Back from the Blender Conference | Hacker News Reader

Back from the Blender Conference

valdyas.org

115 pointshiena0357 comments

Since the article spends a lot of time talking about Krita, I want to throw some praise its way as well.

I moved over to Krita from Clip Studio when I decided to go pure-Linux on all of my computers, not because I thought the software was particularly competitive or good, but because I thought it was the most likely candidate to become competitive and good.

Krita still has a lot of rough edges, but I see it on a similar trajectory as Blender. It's very rapidly improving, not just in the sense of adding new features, but in becoming a more stable, reliable product.

I was using Blender a lot earlier in its development, and I remembered the comparisons people would make to Maya, and it's just... honestly, it's a little surreal to see it now well on its way to becoming something of an industry standard. Blender used to be really annoying to use, and only niche hobbyists were using it. All things considered, its rise in popularity/quality happened fast.

So similarly, I'm very bullish on Krita, and it's very encouraging to see members of that community looking at Blender and taking lessons from its development. Krita is one of the more promising and encouraging OS software projects that I follow.

They've done an admirable job on Krita, but I don't feel the same optimism about adoption. There is a social element to these things Krita is severely lacking.

Most of the migrations I've seen have been from PS/SAI to CSP. There are a ton of artists creating custom brushes for CSP now, and you can't import them into Krita. If you're in a particular artist community/discord and want help, no one is going to know what you're talking about. That's classic lock-in.

CSP and Procreate are the hot tools on the Twitter/Instagram scene right now and I don't see the growth slowing. Krita ATM is unfortunately perceived as an outlier tool, like Medibang or FireAlpaca.

It probably isn't an issue to you personally, but those are the factors I see stymying adoption (and to be honest, I have literally never met another artist using Krita)

Well, to be very blunt, CSP is still a better painting program than Krita is. So of course more people are using it.

That being said (and keep in mind I am only one fallible data-point) the community situation is roughly equivalent to what I remember Blender being like when I first started using it (maybe in 2007-2008).

- Very few professionals were using Blender, if any.

- Most tools were being written for Maya. Every time someone would link a cool tech product with crowd simulations or plant growth, it would always be a Maya plugin.

- Any college/high-school modeling/animation courses you took would be using Maya.

- That meant that if someone wanted to learn professional animation/modeling, they'd learn Maya's paradigms, and then Blender would seem extra weird. This was back when you just had to get used to the weird 3d cursor.

- Compatibility between the two programs was awful. I remember wrestling with the options to get 3D models to export in a format that wouldn't be completely messed up when imported in Unity. I don't remember if I ever got it working.

There were people who used Blender because it was free and swore that it was just as powerful once you got used to it, but they were largely techy people. And even if they were right, none of the digital artists or studios who's work I really respected cared at all. If you wanted to get an actual job with animation, you used Maya, period. Only hobbyists could afford to spend the time learning a separate program.

So it doesn't bother me too much to see Krita in the same position. I care more about the trajectory/velocity of development, and who I see using the software. I see a lot of programmers using software like Gimp, but I see a larger focus from Krita on actual artists, and a more pragmatic prioritization of features artists use.

If the core community stays really friendly for artists, and they keep on releasing at their current pace, then I suspect adoption will eventually come. Or at least, I think it's a decent enough bet that I'm willing to use the software in its current state and regularly throw money at its development.

One factor you're not accounting for is Maya's cost put(s) it firmly out of range for hobbyists.

Considering the tools required for digital art (a Wacom, iPad Pro, or even a cheaper Huion etc) the art program is probably the cheapest purchase you're going to make, and they're all perpetual licenses (sans Adobe CC)

>I moved over to Krita from Clip Studio

You just convinced me to try out Krita again. I've had a lot of problems getting away from Clip Studio, each time it felt the options weren't ripe enough yet. I remember trying Krita a very long time ago and it was barebones, but sounds like it's had a lot of good development time.

I wish there were an open source design app with as much thought behind UX as Krita.

I got Affinity Designer just as soon as I found out about it after many failed attempts to understand Inkscape. There's just no comparison. Inkscape needs to have that moment of focus that got Blender over its UX wall.

Boy I wish Affinity would make Linux ports of its products, but it will probably never happen.

Got $500k laying around? https://forum.affinity.serif.com/index.php?/topic/626-affini...

I'm sure they're in better financial shape four years later, but that's a lot of money to invest without assurance at least 10,000 people ($500,000/$50 sale price) will buy it.

This blog post is much more interesting than the (ambiguous) title might suggest. The blog post is really all about comparing the funding models of Blender with other OSS projects - and aiming to demonstrate that the Blender funding model is successfully creating a valuable piece of OSS. Wrestling with funding is a major challenge for OSS projects, even if they don't believe they require funding.

Well, I didn't want to post click-bait on my blog :-)

Linux Desktop users do indeed seem to spend a disproportionate amount of time bickering about which popular apps best fulfill their purity tests:

whether the app is "free" enough for their liking (MIT vs GPL-based license, etc),

whether it picked the "right" frontend in the never-ending GTK vs QT battle,

whether it is written in an "acceptable" language (Electron being Satan Incarnate, of course),

whether it has been properly packaged as a DEB/RPM/Snap/Flatpak/AUR/AppImage, etc.

Fail any one of these things, and the app is instantly "unusable" and we should all use ncurses-based obscure-thing-I-found-over-the-weekend-in-some-dude's-PPA.

Meanwhile, an increasing array of "free as in freedom" apps that are widely available and cross-platform (e.g. Blender, VLC, Audacity, Calibre, etc.) serve the larger community by just letting people get shit done, and build up name-recognition as a result.

> whether the app is "free" enough for their liking (MIT vs GPL-based license, etc),

I know it is tangential to this point, but a lot of developers, including long time Free Software developers, don't get licensing at all. While packaging software (FOSS) for a distribution, I've had to poke multiple people to fix licensing problems that invariably occurred (most common: incompatible licensing of parts of a project, perhaps imported from somewhere else).

> whether it is written in an "acceptable" language (Electron being Satan Incarnate, of course)

There are legitimate technical reasons for not liking Electron, not necessarily due to "disproportionate amount of time bickering" spent by "Linux Desktop users".

> a lot of developers, including long time Free Software developers, don't get licensing at all

Completely agree; it's a complex part of producing software that people (myself being guilty of this) incorrectly assume they can ignore.

> There are legitimate technical reasons for not liking Electron, not necessarily due to "disproportionate amount of time bickering" spent by "Linux Desktop users".

There are legitimate technical reasons for not choosing electron, but I can't think of any for not liking it. It's a perfectly fine and useful way to make a cross-platform desktop application. If it doesn't suit your project's purposes because of performance, size, or current acumen, that's fine.

But I think OP is referring to "Electron is awful", "I hate Electron", or "Never use Electron" levels of ire that crop up occasionally, which should be exclusively reserved for something that no one should be working in.

> It's a perfectly fine and useful way to make a cross-platform desktop application.

My main problems with Electron are:

- The size of the runtime

- The fact that every application has to bundle it and it means that multiple Electron applications on the same system do consume a lot of resources if not handled well

- The fact that unless you (as a developer) play due diligence, you might distribute your app with a version of the runtime which may contain vulnerabilities (one of the reasons Linux distributions don't like library bundling that much).

I can't deny it's probably easier for many (not for me, but although I write Python code all day, I'm by no means a "developer"), but it can be potentially wasteful, to say the least, and require far more resources than what you'd actually need with a another toolkit.

That said I for sure won't point a gun at someone who wants to use it. ;)

This comment is so beautifully in-character for the thread, and completely fulfills OPs characterization of people that "do indeed seem to spend a disproportionate amount of time bickering about which popular apps best fulfill their purity tests".

If they didn't care about software they'd just use whatever came pre installed on their computer

We are here because we are deeply interested the software that we're using. If one group of programs does something that tends to suck, then we're going to call it out.

Just remember, you get what you pay for.

To be honest I spent at best 3 minutes writing that post; compared to the time I put in FOSS projects...

> but I can't think of any for not liking it.

I distinctly remember a post on here about three or four years ago where a blind person detailed losing their job because Electron does not integrate with desktop accessibility functions.

> whether it has been properly packaged as a DEB/RPM/Snap/Flatpak/AUR/AppImage, etc.

Not sure if you're referring to a choice between packaging formats, or not packaging at all. If the latter, well, it's pretty tiring looking through a list of filenames which contains the platform and arch, downloading the correct one, then going to the project's wik and copy/pasting commands to create users, create systemd files etc. (Looking at you, Go devs). Oh, you want to upgrade it along with the rest of your software on your system? Fuck you.

I get it, packaging isn't painless. But if you've worked hard writing a piece of software, it seems bizarre not to spend some time on packaging and getting distros to accept it.

We like packaging, because it's awesome. A large collection of extremely trustworthy software, all a single, simple command away.

> […] it seems bizarre not to spend some time on packaging and getting distros to accept it.

My personal experience is that it means, at the minimum, having a VM for that distro hot and ready to go, reading a bunch of documentation about how that distro does packaging, consulting an oracle to tell you which versions of that distro you should target, learning how this particular distro wraps your particular build system, translating all the dependencies, etc.

So it doesn’t seem like just “some time” to me. If you pass this off to a maintainer for that distro, they’re usually fairly experienced with that distro’s ecosystems and the tools.

Blender is an example of a project that is laser focused on their users. Every aspect of their development is guided by that focus. The Open Movie projects are a way of ensuring that what they work on is something that professionals need. You can what you want about blender but their continuing growth in industry demonstrates that their focus is paying off. They are quite literally closing the gap with tools like Maya at ridiculous speeds.

You got my point :-)

IMO it is however easier than other "desktop" applications, because it has an narrower focus, like Krita. Some applications are easier to focus than others (where usage patterns may vary wildly).

Excerpt:

"Blender’s development fund currently brings in about 1,200,000 euros a year, which funds 20 full-time developers. That’s not the only source of funding. Blender has about 172 developers in the past year, and 550 over its entire existence, and 64 in the past month, same as LibreOffice. Looking at the last number, it means that there are anyway more volunteer committers in the Blender community than paid developers. Funded development hasn’t eaten the community.

Let’s hazard a guess: Blender has four times the installed base of AutoDesk Maya. This is pretty rough, of course, so ingest with salt.

My thoughts:

These are important non-software aspects of Blender which are critical for acceptance by new users, which is critical to maintain a healthy user base, which should in turn yield a well-community-supported codebase...

In other words, this makes Blender a safe choice in corporate environments where a choice may exist between it and commercial offerings, such as Autodesk's Maya...

> about 1,200,000 euros a year, which funds 20 full-time developers

Wow not paying them a lot are they.

60k Euros is not that bad. You can live comfortably with that in Europe.

I think employing someone has about a 50% overhead, so it's really $30.

I'd imagine they aren't paying any cut to recruiters. I guess that leaves real estate, equipment, contractor, and legal/accounting (assuming outsourced) overhead.

If it's Europe my understanding is there's substantial expenses paid to employment security overhead, it's how they pay for things like the exceptionally good maternity leave and unemployment benefits type stuff.

I would definitely work for blender for 5k a month. That's a top 1% income bracket in many great countries in Europe.

That’s $5k a month to pay all taxes and employee expenses... not $5k take home - you’ll get hardly any of it at the end!

Even so, his point still stands. The upper range of developer salaries in Czech Republic, before taxes, is $3000 to $5000. The country is considered Central Europe.

If you go for Eastern Europe, Hungary/Poland/Ukraine, that's enough to afford the best people.

Income tax (at some bracket) is a percent of income. What does it matter if it's "5k to cover taxes" lol, it's not like its a flat expense that 5k may or may not cover.

Even 3000€ or even 2500€ take home pay per month would let me live a very good life (for me) in a sunny, safe, beautiful place in Europe. All the while I'm working for a free software project, one which I'm a fan of (not working to make my boss rich, not inventing better ways to make people click on ads). What else could I possibly ask for.

What I’ve been impressed with following Blender for a long time is they communicate their development well with their users. They provide demos, give discussions, give example files and videos, and make you realize that the donations are going to something tangible. Other projects don’t do nearly as good of a job with this.

So, they have invested in marketing.

I think early FOSS programmers didn’t value marketing enough, which produced excellent software with difficult adoption. Now, for any software to succeed, a nice UI and good marketing material is the baseline. FOSS that is born since 10 years was much nicer (from Angular to Kibana).

It is even visible in security breach: At one point after Heartbleed, there was a discussion on whether security bugs would now require a nice single-use website with graphic artifacts, in order to get heard by developers.

This is the first time I hear Krita but I support all good open-source projects like Blender

What I like in Blender, except its invaluable community, is its UX, they follow a good logic in interfaces. If someone is familiar with 3Ds Max or Maya, in less than an hour can understand the whole software and how to work with it even without any help or tutorial

However, I'm not in the VFX industry and I'm an architect and Building Information Modeling/Management - BIM technologist, so recently we've started to improve Blender for Digital Built Environment industry, an open-source plugin (interfaces) called BlenderBIM [1]

So hope we help Blender and Blender community to speed up its success

[1] https://blenderbim.org/

This is something I really love with Blender and other Free Software project. You can create new plugins to extends the scope of the project.

Btw did you know that the icons in blenderbim.org homepage are icons from the Breeze icons theme developed by the KDE community (the community behind Krita)?

I installed Blender at the school I work at because it had an MSI installer. Did not install Krita because it did not. If you want it to be the default used in schools, make it as easy to deploy as possible.

The gimp fork Glimpse is planning an MSI installer, so that'll be installed soon

This is a little bit surprising to me, at $DAYJOB i had some requests for an MSI installer. I knew there's some AD/GPO features or whatever but your level of seriousness here is unexpected.

Is it really easier to install compared to an exe installer (say nsis / inno-based)? How so?

Way easier with the default tools. Companies with money will buy sccm etc. But for those orgs without, MSI or writing a script are the only default options. And scripts are unreliable

Imagine the difference between software distributed as a tarball and the binary package. Both totally work, but one is so much easier to deploy in bulk. (This is obviously a very loose analogy, but I think it's roughly correct)

Exe installers support unattended, headless installs with a command line switch.

But how you do run that command only once on every computer? Including ones that are currently not in the building, or on.

Same way as anything, use group policy or a login script that checks first.

But how do you know if it's installed correctly?

What does correctly mean? On windows usually a registry key is checked.

Exactly, changes from software to software. And how to uninstall?

Much easier to add an MSI to GPO, and then just as easy to remove it again and it's uninstalled.

Not sure how writing scripts to do all that myself for every piece of software is better.

Msi is deprecated.

So you keep saying, although don't see any evidence for it. MS want you to move to using the MS Store, because that's not terrible. Or manually download multiple files, put them all in specific places and use a powershell script to install it, hah, so easy.

MSIX is just a wrapper around an msi, exe, etc installer anyway.

In conclusion, just make an MSI file, they're much better, look at all these great features https://en.wikipedia.org/wiki/Windows_Installer

Msi is deprecated.

I am not sure if this is relevant to you, but Krita is available from MS Store, so you could get automatic install and auto-updates.

Can't think of anything more horrifying than trying to deploy apps via MS Store

On Linux Desktop funding: companies like Red Hat and Canonical fund much of the development on things like GNOME.

The real budget is likely much higher than the article suggest, especially now that Ubuntu fully switched to Gnome.

Yep. Red Hat employs quite a few people who work on Gnome and Gnome apps, as done Canonical. Despite being employees for a company, they are extremely involved in the open source Gnome community. It's an amazing example of GPL success IMHO.

Disclaimer: I work for Red Hat but not on Gnome

A minor correction - the article makes a couple of mentions of "Handbrake donation[s]", but I think the author was referring to the donations made to KDE and Gnome by the Handshake project.

https://www.gnome.org/news/2018/08/gnome-foundation-receives...

Reply on news.ycombinator.com