender smoke 4 (performance) video is pretty crazy!
OH WOW...
@Lich
This is realtime simulation, not baked. You just need the right hardware.
WOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO!
This was one thing stopping me from using Blender in production.
--
Kevin
Wow,
There are so many features coming I forgot about this one ><
so unreal :D
Greatness added upon excellency. Now if they would just make a simple script that would allow one to define units better... to make import/export to other apps more accurate/easy/efficient.
And an october release for first version of the new system is quicker than I anticipated. Thanks Team!
Very awesome!
Now my only question is "How long did it take for the to calculate the simulation for that smoke?" If it is anywhere as slow/memory intensive as water it will require an impressive studio computer to make any use out of it.
@wispyred
What kind of hardware is required?? navidia Quadro??
I have an nvidia 9800gt- Q9300 OC 2.7- RAM 8gb 1066- I reach real time but much les resolution.
------
other thing, those are particles? I mean the debris. . . because dont cross the floor (no se como decir atraviesan o rtaspasan) or are rigid bodies?
SMOKING!!!!!!!!!!!!!!!
O_O
@random-
To me it looks like he's done a physics sim using separate objects, using the game engine and an initial velocity to calculate them. An explode modifier wouldn't be able to get the bits rolling around on the floor right, and it would also tend to make the inside edges look a bit funny. Of course, he could also have done it entirely manually if he had enough time on his hands- which I certainly don't- and going by how you expect the internet to jump at your command and answer all your questions instantaneously neither do you. Seriously, we're too busy drooling at this new functionality to notice or even care about some random who can't even take the time to use basics like spelling or grammar...
Marbar, crusader for the English language
Experiment Garden,
He is using billboards to project the smoke and the simulation is done in realtime. It is simply amazing stuff! :)
It's looks nice but what I would love to know is :
- how long did it take to bake, then render
- if it really is realtime and depend on the hardware, how restrictive is the hardware requirement ?
- is it something that can be unified with other volumetric projects from Blender ? (because that would be awesome)
- is it available for testing and where can I find a test .blend ?!!!!!! :D (omg I peed myself gotta clean up)
Well done!
Blender is improving greatly :D
@aws357:
1. There's currently no caching/baking for the smoke simulator, but I'm guessing for the ~125 frames in this simulation he averaged somewhere around or slightly below 1 fps (I can't say for sure though). What the artist did was capture every frame in the viewport and then compose it into a realtime video. And this is not rendered, it is a viewport preview that allows you to rotate around the volume like any other model.
2. It is not realtime, but most people are getting decent framerates depending on their hardware and the complexity of the simulations. You'll need some pretty nice hardware if you want fast results (more than 2 gb RAM, greater than 2 GHz processor), but the developer, Daniel Genrich, recently posted a 400% speed improvement.
3. Yes. The sim_physics branch with volumetric rendering hasn't been merged with 2.5 yet but will be down the road. Otherwise there would be no point to this :)
4. Yep. Get the latest 2.5 build at http://www.graphicall.org, and you can find a sample file at http://www.wxtools.com/wp/
Today my screen was scratched, sob sob...but this puts awsomeness back into my day!!! thanks!
You hear realtime and you think, hmm... a supercomputer could run Lux in real time... what hardware is this going to cost me?.. but shock on me!
excellent stuff!
__________________________________________________
paypal is as illeagal in my country as what your granny takes on a sunday afternoon!
AWESOME!!! :O
aws357: graphicall.org. Its in the latest svn builds.
Very much cool :)
Very interesting stuff. One thing that Blender has an advantage over Houdini is the simulation feedback speed. Dont know the accuracy for this smoke and how well it works with other dynamic effects but many times speed is the key and exact realism comes second. Alos really interesting to see that everything has gone so well when looking back one year. It seems it was planned all along to be apart of Durian. Cool!
http://www.blendernation.com/realistic-smoke-coming-to-blender/
Will this feature be available during playback in game engine as well ?
Okay, this has dunnit...I'm getting a blender-tattoo ;-)
Marbar, thanks =/, i think.
Oh man that looks cool!
WOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO!!!!!!!
O.0 that was so AWEsome. Never seen anthing like it. Daniel is a very great programmer. I like to grow up to be like him.
Nice! Very happy to see this! It looks awesome!!!
Wow.. Seeing the video is one thing, but getting the current svn build and seeing the example file is completely awesome !
Hardware requirements seem to be surprisingy reasonable: Animation preview with calculated smoke movement is still fast enough for a preview, on low smoke resolutions, and graphics hardware isn't an issue: a gf8800 is able to draws the smoke in the viewport in realtime.
hmmm I noticed after a while, the speed comes down and blender gets bogged down even with no animation going... bug it seems. animation starts of at about 10~15fps then after a few animation runs you cant even get 1fps. I know it's beta so I'm not suprised just mentioning.
____
Vista32 ATI:3200HD build:22326
Wow, that was amazing. I can't wait for it to be released! I think 2.5 could finally push Blender into the mainstream, and a direct competitor with the big names. Keep those features coming!
@hitechboy722:
I've noticed it aswell. however this seams normal. Every fluid/gas simulation gets slower after it gets more complex. Realflow has the exact same thing :) If u have 200 particles generated over time it gets slower.
I hope we can cache it all some time soon :D
WOW, that's so cool. Can't wait for its release. This gives all those big boys a run for their money.
Nice But...
As a casual observer and not an expert in Blender not 3d animation, I do see an an issue with the way the smoke is generated. The smoke appears to be generated or at least tied to three shell fragments after the initial explosion. For me, this looks as though the fragments are the source of the explosion and not just the results of one. Wouldn't it be better for the smoke to originate and act independently of the shell fragments? In real life, the smoke is from the burning of the explosive material and not from the encapsulating material, unless that material also caught on fire.
Just my observations.
@John:
Theoretically you could tie the smoke to invisible empties. I think that perhaps the reason why it is tied to the fragment is that the fragment pulls the smoke along after it because moving things create a low pressure vortex behind them.
But you are right.
It doesn't look quite realistic. I'm sure that better results can be obtained though with a little more tweaking.
Sounds to me that Blender needs an "Atmosphere" object that is basically unbound in which physical properties can be varied throughout. An example is varrying air pressure in different quadrants to create air currents. This would allow collisions to act on objects like making trees move instead of moving the tree to make it look like there is wind.
Also, in an explosion, pressure behind an object is greator that pressure in front of it so it's not really possible to to "draw" smoke with it on it's path.
That is so cool. I can't wait for 2.5
whooosa - amazing! :D
Good God. That was fantastic!
At the beginning of the project, tasks always took place after school hours for which teachers received no pay. ,