Blender 3.2 | Hacker News Reader

Blender 3.2

blender.org

423 pointsvictornomad89 comments

I know the potential audience is most likely a lot smaller than Blender but we're really struggling to grow the volunteer community around Open Brush (the open source fork of Google's Tilt Brush).

I was expecting it to grow organically but it's actually gone quiet recently - despite continuingly healthly download and usage numbers.

If anyone has any suggestions that don't involve me spending all my time on community building or PR then I'd love to hear them.

https://openbrush.app/

https://github.com/icosa-gallery/open-brush/

That's unfortunate, because this is such a killer app for VR-based creation.

I expect advanced 3D workflows of the future to look less like Blender and Unreal Engine and more like Tilt Brush.

Full degree of motion of both hands is so liberating. Plus it's fully immersive. When VR UIs improve around knolling and contextual tooling, this will become more obvious to people.

I agree to a large extent. Although not all interactions have been cracked for VR and there are things I'd still prefer to do in pancake mode with a keyboard and mouse.

But anything spatial - most definitely. On the whole trying to arrange things fluidly in 3D space via a 2D monitor is like typing with gloves on.

is Multibrush related to your project?

They are closed source and don't participate in the community any more. We're working on our own multiplayer functionality that will hopefully be less buggy and more flexible.

very happy about that, would much rather have an open source implementation to recommend to people!

I suspect it's because the Metaverse was a flash in the pan.

https://pageviews.wmcloud.org/?project=en.wikipedia.org&star...

AR/VR is a headache. VR goggles are proprietary and locked-down, laggy, and make you dizzy. Your eyes try to adjust FoV when there is no adjustment needed.

How is that relevant to anything?

This is an art program for (mainly) individual users and predates the metaverse hype by multiple years.

I dislike the metaverse hype as much as you but I love content creation tools. There's no need to use any mention of VR to rag on that.

I also mentioned our usage numbers are very good so your point doesn't even really make sense.

I modified my comment; indeed the hype was somewhat unrelated, but I think the reason it didn't catch on was the poor experience of current headsets.

Lack of users isn't the problem. VR growth is fairly healthy and niche applications especially so.

I don't want to get sidetracked into a general "defending VR" thing because that got still about 5 years ago.

I was hoping to get people thoughts on how to build a dev community around an already popular content creation app.

> laggy, and make you dizzy.

This is largely a function of both frame rate and IPD.

The latter is one of the reasons why I feel Meta has done an absolutely huge disservice to VR adoption by making the Oculus series fixed IPD. They claim to be able to software compensate, but my experience is that IPD on a headset being even just a millimeter or two physically off makes the difference between a comfortable VR experience and one that leaves me with a headache after the fact.

The fact that they reduced screen refresh rates on the Oculus S and the original Quest didn't help at all either.

I've had an Index since shortly after they became available and can use it hours at a time without any sort of discomfort provided my GPU is able to serve up enough frame rate for a particular title. Available GPU power is holding back VR currently more than anything, IMO.

Antecdotal:

Oculus Go and Rift S were the only fixed IPD headsets they produced, and I was able to use the Go comfortably with my abnormally wide IPD (69-70mm.) Quest 2 has 3 digital settings for IPD, and that also works fine for me at the widest setting.

(And both of the other consumer-release Oculus headsets (Rift CV1, Quest 1) had analog variable IPD.)

Didn't realize they'd added it back to the Quest series. It was a mistake in the first place on the Rift S, good that they reversed course.

I honestly haven't paid super close attention to their hardware since the Rift S was such a disappointment and the Quest had the lowered screen refresh rates.

Rift S felt like an in-between step to partially satisfy PCVR users, and it was dropped disappointingly quickly. It's obvious that Facebook is moving towards standalone headsets now.

Quest 2 is designed to hit a very low price point, and it certainly shows. But despite that, it's a very competent high-resolution, high-framerate (90, with experimental 120Hz support) VR headset; the only major drawbacks are low FOV, poor color quality on the display, and lack of uncompressed PCVR.

You might just be too early right now. VR headsets just aren't very popular. Personally, I find the hardware atrocious and am waiting for someone like Apple to "do it right". As a late adopter of VR, I may not be your target audience, so take what I say with a grain of salt.

I was surprised you didn't have any videos of interacting with OpenBrush on your home page. Normally, I would have clicked away if I stumbled on that page, but since you posted on HN I searched it up on YouTube.

Do you and your volunteers use Open Brush regularly? If so, toss up some casual live streams on Twitch or other platforms. Post on Twitter and Discord, etc. before you go live, then upload the recording to YouTube for people to watch later.

Part of what makes Blender so damn accessible is the huge number of YouTube tutorials. Virtually every feature of Blender has at least one high quality tutorial video, walking people through every step.

You make some interesting points.

Personally I was always assuming that any potential contributor was already fully aware of Tilt Brush. I find it hard to imagine that someone would be far enough removed from our scene that they would need to be informed about what Tilt Brush is about, but engaged enough to want to contribute. Am I wrong in this? Tilt Brush has pretty decent mindshare for anyone interested in VR content creation.

Another problem is that I just don't really want to spend my time making videos. Plenty of other people make videos about Open Brush / Tilt Brush. It just doesn't turn into "increased engagement from potential volunteers".

I really just want to code new features. Everything else is a distraction that I do out of neccesity.

Seconded on all of this.

I'm aware of tilt brush and would use it but I've still yet to have prioritized setting aside the money to even buy a VR headset, I don't much care for the companies doing it right now, and I'm not interested in just having a phone do it either. I keep hoping for the tech to improve and branch out beyond what it is right now.

> C# 84.5%

Sadly, .NET/Mono apps not so good for Linux.

We ran fine on Linux until recently. We still do under Valve's emulation layer but there's no working native OpenXR runtime I seem to recall.

C# isn't the problem. Unity takes care of that.

What is?

From memeory - it's the lack of a working Linux plugin for Unity's XR plugin framework. We used to use SteamVR on Linux but that's not currently working with the new framework and we need to move to that to support a wide range of hardware.

I'm sure it will get fixed - if not by Valve then by Monado who are doing great stuff for open source XR support.

https://monado.dev/

(some of the above might be inaccurate. I haven't had much involvement with the Linux side of things)

Did you have a look at https://monado.dev/ ?

Consider detaching your GitHub repo as a fork of tilt-brush.

Whenever I come across a fork on GH, my first assumption is that the fork is aiming to be merged back into the main repo in the future, and my second assumption is that the maintainers of the fork have less of an interest in the project than the original project’s maintainers. It’s a signal of lower quality IMO. You should keep mention of the original project in your docs, but I personally don’t think it’s necessary to keep the repo as a fork.

You can use the GitHub virtual assistant to request this: https://support.github.com/request/fork

I am pretty sure that issues, PRs, stars, and everything should be preserved, but don’t take my word for it.

That's a really interesting idea. There would be costs - We'd potentially be less discoverable for people forking the original. I'd have another place to check for interest potential forks (and therefore contributors).

I'll give it some thought.

It's probably a good idea because some GitHub browser extensions will automatically hide forks from search results, under the assumption that basically all "forked" projects are drive-by patchjobs. I'm not sure this is stopping anyone who would be useful to OpenBrush, but maybe it is.

Another good reason to defork is that, as we saw with the youtube-dl fiasco, there are some non-software-related reasons that MS may feel inclined to offline the entire fork tree. That'd be the bigger concern for me.

Got a Quest2 last July, right around time that Google was killing their Icosa-type-thing.

OpenBrush was imo one of the best apps for it and I wanted to start playing around with the code, but had some friction even getting a build started and eventually lost interest. Installed several SideQuest builds, of course, but after I couldn't get a working distributable with built-in Icosa even all the way into September, I got annoyed and I haven't started the app or looked at the project since.

Just getting the Oculus dev env set up to the half-assed extent I did, it's risky, dubious, and feels bad. I never know which agreement I might accidentally click that'll allow Zuckerberg to Quest2 into my house backwards, or whatever.

I guess this is probably a lot less helpful than I thought it would be when I started typing. Sorry!

Thanks. This comment prompted a discussion on how to improve the initial developer experience. The readme definitely needs a tidy up. We tend to assume that anyone interested in building it will be a moderately experienced Unity developer - but I think it's worth reexamining this assumption.

(It's a brave soul who tackles a codebase of this complexity as their first Unity project! But it was my first big Unity codebase and it was a great way to get myself out of the training pool)

It never ceases to amaze me how much improvements Blender is getting all the time. I've never used it much, simply because I got burned in the past by the (legacy) esoteric UI, but it has increasingly become a joy to get started with and just noodle around with.

There are other cool open source tools of course. Blender could've easily remained another one of those hard to use niche applications, but they've definitely managed to transcend that.

Blender was one of those things that I got into when I was in my early teens. This was around version 2.3X.

It has always had a great community, and made amazing development progress.

Version 2.5 was a big ui shift at the time.

These days with 2.8 then 3, they have managed to make some amazing progress, in a much faster turn around time.

I commend Ton's direction and the amazing work by the devs :)

The EU and the Dutch government does not get enough credit for sponsoring Blender until very recently when corporate sponsorships took off.

In the US the commercial competitors would have sued the government for anti-competitive practices if they had sponsored Blender.

> In the US the commercial competitors would have sued the government for anti-competitive practices if they had sponsored Blender.

Can you give an example of where this happened?

Heard of Google? The latest, and biggest one, is probably when 3K+ companies sued the Trump administration over the Chinese tariffs.[0]

[0] https://www.reuters.com/article/usa-china-tariffs-idUSKCN26H...

Maybe they are talking about SpeedTax etc?

This sadly happened in Germany, actually. We have a state sponsored weather forecast service, who also released a free Android (and possibly IOS) app. They were successfully sued by a for-profit commercial weather app (that even uses the publicly available data from the state sponsored service).

Now the state sponsored app is required by law to not be free. Last I checked it costs about two Euros.

I have trouble keeping up with the updates lately. To be honest, compared to quite a few CAD programs common in mechanical engineering, Blender doesn't have to hide here. Quite the contrary indeed.

Honestly hats off to the Blender dev team for taking their technically impressive but extremely tricky to use software and managing to over just a few years turn it into the one piece of 3D software everyone is suggesting to beginners and the fact it's free isn't even what you mention because the software is just great.

Really showing the great user experience on open source projects is possible and that the benefits are worth it.

Well damn, I didn't know that it had gotten easier to use but I am so glad to here it! I will definitely be giving it another go.

Oh I had no idea it was improved now. I tried it about 5 years ago and couldn't make heads or tails of it. I'll try it again.

it's improved by quite a bit, I wouldnt call it "intuitive" to an existing 3d user, it just goes against so much pre-existing muscle memory and assumptions from other apps... but so does Z-brush and people love that,

I'm using it more and more these days for little tasks, processing meshes, plug ins here and there, it has a beautiful and easy to use renderer but I dont work in a field where rendering matters.

I dont really want an alternative way to sculpt texture or model, I have several already, and really my industry (film / previz) is moving to rendering in Unreal or Unity (much more unreal) but as a tool Blender is quickly becoming much more impressive.

having tought myself zbrush before it was useful (2.something?) blender feels like it's just gaining that usability now to expose useful features... but there's so much more that can be done.

As a career developer 40 years into specializing on 3D graphics and animation, Blender's progress and accumulative feature set is stunning. I realized about 6 months ago Blender had cross a threshold of usability, and in those years of being "difficult to learn and use" the Blender community quietly built out a feature set that rivals, if not exceeds every commercial 3D modeler, animation, node based procedural subsystem, compositing, and scientific visualization product on the market. Due to Blender's deep Python integration, deep learning libraries can be imported and integrated to significant mad computer wizardry. Blender has come of age, and it is going to eat several commercial software publisher's lunch, if they don't wake up.

I think I read the focus this year is also live action tools / production tools, so lots of things coming that fit a professional environment.

I would highly recommend trying it again -- it's improved drastically in the past 5 years!

I've been using it since it first went open source, and while it was (and still is to some degree) a beast to learn, it was pretty much always the best 3D software feature for feature compared against other (quite expensive) software I was using / learning at that time. Over the intervening years it's improved in every possible way (except that it's still extremely complex and powerful, but no getting around that). The thing that helped me learn it then, and still helps now, is that there's a plethora of excellent tutorial videos and websites available (as well as these days many great books and courses), all of which can be easily found with your favorite search engine. For some really great resources, check out Blender Guru, CG Cookie, Blender 4 Noobs, and other such channels (found easily on YouTube and other video hosting websites, and on their own websites as well).

Something about the project is very interesting indeed. It grows at a regular pace for non trivial ideas. And it seems all joyful and light.

Blender is amazing. The other day, I needed to create a video of a rotating image so I looked around for potential software to use. Very quick and dirty job.

I didn't want to download anything large like Davinci Resolve. I saw OpenShot but was not sure it can do what I needed easily. It's not even lightweight.

Then someone mentioned that Blender can do video editing. Can you believe that blender is only around 200mb? I downloaded it, followed a quick tutorial on YouTube to figure out keyframes and how to render. 10 minutes after and I'm done.

It's great to see even more features to the video sequencer in this update. Will probably use again for my next video editing needs.

I love Blender but NLE functionality still isn't there yet. Basic things like keeping audio in sync with video breaks all the time. I had to switch to Resolve for my little projects. I will continue to evaluate Blender, but for now it still isn't ready for primetime.

Agreed about the NLE. I tried using blender (iirc 2.9) to make a family vacation compilation and it was a lot of work and not very intuitive. I barely do any video editing though so I may just not have had a good workflow. I wasn't able to find very good workflows either, I'm assuming because blender was still awkward for this use case and others usually reach for another tool.

I'm very much looking forward to further improvements and judging from the breakneck pace of features and how much love the project gets, I'm very optimistic about the future!

I had some issues with audio syncing to variable FPS video. Which is mostly used to save some space and battery when taking videos on consumer cameras, but probably something that profesional movie makers do not use at all. Still it would be nice for home video makers if that thing just worked out of the box.

This problem is present in Premiere Pro too. The recommended solution is to convert the variable FPS video to constant FPS before using it as a source.

For a free, lightweight video editing with a UI similar to Premiere, I recommend Olive 1.0:

https://www.olivevideoeditor.org/

It's small, opens fast, and does a good job.

I use Blender on macOS because it's the simplest video editor. Most importantly it supports webm.

To be honest, the video sequencer needs heavy refactoring. Yes, it can be used, but the UI isn't the best. Text support is very basic, so for titling it's not very useful (text is a general issue with blender). Also unfortunately, one cannot use shaders directly in the video sequencer to create their own filters for instance. Same limitations with the compositor (which isn't GPU accelerated so it is really slow).

But hey, it's free.

They are remaking the compositor now to be GPU accelerated. I saw some videos and looks very impresive!!

There really needs to be a "the architecture of open source applications" type writeup about Blender.

However it is that it is architectured internally seems to have helped it grow over time and not collapse under the weight of 30 years of hacks and poor decisions.

Completely agree, I wonder if it could be proposed and crowd-funded, or perhaps supported by a grant?

Link to AOSA for others' convenience: http://aosabook.org/en/index.html

The lack of corporate pressures to just get features out the door and clean them up later (never) I assume is a huge factor. I agree, Blender releases new capabilities at such a rapid clip that there's certainly a solid base on which they're building that we could all learn a few lessons from.

> However it is that it is architectured internally…

Countless hacks hanging off one giant MVC.

And having a few core devs with veto powers has helped a bunch with it never getting bogged down with bad decisions. I tried to get some iffy stuff tacked on but they (mostly Campbell) would be “umm, what’s this good for?” And don’t even think about adding a null check to prevent a segfault without getting permission first, like, signed in blood…in triplicate.

One of their very early decisions was extremely good in my opinion. Each blender file is saved with a complete schema at the start which basically describes a bunch of C structs, and has a marker for byte order. This means that .blend files can be backwards and forwards compatible, and in the most common case of the schema matching your memory layout, can have their data structures copied directly into memory and then only pointer patching is required. It's quite remarkable.

Yeah, agree. Very very handy if you are writing a parser!

the blender development wiki is already really good plus I think there's already a book on working in the blender code base 'core blender development' I think

I spent a lot of time during COVID learning Houdini as a hobbyist, because I really liked the concept of procedural modeling and node based development. However, with the continued iterations on Geometry nodes, it feels like a foregone conclusion that Blender will replace Houdini for hobbyist procedural artists. Cycles is a really nice GPU renderer and I don’t have to pay a subscription like I do with Redshift.

Houdini still has some strong advantages built over decades such as dynamics and KineFX and is the industry leader for FX. But I wouldn’t be surprised if these tools appear in Blender in a future version, though.

IMO geometry nodes are a bit lacking atm. No loops, no procedural uv unwrapping, no compact maths expressions, bare bones standard node selection in general etc. It feels like they've been concentrating on particle system type use cases at the expense of things like procedural architecture. However, its still a relatively new feature & I'm excited to see where it will be in a couple of years time.

The great thing about blender 2.8+ is that it tends to be good enough in a lot of different areas even if it isn't best in class in any one area. If it can get to that state with geometry nodes that would be massive.

I was reading this article yesterday: 'The best 3D modelling software in 2022'

As a blender studio subscriber & hobbyist, I'm so grateful such incredible software has no barrier to entry, especially for creatives in developing countries.

You can see this reflected in this great video of the worldwide blender community from last years blender conference https://youtu.be/uEjmbsiflMU?list=PLa1F2ddGya_8Wzpajwu1EtiS8...

  Autodesk Maya $4590 per year
  Zbrush $895
  Houdini $4495 per year
  Cinema 4D $3495
  3ds Max $1700 per year
  Modo $1596
  Lightwave $995

https://www.creativebloq.com/features/best-3d-modelling-soft...

To be fair you can get 3DS Max or Maya for $250 a year with a hobby licenses. You can get Houdini for free or super low price as well.

Zbrush and Houdini are still in their own category, untouchable though. Blender is getting close to the sheer pleasure that is sculpting in Zbrush, but there's still quite a ways to go. And Houdini's particle work is unequaled.

For Houdini, you listed the price for the version used by major VFX studios.

Houdini Indie (with no feature limitations) is only $269 per year, and you can also learn Houdini for free too with Houdini Apprentice (again: no feature limitations).

Independent VFX artists earn well under the $100K/year revenue ceiling for Houdini Indie, so $269/year is their actual "cost."

AMD HIP support on Linux now works without proprietary drivers on RDNA2 cards! My 6700xt is great as all I had to do was just install hip-runtime-amd package on Debian from AMD's ROCm repo¹. I am glad they are putting the work into supporting AMD hardware and I no longer have to use opencl and be stuck on older Blender versions.

¹ https://repo.radeon.com/rocm/apt/debian ubuntu main

dreaming for the day blender has more support for 2d animation. software like toon boom harmony are way too expensive for hobbyist work and there aren't many alternatives for that paper cutout type of animation.

I love blender.

I've been using it to model the renovation for my house, and I have no prior skills or experience in 3d modelling or CAD.

it makes me want to buy a 3D printer so i can make things for real.

I bought a 3D printer then had to learn 3D modelling so got into Blender.

Blender certainly works. Though keep in mind that mechanical parts - gears, bearings etc - are often easier to model in other CAD though plugins are changing this for Blender.

The artistic side of Blender however allows things that other CAD can't dream of. Make an object, turn it liquid, turn on gravity, run a few frames of simulation to get a melted effect.

There is a release party stream that you can follow here

https://youtu.be/0yGoeMfTY8Y

Seeing this announced made me realize it has been a few months since I did anything with blender. I'm not doing much gamedev right now (and even when I am I started using all the assets I bought cheap in various humble bundles/etc for prototyping) but this is giving me the itch to build SOMETHING just to do it.

I am far from a visual artist, but Blender is so satisfying to play with.

> Gas simulations now support motion blur

What? Isn't motion blur always done in post anyway? As far as I know, motion blur is just a property of the camera, it's caused by after images that show up in the picture when things move too fast, it has nothing to do with the actual 3d world out there.

You need information about the motion in order to simulate motion blur. It makes perfect sense for this to be generated by the 3D renderer rather than trying to guess at the motion after the fact when all you have are 2D frames.

Does 3D motion blur look better? The standard of realistic motion blur is probably real cameras, no? Real cameras don't need to guess the motion it's just an afterimage in a particular frame while the shutter is open.

I don't have any knowledge of this specific implementation, but my guess would be that it is a question of optimization. If you were to implement something similar to what's physically happening in a camera with a long exposure, you'd have to do a large number of oversamples (e.g. generate 10 sub-frames for every output frame) and merge them together. That's a ton of extra rendering. If, on the other hand, you can get the 3D renderer to generate "smeared geometry" (based on its knowledge of the motion speed, direction, and virtual shutter duration) and render each output frame once, that will get you faster render times.

Bit late to the party, but I can shed some light on this. Since a real camera has the shutter open for some duration, any moving light will smear across the sensor. If you similarly "smear" a path traced object by stochastically randomizing the position of the object while the path tracer is gathering samples for the frame, you get exact physically plausible motion blur without having discrete "ghosts" corresponding to sampled subframes.

Yes, everyone knows where motion blur comes from in a real camera, but in a rendering view, there is no real camera. The renderer is what applies simulated motion blur. There are some simulated optical effects must be accounted for in order to render a realistic motion blur, like depth of field. Motion blur is quite often added to stylized animation in an equally stylized sort of arcing cloud, which requires the renderer to also have knowledge of the literal 3d model that's being animated.

> Motion blur is quite often added to stylized animation in an equally stylized sort of arcing cloud, which requires the renderer to also have knowledge of the literal 3d model that's being animated.

You mean smears? That's the animation technique that literally deforms geometry, it's a kind of motion blur, and it's based on the 3d model of course. But I don't see how a gas simulation benefits from smears.

The pace of development on Blender is extremely impressive. Bravo!

Question: do companies / startups hire people who know just Blender well (and Gimp/Photoshop), nothing else? (Asking for someone else, not myself.)

I'd say the most important thing is that you're a good 3D artist and have a strong portfolio. Smaller studios have a good chance of being software-agnostic in the 3D department, unlike big AAA ones.

The thing about 3d modeling and animation is that you should actually know a few tools. Most places ask you to have a portfolio with 3d projects. They almost never ask for specific applications. If you go to artstation.com and look at the tools modelers use they typically use a few at a time. Blender is insanely good for just modeling. you'll see lots of artist use blender just to model, then substance painter for materials then import into 3dsmax, cinema4d, or maya for the renderer.

A cool note I just learned recently is that game companies are using blender to make hard surface modeling easy. Like EA is using blender to make all the levels in the Deadspace remake and importing that into Frostbite(their game engine)

real caustics!! finally we don't have to make a mask with light nodes.

A couple of gotchas that caught me: doesn't work with flat shaded objects, so no crisp gemstone caustics, objects need to have a decent amount of geometry - three levels of subdiv did the trick for me.

Reply on news.ycombinator.com