Blender Presentation at Siggraph

Information of blender development and Mango news & timetable

Thanks for posting this, very informative.

Nice little autodesk jab at the end as well, I’m curious as to what was said along with it :slight_smile:

I know I little bit meticulously, but may be in the official paper Nurbs tools must spell like NURBS tools? (24 page)

Dependencies!! Yay! Animation updates throughout the work chain, would be cool. And with project Mango just around the corner, I can’t wait for the coding sprint with improvements to the poor old VSE.

Did anyone see this talk at Siggraph? I ask because seeing “Future of the game engine” listed under “Challenges” makes me nervous…

A bit off topic and early to talk about it but regarding Mango, 3 weeks greenscreen in the studio, does that mean we can expect some Live Monitoring solution in Blender from a camera feed, the OSS MLT app already includes Black Magics SDI / hdmi SDK combine that with Blenders OpenCL Compositor to be for realtime greenscreen keying (not for final comp), some scopes and view LUT ability via OpenColorIO that’d be neat.

Hope theres some on location shooting planned in there, can’t help thinking any real credibility from the VFX community is going to come from ability to intergrate VFX into real locations, rather than all greenscreen. More sort of Monsters rather than The Spirit type approach.

Yes, I was there this afternoon. Really interesting meeting, with lots of new info (at least for me). The dependency graph is going to be awesome. One of my questions was related to exactly that: what happened to the massive crowd simulation (mentioned at the beginning of Durian development)? Well, without the dependency graph it was not really possible to implement, according to Ton. Those sort of dynamics WILL be possible with dependecies - right now having 1000 cloned warriors would take up as much space as 1000 warriors; with dependecies this would be reduced to only 1.

To answer your question, Ton mentioned that the game engine integration as it stands now, is limited: you just cannot enter the game engine directly with dynamics, hair, cloth, and so on. First you need to convert these effects into something the game engine can handle. His opinion regarding this was a simple one: it is too complicated and messy, and requires extra steps. Ideally we should be able to just enter game mode without any conversions or modifications.

As for the future direction: he was somewhat vague, I feel. I think he and the other devs are still debating which path to walk. He mentioned the game engine is now 10 years old, and no matter what, requires an update. Another option he mentioned would be to create a game pipeline to another modern game/realtime engine (open source of course), and making sure the integration between that other tool and Blender is near-on perfect (he touched on this by commenting how painless the Unity<–>Blender connection is).

I gather we would still be able to develop in Blender (including Python) and seamlessly connect to that game engine, and have a working game as output. The game engine might not be integrated in Blender’s environment in that case.

Again, I am speculating in that last paragraph - not Ton’s words, just my ‘brainstorming’. The fate of Blender’s GE is still being discussed about; at least that is what I got from his words.

I agree, studio lighting on greenscreen can be soooo hard to get right (unless you are ILM). However building complete 3D environments has its appeal too. Imagine getting high speed Cycles reconstruction of environments to act against while your live key takes place. I guess we could use the live Mocap (a Blender GSoC branch) for camera position capture.

Effectively enlisting game engine for on-set production.

With so much todos concerning animation on the table I am less than confident (british understatement) about BMesh and NURBS fates; unless these two items are mandatory for Mango, I don’t see them coming next year. Imagine: being to be able to do Hollywood-level VFX while, at the same time, not being able to create a pentagon…

Yeh I did notice that one, I commented on it in another thread:

Interesting… There certainly are a lot of options to consider. BGE is very old and inflexable, yes, Id be excited to have it updated (which would of course be a lot of work) rather than all game related stuff being turned over to a different engine.
After the issues with Crystalspace during the Apricot Project, it seems like we’re almost back to square 1, at least with the issue of what to do next
(not to forget all the recent advancments with BGE of course)

I think Bmesh has quite a good chance to make it into trunk before Mango (it’s 4 month until the mango development sprint starts). There’s still issues left to be resolved, but the branch is quite stable. I hope Ton will push the development so the last things are wrapped up. It would be very unfortunately if bmesh would had to wait because of Mango. NURBS is a different story, since there haven’t been any active development lately I don’t see it coming anytime soon.

From my experience while tight integration with Blender and reusing features provides a lot of benefits, it can also provide a lot of problems. For example, most, if not all, of Blenders tools have the focus of accuracy over speed. The BGE is supposed to favor speed over accuracy! Problems with this are evident in things like the animation system which the BGE borrows. Blender’s animation system, while great, is a little slow for the game engines.

Also, to address the engine being ten years old: it has been receiving updates. While some parts may need more work than others, it’s not like the entire code base is ten years old. For example, the modern way to use OpenGL is to use GLSL shaders and VBOs for everything. The BGE currently makes use of GLSL shaders and has experimental support for VBOs in Cucumber.

I’m not sure if Ton consulted any BGE devs prior to making this part of the talk, so I’d take his comments on the BGE with a grain of salt.

Actually, at the meeting Sergey expressed interest in adopting the Nurbana branch in September, and Ton gave a green light. S also mentioned that Eman would probably have some time over to assist him. Rejoice! =) On another note, Howard Trickey mentioned that he was working on the todolist for Bmesh, and that people were welcome to help him out, but none replied :confused: Well, go Howard (commited a bugfix today)! =)

That’s great news! Sergey is definitely the right man for the job. I hope I’ll have time for NURBS along Motion tracking coding and regular BF work…

To me, when I hear of the what the plans for the BGE are, the most obvious outcome I could see is this.

When the new render API for Cycles is done, the BGE itself will eventually be decoupled from Blender and plugged back in using the API, so it will be like the Cycles engine in that it would be an addon, but supported by the Blender Foundation, I wonder then if the API will be built to allow custom space types so the Logic window will not be available when a render engine is chosen.

This would essentially put the BGE on even ground with the other GE options available to would be game designers, plugging into Blender as an addon via the new API, I can see this being the future of outputting content from Blender as they’re already moving in that direction with Cycles.

@AE that would be awesome, many of cycles materials will be impossible to do w/ OpenGL in real time.
so yeah why not have a Game Engine API also, …

oh and BTW Doom3 source is up for release anytime soon ;D that could be sick if you could hook such engines straight into blender. Guess there’s toooons and tons of loose ends to connect, bones animation, logics, textures … but would be sweet

doom3 goes open source

Btw, kinda funny how ‘bevel’ and ‘knife’ were added to the list of things to finish up. With the ending of the 2.5x rewrite and bmesh, I think we will see a surge of new functionality. Looking forward to that! (even Autodesk is feeling the heat, what with that Siggraph slogan that they’re using: “don’t blend in…”

Nice blender video from sig:

Eheh…really looking forward to Mango.

BTW: is the Siggraph reel available now or maybe later ?

i hope that the open movie will be finished for siggraph. i think it would be a great suprise for all the industry to see what blender will be able to do for live action effects movies