Blender Developer Notes: June 14, 2015
Ton shares the notes of the developers meeting in irc.freenode.net, #blendercoders.
Hi all,
Here are the notes from today’s meeting in irc.freenode.net #blendercoders.
1) Blender 2.75 Release Candidate
- So-far no big showstoppers, but it’s possible we do an RC2 this week. Tuesday we decide. But even without RC2 we’d need a week of more tests and fixing.
- Everyone keep testing! Get your old .blends and try it with RC.
- The release logs are still in progress. Everyone who added something in past release should make sure the feature has a log here.
- Splash has color space error (10% oversaturated). Official build will be corrected for it.
2) 2.76
- Updated the planning with a 2.75 BCon proposal.
- We’ll try to get a SIGGRAPH release ready (sunday 9 august)
- Bastien Montagne has File Browser enhancement project ready.
- Gooseberry targets? Ton Roosendaal will sit down the three Blender Institute devs to discuss it.
- 2.76 could become the “Industry Standard” release, adding these features at once: PTex (Disney), OpenSubdiv (Pixar), OpenVDB (Dreamworks), Alembic (ILM). But OK – probably too much for one release!
3) Other projects
- Martijn Berger proposes to also release Cycles standalone on blender.org. Needs a bit more details filled in still.
- Viewport project: new mailing list gets attention and feedback, not decisions yet.
Thanks,
-Ton-
You guys ought to see the ad that I’ve got on my screen for this page. For some of you, it’ll make you start itching.
~
*WARNING: Anyone who experiences trypophobia, or the phenomena of itching at the sight of tiny holes, look away now. Hahaha!*
~
http://oi58.tinypic.com/2ln9t09.jpg
I had that ad too
That “Industry Standard” release sounds absolutely awesome! Add to that a Standalone Cycles renderer, the possible pipeline opportunities are mind blowing.
Getting all of thous standards into one build might be a lot or a long ask but I think it would raise some heads/eye brows. I don’t know if that’s important in the grand scheme, but its really exciting to see blender moving in this direction.
We’ll see. It takes more than just adding PTEX and Alembic and such, to become an industry standard. It’s interesting, to say the least, but I find myself holding my breath a bit here, only for the fact that there’s still too much Blender has to address before making such claim.
It also takes other factors, such as incorporation with other software (that is, playing well with other software), providing more familiarity along with other standard software (such as offering complete “Maya”-style controls for industry veterans, while still offering the option of traditional Blender controls for Blender veterans), and easy user flexibility (that is, fully-customizable UI).
Not to mention, we’re talking about PTEX, but Blender doesn’t even handle 4K textures well. It often stutters in trying to load 4K textures, and often crashes when trying to paint 4K textures. What’s the point of next-generation texturing if it’s all limited to 2K textures? Though, this isn’t to undermine the PTEX addition–this is just to point out the nature of prioritized development. Putting PTEX in Blender before better higher-resolution stability is much like putting the cart before the horse.
Blender also needs to tighten up integration a lot better, if it dares claims to be ready for “industry” status. For instance, as soon as you’re done modeling, you should go into sculpting with ease, and then to retopo (which really should be an independent Mode in Blender by now–a Retopo Mode) for easier UV and retopo, and then easily into Texture Mode, then animation, and all the animation already sitting ready for you within the Compositor for ready editing.
MODO does *precisely* this, and their developers take a lot of time and pride in their work to ensure that working in MODO is as seamless a pipeline as they can make it. It’s not perfect (they always continually improve and innovate), but it’s an extremely well-integrated and unified product.
Blender *has* this capacity to be such, and it’s a truly awesome potential, but we often see so much segmented development that there’s rarely enough unified focus and vision enough to do this. We just keep adding new parts into Blender, hoping it all fits somewhere along the overcrowded UI, continuing in ad-hoc style at rapid pace.
And because Blender seeks to be the one-stop suite for all aspects of 3D creation, its development needs that kind of carefully-though, oneness-of-mind, careful development to truly rise to that potential, even more than MODO. Because Blender seeks to be the lone hope of the FOSS world, it has the most pressure to rise to the occasion of not letting its FOSS nature get in the way of it feeling like a unified professional product.
That’s why it simply is not enough to just add PTEX and Alembic–which, by the way, others have had available elsewhere for several years now. While Blender’s adding PTEX and Alembic, other software are moving towards the next-generation of pipelines, including Physically-Based Rendering pipelines, seamless integration with software like Substance Designer, and emerging new innovations like geodesic voxel binding (which is *the* next generation of character skinning, and we won’t remember life before it).
We’re a new age now, where streamline pipelines are essential and software must now adapt to the users, instead of the other way around. What’s going to truly make the difference in adoption with Blender is how well it keeps itself integrated, both internally and with other software.
Otherwise, people are never going to take Blender seriously, which would be a shame, because Blender is a very noble effort. I’d hate for all this good effort to go by the wayside, just because Blender development isn’t as well organized and prioritized as it should be. What Blender needs is for the next Blender Foundation project to focus squarely on solid integration.
(And as a disclaimer, I give this criticism not out of cynicism, but out of genuine hope for Blender. I speak hoping the needs and challenges get noticed, and dealt with all in stride. I’d find it most interesting to see Blender rise to that challenge of truly becoming an “industry standard” player. It’d prove much about the potential of FOSS–that it is possible to meet the demanding needs of professionals, out of noble pursuit of keeping free cost.)
No software is perfect and never will be – that is why pipeline integration should be the highest priority – so other software can fill missing features. Blender has a lot of cool things to offer, but if it can’t play in the team then it will forever be left out.
True, no software is perfect (of course, perfection is always only in accordance to a purpose, even if one purpose at a time, but I digress). It doesn’t have to try to be the all-in-all software, to perfection. You’re right that there are other dedicated software to fill in for missing (or unnecessary) features in Blender.
Though, part of playing in the team with pipeline integration is ready flexibility for the user (read: no-fuss UI customization), better getting along with mainstream software (like completely-seamless transition into ZBrush and working with Allegorithmic for some sort of basic Substance support) and unifying all the main parts within Blender with better self-integration (i.e., transitioning between its various Modes with readiness and relative ease ).
And not just a “crack open the source code yourself” flexibility, but a “Welcome, non-technical artist–we’ve been expecting you–here are some ready options” flexibility. Blender can still benefit from its openness, but what the industry values far more is readiness than openness.
In many ways, Blender still does things the long way, in places where a little better focus in development would do wonders for the purpose of integrating Blender as a dependable player in industry pipeline.
Yes, but all this makes sense only when blender is ready to be in the pipeline.. otherwise it even does not make sense to start learning and using Blender if you know from the beginning that you will be left out and will never use blender in the industry.
Well, I’m only bringing all of this up solely on the basis that if and only if Blender’s now trying to shoot for “industry-ready” status.
Again, I’m being very careful here to hinge my argument only upon that condition here. IF we’re talking that subject, this is some of what Blender development needs better focus on.
And IF that “industry standard” status is among Blender’s goals, then it’s best to start thinking about them now, rather than later. Future design is at stake. Because it’s too late to work on problems tomorrow that should’ve been addressed yesterday.
Fairly certain he was meaning it would be a release dedicated to adding all of those Industry Standard formats, not that Blender would become the industry standard.
Yeah, I got that sense from the quotation marks. It’s a bit tongue-in-cheek. Though, it’s still a consideration brought up, of making Blender more accessible and capable for wider use. The topic’s still brought up, regardless.
Though, admittedly, it’s one brought up more so by use Blender users than the actual developers themselves. Blender users are often the ones entertaining the notion of how much the industry should incorporate Blender, more than the core developers themselves.
WIth that said, it’s also been no secret that Blender Foundation are trying to prove Blender’s merits as a film-industry-capable piece of software. Projects like Tears of Steel and Project Gooseberry try to show what direction of production Blender’s capable of achieving, something as a sampler of its potential.
Blender crashed on 4k textures?
I painted 6K without any problems. Strange.
I’ve got an 8 GB RAM computer with an i7 processor, and when trying to even LOAD a 4K texture sometimes, let alone try to paint at 4K resolution, Blender stutters on me more than Porky the Pig.
I tried painting in 4k in Blender. Man it is the worst painting experience ever on current hardware. Blender tries to update the whole texture even if I only painted 1 pixel. Then I try other software. On an old 4GB ram machine I can paint upto 16k without slow down. The I tried real industry standard software an 8GB ram machine with i5 cpu, I can paint 5 16k texture without noticeable slow down.
Blender cannot paint even 1 4k texture without slowing to a crawl.
Other software can paint 5 16k textures at 1 time.
Alembic + OpenVDB would change everything! I believe it would be turning point when suddenly a lot more studios would consider using blender. I don’t understand why this has not been the highest priority task so far?!
“Priority” is the right word there. You hit on to the heart of the problem. It’s nice to see Alembic and such making it to Blender, but you’re right that these really should’ve been much higher priority, and thus incorporated much sooner–that is, if a pursuit “industry standard” has been some initial goal.
I would say they haven’t been the highest priority, merely because the pursuit of joining the ranks of “industry standard” hadn’t really been much a priority in itself. “For the community” has always taken the highest objective with Blender, which isn’t really well-suited to professional need.
Even the hot-topic issue of Blender’s UI controversy got summed up largely as “The community likes it, so we’ll keep it, largely just the way it is. If you come to use Blender, just learn it.” This hermetic attitude got Blender nowhere with respect among non-Blender professionals, who simply need software to revolve around them, and not they around the software.
The potential with Blender is definitely there. It just needs the right priorities there to match. Fewer “just for the community” priorities, and more “for the professionals” priorities–again, if “industry standard” is now Blender’s aim in direction.
I’m not sure what’s wrong with the interface? After using any other application by Microsoft, Adobe, or Google, it is always a relief to get away from their clunky methods and get back into Blender, which becomes an extension of my thought process. In general, Blender’s infinitely customisable interface design is years ahead of those big three.
Interface is definitely not an issue. Sure it can be improved and will, but this is not so important – every software has its quirks and each feels different.. some people prefer one while others are used to something else. Just compare Maya, C4D, Max, Modo – they are completely different in a lot of ways, but each is being loved by their users.
Interface is not what holds back from integrating software into pipeline!
Well, interface is certainly a big factor as to why Blender doesn’t see wider industry adoption. Sure, every software has its quirks, but most have a way around helping to alleviate having quirks.
For example, MODO and ZBrush are very unconventional in design. But they compensate for that by either offering the most industry-leading cutting-edge features enough to justify just putting up with the quirks (ZBrush) or by allowing for absolute user-friendly customization (ZBrush, MODO).
Again, what use is it to put Alembic in Blender if the other users hate using Blender? Part of pipeline is how well you can use the software–software preference plays a part of pipeline, just as much as file formats and new features.
By the way, it’s very hard to improve an overcrowded UI. Blender’s adding more stuff than more room to place it all.
I don’t think that users hate using Blender – it’s quite the opposite.. We are here (at least I am) because we like Blender.. I have been using other software for 10+ years and started using blender only couple years ago and stuck with it because I just love the workflows and how it is designed in some ways (although there are things that bother me – like useless outliner and other things). Really – I have also “converted” several long term Maya users and they just love Blender and can’t look at Maya anymore.
So when talking about industry standards, i think it’s not so much about attracting new users in first place, but to allow existing users to have much greater benefit from using blender – to use it along people using other software, get jobs in industry etc.
I think you’re perhaps a minority, who sees others among the minority as a growing majority.
People accustomed to Blender generally don’t hate using Blender. But we’re not most people. There are more people (namely, independent individuals) picking up Blender, but most other people (namely, industry professionals) still by and large avoid it.
The number of people who try Blender and actually like it is probably a lot lower than those who try it and prefer it over more traditional 3D programs.
The most common complaint from other people outside Blender is how much they hate how everything’s organized, with its unorthodox use of horizontally-scrolling context menus, to its inefficient use of layout design (as you said, the Outliner is pretty inefficient).
Of course, we experienced users to are adjusted to Blender’s unorthodox and unique design learn to adapt to it, and with time, even prefer it in some ways over other more rigid menu-based programs.
My point here is, though, that Blender would do well to allow for greater user flexibility, with drag-and-drop ease. Even as a long-time Blender user, I wish I could do this in Blender sometimes.
Merely dividing up the viewport isn’t enough sometimes–sometimes I need a menu to dock to the side of the screen, much like the Toolbar, or to tear something off the Toolbar and place it directly into the Viewport as a floating menu.
~ “I’m not sure what’s wrong with the interface?” ~
It’s not so much of what’s *wrong* with Blender’s interface, as much as it’s becoming a problem of why it can no longer continue to ignore the nature of its design limitation: The UI’s getting overcrowded fast.
Though, I do have to remind people that the Blender UI originally wasn’t purposed as a long-term solution. If folks may remember, back in Blender 2.5x, this UI was billed as a work-in-progress.
And it’s always been a WIP. But we’ve stuck with it for so long now, that we’ve become acclimated to it, despite its shortcomings and limitations. I wouldn’t say it’s a broken UI, or something *wrong*. It’s just one needing some serious thought to a growing issue. We need dedicated look at the inherent problem.
Again, if “industry” is what we’re shooting for here, that is.
~ “After using any other application by Microsoft, Adobe, or Google, it is always a relief to get away from their clunky methods and get back into Blender, which becomes an extension of my thought process.” ~
Do you see the key issue there with that statement? We’re talking about YOUR thought process, as a long-time Blender user. Blender is fine…for Blender users. Blender should remain fine for Blender users.
Though, what Blender Foundation really needs is a dedicated UI team–professionals in user interaction–to retain the desire of the community, while addressing the needs of the industry.
Greater UI flexibility would be a nice start. Too much design of Blender’s usability come from people not ideally suited for the task. A great programming isn’t always the best designer.
~ “In general, Blender’s infinitely customisable interface design is years ahead of those big three.” ~
Well, sort of. What I mean by “customizable” is being able to literally rip any menu from anywhere and to place it anywhere you need. Drag-and-drop ease, with the ability to lock the custom UI.
Technically, that’s not what Blender allows. It allows for window set-up, more than actual customization. If Blender’s going to operate in its own specialized way, the way to alleviate this issue is to take a page out of ZBrush and MODO’s playbook,of just letting the users arrange the workflow however they want.
That’s how those other two help account for their otherwise unconventional designs–they allow absolute customizable (and I mean absolute–everything’s customizable).
Of course, the Blender community doesn’t really have such need, and honestly, probably wouldn’t even care for customizing Blender beyond recognition. Thus why customization and refined UI would be an option, and not a mandatory change from Blender’s default. (Options are nice.)
But again, if the topic’s about “industry standard,” we have to start thinking in terms of needs beyond our lone hermetic desire. Otherwise, we’re fooling ourselves with this goal of making “industry” status.
I’m not sure if the goal for Blender is to be “industry standard”, but I believe that would be most beneficial thing for all blender loving users even if they don’t know it :)
I mean – for all people who have committed to learning blender – this would open a lot of doors and possibilities in the industry even for people who initially see this as just a hobby. This could be a life changing thing for many.
At least I feel really frustrated when I have spent a lot of time learning Blender, creating addons etc., but then I realize that actually I can’t use it professionally and I have to re-learn other software. It’s just heartbreaking!
I feel this would be just fair for the community – to all the people who have spent years trying to make blender better – to give everybody chance to do something more with Blender.
Please!!! Please!!!!! OpenSubdiv, please!!!!! T.T
.. also UDIM (multi tile UVs) support would be more beneficial than PTex in my opinion.
I would love UDIM as an additional option. You’re right that it has quite a number of benefits (esp. in terms of production pipeline) that PTex doesn’t has.
UDIM is more adopted by the (film) industry than PTex–which really hasn’t seen much use outside Disney’s own studio use, as folks figure out how to best utilize the technology.
Though, both options offer unique benefit to texturing, both have their advantages, and both are totally welcomed. Glad to see Blender making some strides into embracing newer standards.
fully agree about UDIM comment!
what does standalone cycles renderer mean and cant u just make a more faster renderer
The Cycles Renderer is Licensed under a different license, one that is compatible with Blender’s license, but is also compatible with Commercial projects. This is enabling them to split the Cycles Renderer out of Blender to function separately for other software to use as a target, essential aiming at being an Open Source competitor to products such as Maxwell, Keyshot, Render Man and others.
It will still be included with Blender, but available in this other format as well. I know of several projects to build pipelines for other applications. I believe someone is building one for Rhino. I hope someone is building one for Daz Studio. One could be built for Maya, 3DS Max or really anything that can export the geometry.
Alembic support is a MUST HAVE if blender will ever have a hope of getting into existing studio pipelines.
Sounds like an exciting release. Keep up the good work.
OpenVDB – does that mean that its the first step to get better fire, smoke, water sims that won’t require as much memory and will be faster than what nowadays Blender have?
OpenSubDiv and OpenVDB are most awaited now.
“Alembic … was initially developed in 2010 by teams from Sony Pictures Imageworks and Industrial Light & Magic”
Sony Imageworks is responsible for it as well. Also they have more OSS technologies that maybe Blender could use as well.
@ Brian Lockett,
Great opinions, man! The more time I spend in blender the more I get the exact same thoughts and I really hope that for the next year or so Blender Institute will act as a conventional software development organization, because Blender desperately needs prioritization, integration and rapid improvements of the many half-baked solutions left behind from the open movie projects… All I want is a good usable software and not another short film… Hope Ton and the head developers do something about this and the next big community effort to be towards making blender better software. The great movies and proofs that this project is something special will follow naturally.
“All I want is a good usable software and not another short film…”
TELL IT ON THE MOUNTAIN! :D
And thanks for your feedback. Yeah, I’m someone who’s always trying to give Blender a chance as stable enough to be my lone software for my game development, but I’ve had to turn to other software, simply because I keep running into the same intrinsic issues, time and time again.
It’s so frustrating, too, because Blender’s got real potential. I use other software for my UE4 pipeline, but I’m still fond of Blender and would like to see it reach a level of professional reliability. It just needs to finish what it starts, before starting something else. It’s exactly as you’ve stated:
“Blender desperately needs prioritization, integration and rapid improvements of the many half-baked solutions left behind from the open movie projects…”
If they stabilize, integrate and polish their existing features, they’ll have much easier time adding newer additions. Because some amazing stuff is heading our way, and Blender will be behind in embracing it, caught in its own way.
Getting this output building Blender Trunk against Clang/LLVM 3.6.1 on Debian Sid/Experimental.
Given: Libboost 1.57. C++11 On.
This breaks the build:
Scanning dependencies of target buildinfoobj
[ 86%] Building C object source/creator/CMakeFiles/buildinfoobj.dir/buildinfo.c.o
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/Expressions/KX_PythonCallBack.cpp:89:38: error: use of undeclared identifier ‘typeof’;
did you mean ‘typeid’?
PyObject **argTuples = (PyObject **)BLI_array_alloca(argTuples, maxargcount – minargcount + 1);
^
/home/mdriftmeyer/Projects/Blender/blender/source/blender/blenlib/BLI_alloca.h:38:3: note: expanded from macro ‘BLI_array_alloca’
(typeof(arr))alloca(sizeof(*arr) * (realsize))
^
1 error generated.
source/gameengine/Expressions/CMakeFiles/ge_logic_expressions.dir/build.make:491: recipe for target ‘source/gameengine/Expressions/CMakeFiles/ge_logic_expressions.dir/KX_PythonCallBack.cpp.o’ failed
make[2]: *** [source/gameengine/Expressions/CMakeFiles/ge_logic_expressions.dir/KX_PythonCallBack.cpp.o] Error 1
CMakeFiles/Makefile2:6225: recipe for target ‘source/gameengine/Expressions/CMakeFiles/ge_logic_expressions.dir/all’ failed
make[1]: *** [source/gameengine/Expressions/CMakeFiles/ge_logic_expressions.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs….
[ 88%] Built target bf_rna
[ 97%] Built target bf_freestyle
[ 97%] Built target buildinfoobj
Makefile:146: recipe for target ‘all’ failed
make: *** [all] Error 2
mdriftmeyer@horus:~/Projects/Blender/cmake-blender$
New warnings with latest FFmpeg:
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:143:10: warning: ‘avcodec_alloc_frame’ is deprecated
[-Wdeprecated-declarations]
frame = avcodec_alloc_frame();
^
/usr/include/x86_64-linux-gnu/libavcodec/avcodec.h:3635:10: note: ‘avcodec_alloc_frame’ has been explicitly marked deprecated here
AVFrame *avcodec_alloc_frame(void);
^
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:239:12: warning: ‘avcodec_alloc_frame’ is deprecated
[-Wdeprecated-declarations]
m_frame = avcodec_alloc_frame();
^
/usr/include/x86_64-linux-gnu/libavcodec/avcodec.h:3635:10: note: ‘avcodec_alloc_frame’ has been explicitly marked deprecated here
AVFrame *avcodec_alloc_frame(void);
^
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:240:24: warning: ‘avcodec_alloc_frame’ is deprecated
[-Wdeprecated-declarations]
m_frameDeinterlaced = avcodec_alloc_frame();
^
/usr/include/x86_64-linux-gnu/libavcodec/avcodec.h:3635:10: note: ‘avcodec_alloc_frame’ has been explicitly marked deprecated here
AVFrame *avcodec_alloc_frame(void);
^
[ 94%] /home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:395:12: warning: ‘avpicture_deinterlace’ is deprecated
[-Wdeprecated-declarations]
if (avpicture_deinterlace(
^
/usr/include/x86_64-linux-gnu/libavcodec/avcodec.h:4849:5: note: ‘avpicture_deinterlace’ has been explicitly marked deprecated here
int avpicture_deinterlace(AVPicture *dst, const AVPicture *src,
^
Generating tr.mo
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:1065:10: warning: ‘avpicture_deinterlace’ is deprecated
[-Wdeprecated-declarations]
if (avpicture_deinterlace(
^
/usr/include/x86_64-linux-gnu/libavcodec/avcodec.h:4849:5: note: ‘avpicture_deinterlace’ has been explicitly marked deprecated here
int avpicture_deinterlace(AVPicture *dst, const AVPicture *src,
^
[ 95%] Generating ar_raw.mo
/home/mdriftmeyer/Projects/Blender/blender/source/gameengine/VideoTexture/VideoFFmpeg.cpp:53:12: warning: unused variable ‘timeScale’
[-Wunused-const-variable]
const long timeScale = 1000;
^
Future Suggestion: Text Object Layer Tool to leverage GEGL/BABL and FreeType rich features to not have to use GIMP for text transitions and much more. Create a non-destructive editing with independent layer resolution once GEGL is fully ported and GIMP has released 2.10 and on towards 3.0.
The bit depth features from GEGL would be huge, not to mention so much more with OpenCL.