Blender2.67a is coming up later this week, with over 100 bug fixes.
Ton Roosendaal writes:
Hi all,
Here's a summary of today's meeting in irc.freenode.net #blendercoders:
1) Release 2.67a issues
- Saturday showstoppers were fixed! Tomorrow EU daytime we call for release builds.
- The list of 100+ bugs that made it to 2.67a so-far.
- Well, even when we're a bit sloppy, we do fix it :)
Lessons to be learnt:
- Better testing! People still don't use or try the Release Candidates enough.
- More strictly respect our "BCon" schedule. During BCon3 commits should be reviewed and only concern work on approved projects, bug fixes, or urgent maintenance.
- We could switch to a different cycle, with odd (unstable) and even (stable)?
- Regression .blend file collection could need update. Need someone to coordinate.
2) 2.68 targets
- Candidate list.
- Campbell Barton has some misc Mesh tool improvements too, not listed yet.
- Martin Felke: his proposal for Explode Modifier was reviewed by Sergej Reich, wont be ready for 2.68. Blender just has shortcomings that better be solved first, makes such modifier much better.
- Added to targets list: Cycles GPU hair render (in svn), new hair shaders.
30 Comments
Great job BF!
Can we please at some time take care of the Modifiers? They're pretty old tools by now -most of them- and some could be pretty awesome. Why can't I have an array of lights or group instances for example? New features are nice and all but improving the existing tools is also important imho.
Dude! Just Friday I was work for on a project at work and I was like damn why can't I have array lights? Something like that and other little modifier updates could come in real handy.
I agree. I deeply appreciate the new modifiers and Cycles improvements and stuff, but if we could sit down and improve the long-existing features, that would be awesome.
For instance, it would be nice if we could use the layers for sculpting. In this day and age, layers in sculpting and texturing are rather imperative, and sculpting without it feels very limited. You end up making compromises to your sculpt to compensate for the lack of layers.
The Texture Paint Mode is heading in a great direction with the stencil mode, but I wish the stencil could be moved and zoomed incrementally--sizing them manually can be a hassle. A bucket fill tool would be extremely time-saving, and I'm a bit surprised that's not a feature yet. The masking pen in Sculpt Mode could be brought over to Texture Paint Mode as well.
It'd also be nice to have some simple operations for the Sculpt Mode, such as a deformations like Inflate and Polish sliders so that you can systematically inflate and polish unmasked areas of your entire mesh instead of manually doing it with oversized brushes (which is what I use).
I think it's very cool to have new features, but it's just downright important to complete existing ones--esp. if we want Blender to be taken more seriously by professionals. In fact, most professionals don't rely very much on bells and whistles, not nearly as much as hobbyists do. For instance, most professional concept artists don't use fancy brushes very much in Photoshop--they stick to plain chalk brushes. They rely more on technique than tool, but they do rely on the tools they do use to be complete.
If I had to define Blender's main weakness, I'd say it's the very thing that makes Blender unique--it tries to fill so many shoes in the 3D world (and I understand why--there just isn't another FOSS answer to much of what Blender does, such as a compositor, 3D printing tools and camera-tracking tools), with adding more new features with each build, that it doesn't truly master any of its long-existing tools to a full completion.
Though, again, I am appreciative of the new features and the developments. I don't want to sound ungrateful or unappreciative here. I'd just personally rather have improvements to long-existing features first.
Couldn't agree more with you Brian.
Maybe they should have a dev team only to improve existing features.
Boy, that'd be nice.
I agree that some feature aren't really good enough to be usefull. I feel exactly the same way about hair sim, no proper collision for hair. It actually makes hair simulation very impractical to use.
Currently you do have "layers" with sculpting. Just use shape keys as layers. You have the added benefit of painting masks for the layers by painting poly groups. I don't think it will work with dyntopo though. Do you even retain your layers in Zbrush when you dynamesh? I am not too fussed about that.
Texture painting without layers is quite useless, to me anyway. Another issue with texture painting in blender is that when you paint very low opacities with a soft brush it creates banding. Texture painting is also kind of slow with large maps. A screen buffer paint option similar to mari would be very useful.
I am very happy with the general direction of blender's development. Especially Cycles. I tried the new sss and min hair thickness and the results are fantastic. All I still really need from cycles is deformation motion blur and motion blur working on hair.
*vertex groups not poly groups
The next best thing to having an array of lamps is to duplivert a lamp parented to an arbitrary mesh.
I can't wait to use the new version!
By the way, did they implement a true micropolygon displacement in blender internal/cycle?
Thanks.
It seems like a very good way to implement that would be to couple the Displace Modifier with dynamic topology, so that way, you can apply the great detail of polygons were you need it.
They should really add OpenSubdiv...
It was looked into at one stage but I don't recall a huge amount of enthusiasm on the mailing list. The main question was - what was the use case for it and what gap would it be filling?
Mind you, I'm not a dev, so don't rely on my memory!
Actually the issue is one of open source license. I'm pretty sure the devs know what the advantages are. Campbell covers it in the latests blender pod cast.
I still fail to see the issue with the licence thou, since it's still a Open Source License, just under MS Public, so it protects Pixar's own Patents associated with their texture handling, which isn't needed for the OpenSubdiv implantation in Blender.
Incompatible with GPL. That's the only reason.
Oh, how we should value these incredible Blender devs! Having just signed the petition against Adobe's CC (Constant Cash) plans*, it is so incredibly refreshing to jump back to Blender land and witness yet another demonstration of just how incredibly agile and responsive good software development can be.
*If you have to use Adobe products, you might want sign this petition: https://www.change.org/petitions/adobe-systems-incorporated-eliminate-the-mandatory-creative-cloud-subscription-model.
Adobe Photoshop traditionally cost about $1000 USD, several models of Adobe Creative Suites cost a range of about $1800 for select collections to $3000 USD for the complete Master Collection, and if you want to upgrade an existing Adobe product you own, it would cost you several hundred dollars. These were professionally-priced, aimed at professionals.
Most people (small fries, mainly--small business and freelancing professionals, as well as average consumers) complained for years that Adobe's professional products was too high. Most people don't use 95% of Photoshop, and similarly, most of the features in other Adobe professional products go largely used.
So even among those of them who bought the products, they were spending more money for what they didn't need or use. But most of these other non-industry people pirated their copies of Photoshop, to the point that most of Photoshop's user base use pirated copies, all using the professional prices of Photoshop as their justification. Something like 60-75% of Photoshop users use pirated copies.
Meanwhile, industry-level professionals got their start as industry-bound students buying drastically-reduced-priced copies of Photoshop, which Adobe offered quite generously to students. Most get high-paying jobs right out of college, and many even get jobs where the companies provide the software--you just need the skills.
Most industry-level professionals simply don't complain about price because they make the money to buy professionally-priced software (and hardware). They know that paying even $3000 for complete software pays for itself in about two months working professionally. Industry professionals don't complain.
Adobe now offers their entire Adobe line as a payment-based cloud-based subscription service, offering their single applications for only $20 a month, which includes all their upgrades. They offer their ENTIRE line of software (which traditionally would be Adobe Creative Suites Master Collection, previously about $3000) for $50 a month. You can pay for the months you want or save money and pay annually.
As this service is cloud-based, it is online-based and thus significantly more harder to pirate. Adobe thought they finally struck a balance between addressing the small fries and offsetting the outrageous piracy rates they experience. You get access to all their tools for small fees a month, and access the tools as you need them.
Those same non-industry-level consumers (small business and freelancing professionals, and the non-professional average consumers) are now complaining that this model is a rip-off, that it will cost them more money in the long run, and that they don't want to pay an "infinite amount" of money for services they otherwise would've bought once as standalone. They complain that they want to be able to pay for standalone copies again to buy once and own forever.
Never mind that MOST of them complaining simply DIDN'T do that all the years before.
Human nature.
Funny enough, if you're not an industry-level professional, you're actually SAVING money, not spending "infinite money." And if three years later, you're not making money with their products, it simply doesn't matter HOW the products are priced--you're not the professionals for which Adobe was aiming. You're not spending "infinite money" if you're making money using their small-fee business model.
Now, let me say that I myself don't particularly like online-only cloud-based services--generally, I don't like ANYTHING that forces me to stay tethered online. I also do prefer to pay once for something and own it as standalone. If I were to complain, it would be more about the technology, not the business model--which isn't the bulk of most other people's complaints on this petition. But I just have to say that I frankly don't blame Adobe for what they're doing.
Feeling talkative? ehehe..
It's just been one of those days. lol
It's not just the higher cost (after 18 months on CC I would be paying more than I would for a CS suite) but the operating terms of CC that is too harsh. If Adobe said simply that you won't get any more updates if you stop paying, fair enough. But the fact that the applications will actually *cease to work* is a step too far. If a user who chooses CC (instead of CS6) then decides to move away from Adobe products, they'll have a hard drive full of AI, INDD and PSD files that they can never open again.
For businesses the implications are even more frightening. By signing up to CC they are effectively signing a lifetime commitment for access to their own design files, without any idea of the future costs that Adobe will impose to 'allow' access.
If you use the software for your job, chances are, you'll be earning enough to pay for it. If not, there are open/free alternatives to Adobes most popular applications... A lot of people I know use cracked copies of the creative suite, but to be honest, they mostly just use Photoshop and more recently Premiere. With it being harder to get cracked copies because of CC, it'll either drive people who use this software for their livelihood to start paying, or it'll force them to try alternatives.
As far as being locked in to the proprietary file formats, pretty sure Gimp can open PSD files, not sure if InkScape can open AI (I'm guessing probably)...
I agree with Brian that most of the noise is probably coming from hobbyists who use cracked copies anyway...
Don't take it too hard on yourselves Blender Developers. It's true that most people don't test the release candidate, including myself. It's therefore normal that most of the bugs will be discovered with the point release.
Good job that you guys fixed a 100 bugs in such a short time.
What's up with GSoC? That *used* to be a really big thing, but I haven't heard much about it this time around...
I believe that's because it all just got started. It's not yet decided which projects will actually be mentored.
True!
I'm always excited about GSoC because every time it's like a box with foreign candies - you can only guess the taste by it's flavor!
BroadStu and Brecht, thank you for Cycles Hair! And now for GPU rendering!
It's a great gift for the all Blender community!
Just five months passed after Hair rendering development started (as I know) and now it looks very decent. Almost (I'm just not sure) the complete feature.
Congratulations! It's very fast as I think.
I don’t test RCs as such. Or even release builds, for that matter. I do my own builds from source, about once a week. I started adding one or two 2.67-specific updates to the “Blender 3D: Noob to Pro” wikibook a few days before 2.67 was officially released.
IMPRESSIVE
THANK YOU VERY MUCH FOR YOUR HARD WORK
I ( WE ) GREATLY APPRECIET IT
Thank you blender developers for all your hard work!!!!
There are so many improvements with each release.
yes, thank you for all your hard work!
and now to the thing with the bugs and the release philosophy...
in my opinion is there no need for "even/odd" releases, even i see any benefits with a RC. no of the customers, i make software for, are interested in RC's or an unstable even/odd release - they do ignore its existence, when they know it is not the final version.
RC's are maybe for the customers of interest,only to get a preview of features or if they need a preview for further testing, to test if it will be compatible with the developing environment of them - but it is not intended to give these pieces of software out to the end-consumer.
i think the quality management should be better.
strict guidelines, how code should be look like, how system resources must be used, ... , and also tiny stupid thinks like variable initializations or how they has to named, ... avoiding using strings/values in the code if it is used more than one time for the same meaning, use a string/value variable/constant/definition/enum instead to bypass problems in case of forgotten string-value-changes, so you have only one point to maintenance and don't need to crawl through everywhere where a value is maybe used, and so on...
there should be also a guideline for the workflow of how to change/fix something - so what it will be/should be triggered - coding, testing, documentation process...
ok, that is maybe a bit too overloaded or complicated for an open software project, where multiple people with different levels of programming skills and contribution intentions come together...