Physics Sprint Results
Posted in Blender Institute, Development by campbell
Just before the blender conference, the physics developers got together at the at the blender institute to discuss how different physics systems can play better together and share more code.
- Daniel Genrich – Cloth
- Jens Ole Wund – Softbodies
- Nils Tuery – Fluid
- Janne Karhu – Particles
Brecht van Lommel, Ton Roosendaal and myself (Campbell Barton) needed to see what was common between the different physics solvers to make into generic functions. We also had to make a short term plan for the peach project.
By the end of the 2 days we agree’d on the underling API as well as new features needed to improve blenders work flow.
A big incentive for this is the need for hair/fur from Janne’s excellent particles branch (not yet included in blender).
Highlights…
- Fast Playback; Automatic caching for physics solvers, transparent to the user with no need for the Bake button.
- Editable Bakes; When the physics are generated, you can bake the results to a sequence of editable keyframes, similar to shapekeys, applied in the modifier stack to allow non destructive hand editing.
- Unified collision detection; All physics systems can make use of this. At the moment each system does its own.
- Janne’s Particles in Blender; Hair you can edit by hand, comb and cut, to name just a few of the features.
For details see the sprint meeting outcome on the Development Wiki
October 15th, 2007 at 9:55 pm
Nice to see the physics taking shape. Thanks guys for all your work.
October 15th, 2007 at 11:08 pm
>> Daniel, Janne, Nils, Jens, To, Brecht and Campbell
Is ‘To’ written without the ‘n’ because he’s only partly on the photo as well? :P
October 16th, 2007 at 12:17 am
He was busy and had to keep running out for stuff too!, thanks for the correction
October 16th, 2007 at 2:17 am
Great news!
October 16th, 2007 at 4:13 am
I’m So anticipating the Hair particle system. The demonstration video was amazing.
October 16th, 2007 at 7:00 am
that is very clever to minimize redundancy in the code.
BTW did you guys thought about creating some sort of an engine to be able to do the redundant thing all over the programmer I am not much of a programmer but I heard brad Peebler talks about a similar idea.
and LoMac could you please provide me with a link for those demonstrations
thank you
October 16th, 2007 at 4:12 pm
Thanks for the work you guys put into Blender!
Now, all we need is a built-in mocap system which works with webcams such as the one presented at Siggraph;-) ….just kidding, but would be fantastic ofcourse:-D
October 16th, 2007 at 5:32 pm
Here you are. http://www.blendernation.com/2007/01/22/combing-brushing-and-cutting-hair-is-in-blenders-future/
October 17th, 2007 at 1:23 pm
Of course, it wouldn’t be completely useless to be able to apply both soft and rigid body dynamics to armatures…
October 17th, 2007 at 1:53 pm
Just a question:
Will there also be some optimalisation to speedup the rendering with the internal renderer?
I’ve done a compare between my desktop (3Ghz Intel WinXp) and a server with 2 Xeon 2.80Ghz) – 4 threads and the difference is not that big… 2 minutes 51 seconds for a certain scene.
October 17th, 2007 at 4:20 pm
I was just wondering this morning what happened during the sprint. Glad to see this news and the developer wiki info! Can’t wait to see this implemented! This will be a great advancement for blender! Keep up the good work!
October 18th, 2007 at 12:16 am
@Wim, this is planned, Brecht’s tests are showing more then 2x speedups in some areas! – though this is in development.
Physics api is such a big area that even to do a small part of it will mean a lot of work, so rendering etc isn’t included in these plans, but will probably happen anyway ;)
October 18th, 2007 at 8:23 am
@campbell, thanks for the info that really sounds great! When rendering a still image the effect of a speedup is good, but it makes a big difference for animations, because of the multiplication in number of frames.
Keep up the excellent work!!!
October 19th, 2007 at 5:02 pm
Great job! Can’t wait to see all the implementions in 2.50!
November 25th, 2007 at 8:11 am
Fantastic news on the physics stuff, guys! The features sound very exciting…
I have a quick question: I’m trying to get a “fluidy” looking fire/flame working in a project I’m doing. I’ve tried the fluid system and the particle system but cannot get the “right feel” for the flames. I’m wondering: How would you guys tackle a realistic, fluid-like flame/fire? (And I dont mean a single,simple candle flame – I’ve seen the tute for that…)