No less than 15 Blender projects were accepted into this year's Google Summer of Code!
CGEffex writes:
Once again, Blender has been accepted into Google's summer of code program, and has received 15 project slots! With such features as Cycles volume rendering, deformation motion blur, and some new shader nodes, as well as improvements in the VSE, motion tracker, rigid body simulator, paint tools, dependency graph, and the BGE (woot woot, game devs!), this is sure to be another awesome summer!
http://www.google-melange.com/gsoc/org/google/gsoc2013/blender
36 Comments
Hmm…
…a "woot woot" for BGE game devs?
I'm still waiting for the improvements of GSOC 2012 and before.
Things like:
→ FBX importer
→ Android port
→ Hive System
→ OpenGL upgrade
GOSC 2011:
→ Inferred Rendering
GOSC 2010:
→ BGE Shader Work
Not in some custom unstable not maintained development branch. But in the stable mainline one.
I would like to note that this year Jason Wilkins will be working with our very own Brecht van Lommel on a 'Viewport FX II' project, that is directed at updating blender's outdated OpenGL system :)
Yeah, I know the feeling.
I mean, don't get me wrong, folks--I always get excited to see what GSoC brings for Blender, but it seems like most of it stays in unstable status. I'd love to see more stabililization of past work before we move on to newer projects.
But with that said, I'm ecstatic to see what's in store for GSoC 2013, most particularly the Cycles volume rendering and paint tools! Hopefully, these paint tools in particular will finally help make Blender the top-choice 3D texturing tool it can be.
I agree, I was especially waiting for the Android port, and was rooting for it to be continued in the GSoC 2013, but sadly that won't be happening. And I don't think it will be done outside this, it's not an easy job. :(
It just stopped, and in a state that it can't be just compiled from latest source...
It's as I find myself always saying, Blender's greatest strength is its greatest weakness: It tries to master so many things that it never truly masters anything.
We can still have the newer features, but Blender would gain much more credibility as a tool professionally if it would sit down long enough to complete one feature at a time. We'd have fuller completions done faster that way.
I'd rather use a complete feature while waiting for another one to be complete, than have a bunch of incomplete features, risking some of the ones started to be lost to the spotlight of the new announced features...which will most likely stay incomplete.
15 is awesome. GIMP got 4 by comparison.
my favorite projects:
-Viewport FX II
-Painting tool improvements
-Texturing for Volume Rendering in Cycles
-Cycles - Add new shader nodes and shading features
+1, these are my favourites, too :)
I hope Viewport FX II means a better realtime (non render preview) with support for cycles materials. Currently the realtime preview blender Internal has is much more complete (e.g. real time Bumpmap or Normalmap display.)
All sounds very exciting, though what particularly gets my top interest among them is the paint tool improvements. There's a sore need for a top-quality 3D texture-painting tool, and I'm hoping Blender truly becomes that answer.
A close-second in excitement to me is the Sketch Mesh Editing. That sounds like a wonderful mesh-editing tool. I always love seeing mesh-editing tools up for development!
Anyways, thanks to all, I look forward to seeing all the new work soon, and good luck this year, guys!
* "...a free a top-quality 3D texture-painting tool..."
Wait... what about Gimp? Does that not work?
I believe he's talking about actually having GIMP/Photoshop level texture editing capabilities in a 3D environment within Blender.
I don't think that's viable or even useful to duplicate what other FOSS tools can already do. Let's stick to pure 3D modelling / rendering improviements.
@BigPilot that wouldnt be a Duplicate of what other FOSS software can already do.
Look at mari for example, one of the leading commercial Texturepainting tools. You would never be able to work that way with GIMP+Blender simply because they are two separate Tools. Even if you have Photoshop at hand, which has some features superior to GIMP you wont't be able to paint textures in there with ease.
I think/hope the targeted improvements here are not limited to additions in painting tool functionality itself, but also some major workflow enhancements.
For example, the current Texture painting system confuses a lot of users, because you first need to create a blank image, save it, unwrap the mesh in a random way and finally start painting.
I could think of a way to automatically get this done by the software as soon as you enter texture painting as a major workflow boost.
Yes, what blazer003 and randomguest have said are both what to which I was referring.
When working in a professional environment, or just wanting to otherwise streamline your workflow, texturing directly on the models in 3D is not only time-saving and easier, but it offers you the tightest workflow possible for texturing.
There's no saving texture-exporting-painting in Photoshop/GIMP-saving-exporting-checking your progress-paint some more-saving-exporting.
This over time can make texturing not only time-consuming, but a headache to deal with when you start working with more than one texture map. Due to its frequent need in 3D work, texturing should be far more streamlined a process. It's a more technical and indirect a process than it really needs to be now these days, in the face of better solutions.
Dedicated texturing products like Mari and the painting features in Mudbox (and even ZBrush's PolyPaint and Spotlight, to reasonable extent) are becoming the standard of the entire industry for great reason. It's far more intuitive and fluid for the artist's workflow. We simply don't have time to try to do the bulk of our texturing in 2D when texturing in 3D gives us so many advantages.
Now, that's not to say that texture-editing in 2D applications won't be needed--that's simply not true. We'll still need 2D image-editing of texture maps to quickly adjust color together, apply filters on top of the textures, and perhaps some will still find it easier to use 2D for doing some very fine details.
But still, there are many things that 3D texture painting offers that 2D texture-painting simply can't, such as texturing with radial symmetry, height-based painting (so that you can exclude painting over cavities or protrusions), tiling with stenciling, and painting across seams in a consistent manner--just to name a few. All immediately and naturally.
Currently, there are no free solutions for such options--heck, there aren't even many commercial applications doing this, despite the growing demand for more of such options. But thankfully, Blender has become closer to that full potential. It needs cleaner workflow as well as more features--I mean, it's been a decade, and Blender still doesn't even have a simple bucket fill tool.
My hope is that they go all the way with is, such in a way that other 3D packages like Cinema 4D has done with BodyPaint 3D and modo has been long doing with their 3D texture-painting system. Heck, even 3ds Max has a decent 3D painting feature--it lets you use layers and Photoshop brushes.
Compositing features and renderer improvements are very nice and all, and I do look forward to them improve in Blender, but truth be told, there are a plenty of other available options for doing such--many of which being better to work in than in Blender anyways. No one's waiting for the compositor and Cycles to be finished when After Effects and V-Ray are already complete, far more robust and prevalently industry-standard. In other words, anyone serious about productions needing industry-level features will use industry standards anyways.
If Blender had better 3D texturing tools, it would immediately gain something that many other applications simply do much. We could also take advantage of improving Blender's sculpting tools, seeing how Blender has something of an advantage that even Mudbox doesn't have--dynamic topology. We could use a true remesh tool, as well as simple features like layers and global operations like Inflate and Polish.
These features would give Blender something that many other competitors simply don't offer. That'd help drive more support for Blender professionally. With more support for Blender, then we could begin to address those other priorities--possibly even from some industry aid.
i´m excited about the VSE!! YAY!
Aye - all of the projects look interesting but the VSE update and the viewport update are the ones that really stand out to me as area's where Blender most needs some love (now that so many of the biggies have been taken care of!).
That said, I also agree regarding the need for feature stabilisation. Blender nominally ticks the boxes on a lot of features but when you start to get beyond the beginner stage a lot of them feel either incomplete or unpolished - fluid, smoke and sculpt tools for example.
Great Great Great!!!!!
I am also most excited about the VSE as this one really needs an update!
To me, I think the VSE and the compositor should more work 'together':
So that basicly the compositor IS the node editor part for the VSE. Then, while using the VSE you can do the simple "drag and drop stuff", whlie in the compositor part you can do the more complex node set ups.
( So it's basicly the same as setting up materials for in cycles. You CAN do everything in the properties panel, but using the node editor makes it a lot easier. )
Also keep in mind that a lot of people are using After Effects for compositing and that AE is a .......... Video Editing program!
AfterEffects is a (layer based) compositing tool, but definitely not purely a Video Editing program!
Blenders node-based compositor is more like Nuke.
Node compositing vs. Layer compositing can be summed up in a sentence: the easiest things are harder while the more complex tasks get easier to overview with a node based system.
So if you complain about people switching to Ae for post editing its just because they prefer a layer based workflow for minor tweaks, that might be a little slower to setup within blender with nodes for them.
Blenders VSE is obviously layer based, but more of a video editor, so its not intended for advanced compositing (that's what the node editor is for) Making these two work "more together" as you proposed would require a mix of Nodes and layers, which has never been done before (afaik). I don't know if that would actually work out, but if they get a succesful mix of Nodes/Layers blender, as a compositor would have the best of both worlds (AfterFX vs. Nuke /// Layers vs. Nodes; plus a full 3D feature set.)
That would really make blender a KILLA comp tool (with the most flexible Workflow)
Also taking into account the limited 3D functionality of both nuke and Ae [with C4D "lite" in newer versions) blender could be superior to both major compositing worlds!! [that would be great :D]
Everything is cool! :)
Deps. graph refactoring... It's very serious and very important. I believe that Blender will become very powerful pretty soon after this big stage will be completed.
P.S.
15 GSoC projects!!! Congratulations! :)
No fluid simulation improvements still ? , unfinished fluid particles ?.
So upsetting.
I hear you brother. I've been waiting for a compleate SPH solution in blender for a long time but it seems like the wait has only just begun.
Seems like with Blender, development builds wider with new features, but sometimes forgets to build taller with old features. Perhaps if we had a division with Blender Foundation dedicated just to the latter... Hey, if Blender would organize and crowd-fund an agenda like that, it'd get supported.
I think we should be able to vote on features we would like to see worked on with the option of donating to it.
While that doesn't sound like a bad idea much at all (and personally, I think it's a great idea), I think the only problem with it is that most people in the Blender community are non-professionals, and a democracy with development might just work against our favor.
As a result of a good majority of people within the Blender community just being folks enjoying dabbling with it as hobbyists for non-professional use, many, if not most, of them will vote for the flashiest new features proposed, hoping for more shortcuts to play around with--rather than those with more project-oriented experience who'd love to see new flashy features but want a focus more on finishing older features first.
But hey, if they did have what you'd suggested, I'd sure use it. And if people proved me wrong with such a system, hey, I wouldn't complain.
the countless discussions on blender artists about this really show there is interest in such a system, there is blenderstorm.org for example, which has been overrun by spambots. However, there are still some new proposals there but there is no active voting and no active Devs interesting in picking things up, even the much upvoted ideas haven't been implemented
Currently the (most official) way to register interest in some features with the community's legitimation would be just asking for feedback on BlenderArtists and mailing it to a developer if there is a lot of positive response in the community.
I've also seen troll people getting tiny new features in over the bugtracker, but thats not the way to do this ;D
Given the fact that this ( https://www.youtube.com/watch?v=rce92SZ1y60 ) is open source AND made by the guy who made the current fluid sim AND Ton said "We're on it" we can definitely expect something to happen on fluidsim, just not in this GSoC :)
Nice.
Not quite super additions for the BGE ,but hopefully there is going to be a good debugger.
Too bad plenty of BGE additions/patches are still not in though..
Inferred rendering is still in works..
Will be nice have the VSE with HLSL material, for fast FX compositing and custom real-time effects like LIGHTWORKS, for me the best video edinting tool.
Improving cycle will take a great deal of time and founding consumption.by then cycle will be an out date rendering technology. I have seen rendering test done on a new render engine Call NuN it very fast! two second with ray trace in any complex scenes develop by independent sources. However I'm sticking with blender...
E X C E L L E N T
The reason most of the features are not implemented is because the blender foundation barley has enough money to cover basic support. All of the fancier feature require time and resources, something the blender foundation doesn't have. So if you want to see them accelerate then I suggest you pony up a few extra ( insert your currency here ) towards the Blender foundation.
...Or the Blender Foundation could master the older features first, mastering Blender's features first before moving on to new ones. I wouldn't want to fund more waiting for simple improvements that honestly should've been there years ago, but got lost to the spotlight of flashier new features.
Check out Steven Shearing's suggestion above. I think something like that could better work--that is, if the community will be responsible enough to put stabilization of long-standing features as a priority before adoption of new flashy features. You'd get more support if people knew for sure that their money was going to solve a longstanding problem they'd been waiting to see fixed for years.
Absolutely. Especially now that there is a more flexible mesh structure as well as an improved API in place, there are so many simple extras that would be simple to code and make a world of difference.
Imagine if the Subdivide tool had the ability to adjust the number of divisions along U|V|W separately, as well as linked to the same value. Similar to the separate parameters used in creating a UV Sphere, as opposed to the single parameter for an Icosphere where only the one is needed.
Or more ambitiously, having "self" as an option when selecting objects for scripted expressions in drivers, or for other uses such as in constraints. This would make it easier to link an action to hundreds of other objects without each of them being stuck with an absolute reference to the original object, in cases where all of the new objects are intended to reference themselves.
Yes there are workarounds, but they are time consuming and prone to mistakes. There is so much that could be gained from just minor tweaks such as these, and with such a comparatively low coding effort for those who have a thorough knowledge of the specific tools to be refactored.
So many really, really nice projects! And YES, the dependency graph! This is going to be great!