Can you imagine if everyone paying monthly subscription for any type of software took 10% of their budget and donated to the alternatives? I'd bet that any FOSS project could reach parity to their proprietary counterpart in three years.
Seconded. I don't know much about the code, but on a usability level it sometimes feels even more polished and pleasing to handle than many big proprietary offerings. I would go so far as to say it's one of the best executed desktop programs of the current era.
Just to share one datapoint - I have not looked at the internals myself, but having talked to a couple of devs who has spent a great deal of time working with Blender at a code level, one takeaway seemed to be that there was a lot of code duplication caused by each dev doing their own thing and not having an overall architect to rationalize some parts of the code.
A particular example that was mentioned was the importers that were each done standalone rather than built on a common framework. This was a couple of years ago and of course things may have changed since then.
Just to note that the developers were still extremely positive about working with and in the code, but it does seem plausible that there would be issues like this in a codebase where many devs focus on a small piece of the overall puzzle.
FWIW, Blender, like other open source projects has been dealing with large bandwidth bills due to AI companies repeatedly crawling them. They've been adding additional checks to try to figure out how to stop spending their limited donations on server infrastructure to support bots.
Getting a lot of undeserved pushback, but I sympathise. I have severe CAPTCHA fatigue. Any time I encounter one, unless it's something I need to use and must access, I back out. Even the proof-of-work ones. Fuck 'em. 10 seconds wasted many times a day adds up, and since I use a "strange" OS (Debian or Gentoo Linux) and a "weird" browser (maxed out uBlock and Cookie-Autodelete), ClownFlare hates me and I hate them right back in return.
It does feel like Blender is eating the world of 3D content creation.*
I know the industry standard of animation is still Maya, and geometry node isn't as powerful as Houdini, but Blender has made such great progress so far... I wonder if there any hobbyist/student who learns 3Dsmax or 3DCoat as their first DCC anymore?
*: Assuming genAI won't deliver and make that world disappear entirely, ofc.
> Assuming genAI won't deliver and make that world disappear entirely, ofc.
I get the concern and the reasoning behind it but having spent most my adult life and career in digital content creative tooling (2D, 3D and video), I believe AI will be short-term disruptive but also a long-term net positive. It should make skilled pros faster and more productive while helping non-pros achieve more of their creative vision on their own.
For over 30 years progress in tools has been about giving creative people the power to realize their visions. To create more and do it faster, cheaper and at higher quality. Of course, they don't always choose to use that power well but the concern that "these new tools are just enabling bottom feeders to create more bad content" has remained a pretty constant refrain since I started in 1989. Even back in the 90s I said "Sure, these new tools will enable 95% more crap but they'll also unleash 5% that's great which wouldn't have existed before." I think that's just as true today as it's always been.
> It should make skilled pros faster and more productive while helping non-pros achieve more of their creative vision on their own.
The problem is, AI is good enough to replace juniors. That means companies are already cutting positions at that level and some are just itching to ditch intermediates as well once quality improves.
But when juniors and intermediates are all gone... how are the beancounters expecting to get new seniors when the old ones go to retirement or hand in their 2 week notice because they are fed up cleaning up after crap AI?
Disruption is disruptive and can suck. But it's also not new. Some companies and people will do short-sighted things in response and pay the price. Other job types and roles will simply be disintermediated and those individuals will need to learn new skills and find different kinds of jobs in the evolving landscape - same as always.
What happens is those companies will end up paying a lot of money for people with no experience when they end up being the only ones available to hire.
Same thing happens in tech in cycles where companies fire all the juniors, and every time theres a boom, they hire them all back at obscene prices.
> The problem is, AI is good enough to replace juniors. That means companies are already cutting positions at that level and some are just itching to ditch intermediates as well once quality improves.
Who needs companies? Those sounds like an unnecessary middleman to me.
I don’t see much risk to creatives and workers (hold the number of workers constant and have them all get more productive, and the world is a better place).
But, I do wonder about smaller teams accomplishing more, which the unit of work-doing-people can be much smaller. The org tree can shrink, probably knock out some levels. Maybe a team of 4 that does the work previously done by 10 finds it easier to just have someone directly interface with the customer, rather than needing a layer of project managers and customer service to receive customers messages and distribute them.
I’d be worried if I was… anywhere above an IC really.
> Maybe a team of 4 that does the work previously done by 10
Interestingly, I heard similar ratios were being discussed (or feared/lamented) in the mid-90s as computer-based non-linear video editing swept the post-production industry and in live TV production as computer-based all-in-one production switcher, effects, titler systems like the Amiga-based Video Toaster disrupted everything. I still remember hearing about how the Toaster was giving unionized TV stations problems because union rules said the switcher guy wasn't allowed to touch the graphics/titling system or the editing system. So three guys would be standing around one chair depending which tab was up on the Video Toaster screen :-).
Having been around the content creation tooling business for so long has given me perspective. Since at least the late 80s it's been one never-ending disruption. And, despite the constant predictions of jobs lost, today there are far more jobs in content creation than there were then. Of course, the job descriptions, types, skills required, divisions of labor and which roles were more or less valuable have never stopped changing and probably won't. Yes, this ends up displacing some people in some roles. Back in the 90s highly-paid senior video engineers with deep expertise in how to wire up and synchronize analog video systems initially laughed at us "young kids" with our "toy" digital video systems. Then they resented needing to call us in to get computer-based gear installed, interfaced and working. Then they resented that us "new kids" were paid less due to being less senior but becoming increasingly essential. Some of them learned the 'newfangled' digital systems while others didn't and opted to retire or do something else.
Over thirty years later, I'm now the highly-compensated "old guy" with once-invaluable expertise that's depreciating by the minute and proven skills at doing increasingly irrelevant things. The particulars change but the theme remains the same, which is why I don't think the shift to integrating AI in content creation will be significantly different. There will be skills that become less valuable and job types that go away and entire industry sectors which get disintermediated but at the same time new kinds of work, industry sectors and business opportunities will emerge in different places and ways. As always, the new job types won't be 1:1 replacements for the old, and this will cause as much angst as it always has. They'll not only look different and have different kinds of trade-offs, they may even have different business and compensation models.
It's Schumpeter's process of "Creative Destruction" and renewal (https://en.wikipedia.org/wiki/Creative_destruction). The destruction part is fast, loud and shocking as what exists today combusts in flame. The creative renewal part tends to be more gradual, unfamiliar and at first may not even seem related. From time to time, even the definition of "the industry" ends up changing along with the jobs and roles. I'm pretty sure the old-school TV station engineers I learned from wouldn't consider a digital nomad working full-time editing 30 second clips on a mobile device for YouTube, Instagram and TikTok influencers as even being "in the TV business."
blender is the python of 3d animation. Its the second best tool for everything. Which btw is not a digg. its really hard to be that level of reliable accross an entire workflow.
Maya hasn't changed much in a decade, and that is a wise choice for training/documentation. They focused on the content part which is its value proposition.
Blender has always had the perpetual Beta problem, as many boring core design issues were never really given priority. Fine for developers, but a liability in commercial settings.
Houdini is interesting, but with Blender Geometry-nodes now working it is unclear how another proprietary ecosystem will improve workflows. =3
The Entagma channel covers a lot of Houdini and Blender bleeding-edge features with short lab tutorials:
I've only dabbled with Blender but from what I saw of geometry nodes it's quite a long way away from competing in the same space as Houdini. Houdini's biggest single feature and the thing that allows film studios to use it at complexity and scale is the HDA system and there's no real competition for that.
I think the main difference between OSS like Blender and the competition is it seems like OSS only gets better. Each version is a little bit better, so projects that were once not competitive catch up. Blender is an obvious example, but also look at Krita, or the entire KDE project. These pieces of software age like fine wine: they get more features, they get faster, and they get more stable.
Closed-source software seems to get... stuck. In the best case. Often, they regress: becoming buggier from version to version with less features. I think of Windows and the entire Microsoft suite of applications.
I think one exception is Gnome. Gnome loves removing features more than they love not implementing popular Wayland protocols.
With OSS, unless every add-on is part of the build tree or package repo testing... it quickly becomes broken as the ecosystem constantly evolves.
i.e. of the $3k of community Blender plugins/add-ons we evaluated last year, only around 60% are still functional in stable blender releases. Additionally, many built-in core features like Fracture became broken in 4.x due to API permutations, and getting split into its own module.
In a production environment one must version lock Blender for the project. =3
Houdini is entirely a node-based system and HDAs are just your custom nodes (on contrary to the built-in ones).
At the end of the day, an HDA is just a 'function' you define like what you do in a programming language. A function can call other functions etc. It might sound nothing special if you're a programmer, but Houdini is the only generic DCC that is built around this idea, making it more like a framework than an app.
Geometry node is not Houdini (yet) though. The main reason people use Houdini is it's easy to transfer data between modeling (SOP) and physical simulation (DOP) contexts. And they're working on integrating rigging/animation too.
Geometry node is quite a separate thing from what Blender already has. Last time I checked one couldn't even create vertex groups in geometry node, and there was no way to create an armature there either. (Not sure if it changed since I checked though)
I think you should've be able to access vertex groups from the initial version of geometry nodes. Certainly, at this point, you can dynamically access (and set) arbitrary vertex groups, including loading the names of the vertex groups you want to access from a csv file, or otherwise synthesizing the names using string operations.
its actually funny, the first version of geometry nodes was actually more powerful in that way. it didnt provide many built in nodes, but the nodes it did provide were arbitrary property setting/getting. the current and newer geometry nodes system based on "fields" is way easier to actually work with but they don't give you raw access to properties. so like in the first version of geometry nodes you could write custom mesh normals just by setting the "normal" attribute with some vectors, but then from blender 3.0 to 4.4 you could no longer set custom normals in geometry nodes and they only added a set normals node in 4.5
I found the Entagma lab videos very helpful in understanding how to parse geometry with nodes. It is not obvious (unless you already get CLI pipes), but so worth it when things finally work... =3
That's often a problem with Open Source. Lots of people steering the ship, and it's hard for any de facto captain to really demand a direction. If they do, they may not have the contributors needed to really see it through.
This can be solved with paid contributors, but FOSS organization don't have the most funding out there. So it can be challenging when trying to attract specialized talent.
> Assuming genAI won't deliver and make that world disappear entirely, ofc.
Even if genAI were to kill the cinema/animation back-end industry, I still think 3D software would be needed to continue videogame development, and godot is far behind in modelling/animation capabilities compared to blender. Hopefully they can latch on this in the long term if it happens that AI keeps eating artists lunch.
My daughter did the Blender donut tutorial and put her 3d animation into the year end art show when she was 10. Having a great community and resources behind the project makes if very easy for students to pick up.
> *: Assuming genAI won't deliver and make that world disappear entirely, ofc.
The current state of things leaves a lot to be desired. I have yet to see a model, or technique that actually tackles real 3D content creation problems, aside from animation of course.
Most solutions are trying to solve things completely orthogonal to what you really need in the real world. They produce bad meshes, if meshes at all, have bad topology or inconsistent point sampling, and are extremely prone to parroting their training data.
It's kind of incredible how far we have come but seriously, to understand a problem space you need to try to use the existing tools. Only then, do new, and useful ideas pop up that solve real issues. But man, nobody in the space seems to have a clue.
Believe me, 3D modelling software is REALLY good at what it does, it is really hard to try and dethrone human and machine working together in these really well thought out pieces of software.
Custom mesh normals is huge, because this used to be a PITA.
Tugged away at the end is defining custom cameras through OSL, which is a very very interesting feature, because it allows more physically accurate lens emulation, with dramatically more interesting bokeh than the standard perspective camera model: https://www.reddit.com/r/blender/comments/1kehtse/new_camera...
yeah man i was faking it by creating custom materials and setting vertex attributes containing the distance to a specific other object, then in my material using the attribute and the other objects info to blend between the normals at the material level
A wonderful app that, I find, benefits from LLMs to help you figure out how to use it. That or you more or less have to dive in for months and months. The people on YouTube that make Blender look so easy … Blender is really all they do. ;-)
At some point I see an LLM more or less integrated into the UI.
At some point I see whole apps written so complexly that an LLM is the required interface.
> That or you more or less have to dive in for months and months.
Waaaaay back in the Quake 3 days I had a pirated copy of 3D Studio Max and decided to try and make a Quake (or maybe it was Half-Life) character model. Found an on-line tutorial that step by step showed you how to setup Max with a background image that you "traced" over. So I grabbed the images of front, back and side views of a human, loaded them into the various view ports, then drew a rectangle from which you extrude a head, arms and legs. Then you manipulate the block headed human mesh to fit the human background image - extruding more detail from the mesh as you go along. In one day I had a VERY crude model of a person. I also found out I dont like 3D modelling. Though I'd say a person who really enjoys it would pick it up in a week or two with good tutorials.
LLM's just cut out the learning part leaving you helpless to create anything beyond prompts. I am still very mixed on LLM's but due to greed and profit driven momentum, they will eat the world.
One problem I find is that a lot educational content has moved into YouTube and videos (monetization be damned). I have no time to watch 10mins of rambling and ads for a quick tip, LLMs are great at distilling the info. Otherwise, I agree, deep knowledge building only happens through doing stuff…
Oh boy. Giving Blender instructions in English, on every level of detail, could be quite a workflow. Like, have it generate a range of chairs, pick the mesh you like, then start asking it to tweak parts of it until you are satisfied. Add finishing touches manually.
In general, model re-topology almost always requires a manual editing process post sculpting and texturing. There have been countless attempts to automate the process, but it usually requires working through an obscure book about rigging with secondary transforms etc.
If you are pushing vertices around to make a model, you eventually have to think of topology issues if you ever want to make a good animation.
And then when you have a good topology... Then You need to rework the UV Maps, Normals and other such data so that it's all consistent.
LLMs are not the right tool for this. I'm sure some AI retopology tool can come around but there is a HUGE difference between a model designed for stills, a model designed for animation and finally, a model designed for close up facial animations.
The topology differences alone are insane.
---------
Now maybe in the future an Ubermesh (like the Uber shader / principled shader) can be made to make these processes consistent. And then tooling can be made over the hypothetical Ubermesh. But today it's a lot of manual retopology that requires deep thinking about animation details.
Ex: an anime style mouth doesn't usually move the jawline (Think Sonics mouth: its basically a moving hole that jumps around his face). Meanwhile, western 3D animation (Baldurs Gate and the like) obviously have fully made lip sync including phhfff, eeee and ahhh sounds.
The 'Kind' of animation you want leads to different mesh topologies. It's just the nature of this work.
And people get ANGRY when you ex: animate Sonic with traditional teeth and jawline (see 'Ugly Sonic'). You can't just willy nilly pick one style randomly. You need purposeful decisions.
--------
I guess if you are a pure sculptor / stills then you don't have to worry about this. But many 3d projects are animated, if not most of them. The retopology steps are well known.
LowPoly look will almost certainly require manual effort (these topology problems tend to disappear with more triangles).
Blender is often used to project the high-granularity sculpted detail onto a low-poly (rigged and clean topo) Normal map. i.e. you can get great looking assets that won't hit your poly budget as hard. When it works its great...
People have feelings, and sometimes may misinterpret others intent. I am happy we've met if we both learned something. =3
I have not tried it. Instead I have been asking Claude (etc.) "How do I create a repeating triangular truss..." or what-have-you. And then I follow the steps they list.
I don't know if it supports HDR on MacOS but, AFAIK, it doesn't on windows and in Linux it is only supported with Wayland.
Though I don't like the Wayland x X11 flamewar, I'm happy to see some modern features are only supported on Wayland.
That may please the crowd that will be able to say "sorry, I can't use x11 because it doesn't support a feature I need" bringing some symmetry to the discussion.
> I'm happy to see some modern features are only supported on Wayland.
Why?
If anything that's a reason to why I wouldn't fully jump to blender.
I have been working on my own hobby game engine for the past 15 years and have been excited to introduce Blender to the workflow. If this is the case I don't like it. Wayland has never work for me the same way as X has.
If the starting point is that Wayland is missing features that X has, the good outcome is not getting to a point where neither option has all the features, the good outcome is that either one has all the features.
That's at the cost of lots of duplicated work by the already sparse number of people capable of implementing a graphics server.
There's also a third option where Wayland is foundational and the X11 network protocol is implemented on top of that for people who need it. Why should a network GUI service implement a driver to talk to a specific model of video card?
> That's at the cost of lots of duplicated work by the already sparse number of people capable of implementing a graphics server.
Yeah, it kinda sucks but this is where we are.
> There's also a third option where Wayland is foundational and the X11 network protocol is implemented on top of that for people who need it. Why should a network GUI service implement a driver to talk to a specific model of video card?
Agreed; I have long argued that it would have been far better to transition to everything on the same backend with effectively rootful XWayland being the only (bare) X server, and then after that try to deal with the rest of the stack (if you really must). And maybe in 2026 we'll finally start to see movement in that direction with https://gitlab.freedesktop.org/wayback/wayback
It definitely does on macOS and I think also on Windows. You have to set the color management for the viewport to Display P3. In older versions this precluded you from using AGX or Filmic, but I think you can actually use AGX with Display P3 now.
I really wish to could use Wayland, but there is too much problems or bugs related to the software I use for work and also play. I will test it again with this new version Blender (that was one of that software with problems).
In general, most repositories use old stable versions of Blender. And often folks are reduced to using snap to maintain version specific Compatibility with add-ons etc.
Also, getting the most out of Intel+RTX CUDA card render machines sometimes means booting into windows for proper driver support. Sad but true... =3
The reality is most commercial software and users are on Windows machines. It is fundamentally a Blender interoperability, and 3rd party platform license compatibility issue. We all wish it wasn't so, as many artists find the Windows file systems and color-calibration concepts bewildering.
Making a feature platform specific to a negligible fraction of the users is inefficient, as many applications will never be ported to Linux platforms.
Blender should be looking at its add-on ecosystem data, and evaluate where users find value. It is a very versatile program, but probably should be focused on media and game workflow interfaces rather than gimmicks.
I agree with you, but I think this limitation is for much simpler reasons, like "the contributor only knew how to make this feature in Linux, and only in Wayland". cross compatibility for stuff a base as color grading can be a thorny issue.
If nothing else, it's better to have some implementation to reference for future platforms than none.
Indeed but this is a discussion about Blender and you posted originally:
> Making a feature platform specific to a negligible fraction of the users is inefficient, as many applications will never be ported to Linux platforms.
All the large studios use Linux, that's why all the third party software that is used in feature animation and vfx is supported on Linux. So I'm just saying 'negligible fraction of users' in the case of Blender (which as a project would like to increase adoption in professional feature animation and vfx) isn't really true.
I am sure Studios account for a small portion of the 4.5 million unique downloads each release. Note that less the 20% of users ever touch film or animation projects, 73% are single users, and most related user applications are Adobe products.
Stats are available from the published 2024 feedback data:
I'm not sure download stats are hugely relevant because that would imply the needs of every person that downloads Blender are weighted equivalently which would make little sense.
Or are you suggesting the Blender foundation has no interest in getting wider adoption among film and animation studios?
I think the foundation projects hold a lot of potential, but what they release as "stable" is rarely ready for a production setting. People do use Blender for small side tasks commercially, but would you honestly bet your company reputation/job on their 31 years of shenanigans?
Updates still permute the core to break parts of the program, and brick countless add-ons or custom code. People turn users into Beta testers, partners into IT support, and hide workflow details under layers of feature-creep kludges.
When Blender updates for feature X, they will usually brick feature Y. YMMV
The Foundation may intend to improve user adoption, but they can't even cover there own unit-tests on internal add-on code. =3
If you get value out of Blender they could use your support: https://fund.blender.org/
Even just a coffee a month will help immensely. Please consider supporting :)
Always support open source when you can, even if you use proprietary software instead of Blender.
Supporting open source creates competition for your paid alternative, so they're forced to give you a better product or a better deal.
Can you imagine if everyone paying monthly subscription for any type of software took 10% of their budget and donated to the alternatives? I'd bet that any FOSS project could reach parity to their proprietary counterpart in three years.
The whole point is subscription software is you don’t have 10% left of your budget to finance anything else.
You can also/alternatively subscribe to Blender Studio:
https://studio.blender.org/
You get access to training, assets, source and production logs like the recent Dog Walk update while also supporting Blender:
https://studio.blender.org/projects/dogwalk/production-logs/
I could tip them even without using it, the value of such a successful creative project is a pleasure in itself.
This is the only project I support year in year out. What they did is amazing. Simply amazing.
Especially software like Blender.
It's easily one of the most well-made FOSS projects out there!
I maintain a Blender fork at my company and Blender quite literally is my paycheck, but I disagree vehemently that it is a well-made FOSS project lol.
So .. can you give reasons?
Seconded. I don't know much about the code, but on a usability level it sometimes feels even more polished and pleasing to handle than many big proprietary offerings. I would go so far as to say it's one of the best executed desktop programs of the current era.
Just to share one datapoint - I have not looked at the internals myself, but having talked to a couple of devs who has spent a great deal of time working with Blender at a code level, one takeaway seemed to be that there was a lot of code duplication caused by each dev doing their own thing and not having an overall architect to rationalize some parts of the code.
A particular example that was mentioned was the importers that were each done standalone rather than built on a common framework. This was a couple of years ago and of course things may have changed since then.
Just to note that the developers were still extremely positive about working with and in the code, but it does seem plausible that there would be issues like this in a codebase where many devs focus on a small piece of the overall puzzle.
[flagged]
FWIW, Blender, like other open source projects has been dealing with large bandwidth bills due to AI companies repeatedly crawling them. They've been adding additional checks to try to figure out how to stop spending their limited donations on server infrastructure to support bots.
I believe they were also outright ddos'd last year, by someone who apparently has a beef with Blender.
This isn't reddit your opinion isn't necessary. Donate or don't but your pithy commentary isn't adding to the conversation.
Getting a lot of undeserved pushback, but I sympathise. I have severe CAPTCHA fatigue. Any time I encounter one, unless it's something I need to use and must access, I back out. Even the proof-of-work ones. Fuck 'em. 10 seconds wasted many times a day adds up, and since I use a "strange" OS (Debian or Gentoo Linux) and a "weird" browser (maxed out uBlock and Cookie-Autodelete), ClownFlare hates me and I hate them right back in return.
Counter-point: Stop subsidizing AI companies by giving them free training data.
Counter-counter-point: I like running free models on my computer that were trained on freely accessible data.
We all want free stuff, but somebody still has to pay for that stuff
Information is not "stuff".
Information is absolutely stuff.
How much does that thought cost? How much does it weigh?
Information absolutely has a cost, somebody had to create it.
Weight is just nonsense. Physical presence is not a hard requirement for value.
It does feel like Blender is eating the world of 3D content creation.*
I know the industry standard of animation is still Maya, and geometry node isn't as powerful as Houdini, but Blender has made such great progress so far... I wonder if there any hobbyist/student who learns 3Dsmax or 3DCoat as their first DCC anymore?
*: Assuming genAI won't deliver and make that world disappear entirely, ofc.
> Assuming genAI won't deliver and make that world disappear entirely, ofc.
I get the concern and the reasoning behind it but having spent most my adult life and career in digital content creative tooling (2D, 3D and video), I believe AI will be short-term disruptive but also a long-term net positive. It should make skilled pros faster and more productive while helping non-pros achieve more of their creative vision on their own.
For over 30 years progress in tools has been about giving creative people the power to realize their visions. To create more and do it faster, cheaper and at higher quality. Of course, they don't always choose to use that power well but the concern that "these new tools are just enabling bottom feeders to create more bad content" has remained a pretty constant refrain since I started in 1989. Even back in the 90s I said "Sure, these new tools will enable 95% more crap but they'll also unleash 5% that's great which wouldn't have existed before." I think that's just as true today as it's always been.
> It should make skilled pros faster and more productive while helping non-pros achieve more of their creative vision on their own.
The problem is, AI is good enough to replace juniors. That means companies are already cutting positions at that level and some are just itching to ditch intermediates as well once quality improves.
But when juniors and intermediates are all gone... how are the beancounters expecting to get new seniors when the old ones go to retirement or hand in their 2 week notice because they are fed up cleaning up after crap AI?
Disruption is disruptive and can suck. But it's also not new. Some companies and people will do short-sighted things in response and pay the price. Other job types and roles will simply be disintermediated and those individuals will need to learn new skills and find different kinds of jobs in the evolving landscape - same as always.
I just wrote a longer reply to a sibling comment addressing exactly this: https://news.ycombinator.com/item?id=44574895.
What happens is those companies will end up paying a lot of money for people with no experience when they end up being the only ones available to hire.
Same thing happens in tech in cycles where companies fire all the juniors, and every time theres a boom, they hire them all back at obscene prices.
> The problem is, AI is good enough to replace juniors. That means companies are already cutting positions at that level and some are just itching to ditch intermediates as well once quality improves.
Who needs companies? Those sounds like an unnecessary middleman to me.
I don’t see much risk to creatives and workers (hold the number of workers constant and have them all get more productive, and the world is a better place).
But, I do wonder about smaller teams accomplishing more, which the unit of work-doing-people can be much smaller. The org tree can shrink, probably knock out some levels. Maybe a team of 4 that does the work previously done by 10 finds it easier to just have someone directly interface with the customer, rather than needing a layer of project managers and customer service to receive customers messages and distribute them.
I’d be worried if I was… anywhere above an IC really.
> Maybe a team of 4 that does the work previously done by 10
Interestingly, I heard similar ratios were being discussed (or feared/lamented) in the mid-90s as computer-based non-linear video editing swept the post-production industry and in live TV production as computer-based all-in-one production switcher, effects, titler systems like the Amiga-based Video Toaster disrupted everything. I still remember hearing about how the Toaster was giving unionized TV stations problems because union rules said the switcher guy wasn't allowed to touch the graphics/titling system or the editing system. So three guys would be standing around one chair depending which tab was up on the Video Toaster screen :-).
Having been around the content creation tooling business for so long has given me perspective. Since at least the late 80s it's been one never-ending disruption. And, despite the constant predictions of jobs lost, today there are far more jobs in content creation than there were then. Of course, the job descriptions, types, skills required, divisions of labor and which roles were more or less valuable have never stopped changing and probably won't. Yes, this ends up displacing some people in some roles. Back in the 90s highly-paid senior video engineers with deep expertise in how to wire up and synchronize analog video systems initially laughed at us "young kids" with our "toy" digital video systems. Then they resented needing to call us in to get computer-based gear installed, interfaced and working. Then they resented that us "new kids" were paid less due to being less senior but becoming increasingly essential. Some of them learned the 'newfangled' digital systems while others didn't and opted to retire or do something else.
Over thirty years later, I'm now the highly-compensated "old guy" with once-invaluable expertise that's depreciating by the minute and proven skills at doing increasingly irrelevant things. The particulars change but the theme remains the same, which is why I don't think the shift to integrating AI in content creation will be significantly different. There will be skills that become less valuable and job types that go away and entire industry sectors which get disintermediated but at the same time new kinds of work, industry sectors and business opportunities will emerge in different places and ways. As always, the new job types won't be 1:1 replacements for the old, and this will cause as much angst as it always has. They'll not only look different and have different kinds of trade-offs, they may even have different business and compensation models.
It's Schumpeter's process of "Creative Destruction" and renewal (https://en.wikipedia.org/wiki/Creative_destruction). The destruction part is fast, loud and shocking as what exists today combusts in flame. The creative renewal part tends to be more gradual, unfamiliar and at first may not even seem related. From time to time, even the definition of "the industry" ends up changing along with the jobs and roles. I'm pretty sure the old-school TV station engineers I learned from wouldn't consider a digital nomad working full-time editing 30 second clips on a mobile device for YouTube, Instagram and TikTok influencers as even being "in the TV business."
Due to Blender being OSS and scriptable with Python (https://docs.blender.org/api/current/info_quickstart.html), it wouldn't surprise me if a future iteration of genAI/agents works with Blender directly.
There is a Blender MCP server https://blender-mcp.com/
blender is the python of 3d animation. Its the second best tool for everything. Which btw is not a digg. its really hard to be that level of reliable accross an entire workflow.
Maya hasn't changed much in a decade, and that is a wise choice for training/documentation. They focused on the content part which is its value proposition.
Blender has always had the perpetual Beta problem, as many boring core design issues were never really given priority. Fine for developers, but a liability in commercial settings.
Houdini is interesting, but with Blender Geometry-nodes now working it is unclear how another proprietary ecosystem will improve workflows. =3
The Entagma channel covers a lot of Houdini and Blender bleeding-edge features with short lab tutorials:
https://www.youtube.com/@Entagma/videos
I've only dabbled with Blender but from what I saw of geometry nodes it's quite a long way away from competing in the same space as Houdini. Houdini's biggest single feature and the thing that allows film studios to use it at complexity and scale is the HDA system and there's no real competition for that.
I think the main difference between OSS like Blender and the competition is it seems like OSS only gets better. Each version is a little bit better, so projects that were once not competitive catch up. Blender is an obvious example, but also look at Krita, or the entire KDE project. These pieces of software age like fine wine: they get more features, they get faster, and they get more stable.
Closed-source software seems to get... stuck. In the best case. Often, they regress: becoming buggier from version to version with less features. I think of Windows and the entire Microsoft suite of applications.
I think one exception is Gnome. Gnome loves removing features more than they love not implementing popular Wayland protocols.
With OSS, unless every add-on is part of the build tree or package repo testing... it quickly becomes broken as the ecosystem constantly evolves.
i.e. of the $3k of community Blender plugins/add-ons we evaluated last year, only around 60% are still functional in stable blender releases. Additionally, many built-in core features like Fracture became broken in 4.x due to API permutations, and getting split into its own module.
In a production environment one must version lock Blender for the project. =3
IMO it's always a good idea to vendor software. But yes, OSS typically moves fast. It's a tradeoff.
Blender is definitely getting better compared to things like Maya but I don't think this argument holds for Houdini & Sidefx.
Blender has solved several key challenges, but the learning curve is steep given many tutorials are version specific.
I usually recommend these courses to users when the under $20 sale is active.
They cover a lot of Blenders non-intuitive workflows :3
"Complete Blender Creator: Learn 3D Modelling for Beginners"
https://www.udemy.com/course/blendertutorial/
* Basics of low-poly design in blender
"Blender Animation & Rigging: Bring Your Creations To Life"
https://www.udemy.com/course/blender-animation-rigging/
* Some more practice rigging
"The Ultimate Blender 3D Sculpting Course"
https://www.udemy.com/course/blender-3d-sculpting-course/
* Sculpting, Retopology, and VDM brushes
* basic anatomy
"The Ultimate Blender 3D Simulations, Physics & Particles"
https://www.udemy.com/course/blender-simulations-physics-par...
* Shader/Texture basics
* Geometry node basics
* Boid sprites
* Hair and physics simulation
* Camera FX, and post-render filters
* Instructions on how to export your assets to Unity 3D and Unreal game engines
What is HDA?
Houdini is entirely a node-based system and HDAs are just your custom nodes (on contrary to the built-in ones).
At the end of the day, an HDA is just a 'function' you define like what you do in a programming language. A function can call other functions etc. It might sound nothing special if you're a programmer, but Houdini is the only generic DCC that is built around this idea, making it more like a framework than an app.
Probably means a Houdini Digital Asset which is similar to a maya reference.
At one time there was a Houdini-Engine Open Mesh Effect plugin, but no idea if that project survived.
Cheers =3
Geometry node is not Houdini (yet) though. The main reason people use Houdini is it's easy to transfer data between modeling (SOP) and physical simulation (DOP) contexts. And they're working on integrating rigging/animation too.
Geometry node is quite a separate thing from what Blender already has. Last time I checked one couldn't even create vertex groups in geometry node, and there was no way to create an armature there either. (Not sure if it changed since I checked though)
I think you should've be able to access vertex groups from the initial version of geometry nodes. Certainly, at this point, you can dynamically access (and set) arbitrary vertex groups, including loading the names of the vertex groups you want to access from a csv file, or otherwise synthesizing the names using string operations.
its actually funny, the first version of geometry nodes was actually more powerful in that way. it didnt provide many built in nodes, but the nodes it did provide were arbitrary property setting/getting. the current and newer geometry nodes system based on "fields" is way easier to actually work with but they don't give you raw access to properties. so like in the first version of geometry nodes you could write custom mesh normals just by setting the "normal" attribute with some vectors, but then from blender 3.0 to 4.4 you could no longer set custom normals in geometry nodes and they only added a set normals node in 4.5
Indeed, but it is non-obvious how to handle it...
I found the Entagma lab videos very helpful in understanding how to parse geometry with nodes. It is not obvious (unless you already get CLI pipes), but so worth it when things finally work... =3
Indeed, Geometry nodes is new, and I have also found documentation on many operations sparse. That being said, it has proven itself very capable.
The mini labs on Entagma's channel were very helpful for parametric surface operations =3
https://www.youtube.com/@Entagma/videos
That's often a problem with Open Source. Lots of people steering the ship, and it's hard for any de facto captain to really demand a direction. If they do, they may not have the contributors needed to really see it through.
This can be solved with paid contributors, but FOSS organization don't have the most funding out there. So it can be challenging when trying to attract specialized talent.
Donated to most projects we find useful, and also tried a few Blender paid plugins.
The other problem is people Cowing anyone that may not agree with their personal opinions how projects should mature. QED: Our karma scores... lol =3
These plugins made Blender usable for a few projects, and I have personally found value in supporting for asset creation:
https://tinynocky.gumroad.com/l/tinyeye
https://sanctus.gumroad.com/l/SLibrary
https://polyhaven.com/plugins/blender
https://artell.gumroad.com/l/auto-rig-pro
https://fbra.gumroad.com/l/Faceit (for Unreal face mocap app with iPhone Pro Lidar)
> Assuming genAI won't deliver and make that world disappear entirely, ofc.
Even if genAI were to kill the cinema/animation back-end industry, I still think 3D software would be needed to continue videogame development, and godot is far behind in modelling/animation capabilities compared to blender. Hopefully they can latch on this in the long term if it happens that AI keeps eating artists lunch.
My daughter did the Blender donut tutorial and put her 3d animation into the year end art show when she was 10. Having a great community and resources behind the project makes if very easy for students to pick up.
> *: Assuming genAI won't deliver and make that world disappear entirely, ofc.
The current state of things leaves a lot to be desired. I have yet to see a model, or technique that actually tackles real 3D content creation problems, aside from animation of course.
Most solutions are trying to solve things completely orthogonal to what you really need in the real world. They produce bad meshes, if meshes at all, have bad topology or inconsistent point sampling, and are extremely prone to parroting their training data.
It's kind of incredible how far we have come but seriously, to understand a problem space you need to try to use the existing tools. Only then, do new, and useful ideas pop up that solve real issues. But man, nobody in the space seems to have a clue.
Believe me, 3D modelling software is REALLY good at what it does, it is really hard to try and dethrone human and machine working together in these really well thought out pieces of software.
Custom mesh normals is huge, because this used to be a PITA.
Tugged away at the end is defining custom cameras through OSL, which is a very very interesting feature, because it allows more physically accurate lens emulation, with dramatically more interesting bokeh than the standard perspective camera model: https://www.reddit.com/r/blender/comments/1kehtse/new_camera...
yeah man i was faking it by creating custom materials and setting vertex attributes containing the distance to a specific other object, then in my material using the attribute and the other objects info to blend between the normals at the material level
this is way nicer
Blender 4.5 LTS Release Notes:
https://developer.blender.org/docs/release_notes/4.5/
A wonderful app that, I find, benefits from LLMs to help you figure out how to use it. That or you more or less have to dive in for months and months. The people on YouTube that make Blender look so easy … Blender is really all they do. ;-)
At some point I see an LLM more or less integrated into the UI.
At some point I see whole apps written so complexly that an LLM is the required interface.
> That or you more or less have to dive in for months and months.
Waaaaay back in the Quake 3 days I had a pirated copy of 3D Studio Max and decided to try and make a Quake (or maybe it was Half-Life) character model. Found an on-line tutorial that step by step showed you how to setup Max with a background image that you "traced" over. So I grabbed the images of front, back and side views of a human, loaded them into the various view ports, then drew a rectangle from which you extrude a head, arms and legs. Then you manipulate the block headed human mesh to fit the human background image - extruding more detail from the mesh as you go along. In one day I had a VERY crude model of a person. I also found out I dont like 3D modelling. Though I'd say a person who really enjoys it would pick it up in a week or two with good tutorials.
LLM's just cut out the learning part leaving you helpless to create anything beyond prompts. I am still very mixed on LLM's but due to greed and profit driven momentum, they will eat the world.
One problem I find is that a lot educational content has moved into YouTube and videos (monetization be damned). I have no time to watch 10mins of rambling and ads for a quick tip, LLMs are great at distilling the info. Otherwise, I agree, deep knowledge building only happens through doing stuff…
Oh boy. Giving Blender instructions in English, on every level of detail, could be quite a workflow. Like, have it generate a range of chairs, pick the mesh you like, then start asking it to tweak parts of it until you are satisfied. Add finishing touches manually.
In general, model re-topology almost always requires a manual editing process post sculpting and texturing. There have been countless attempts to automate the process, but it usually requires working through an obscure book about rigging with secondary transforms etc.
Could check out https://extensions.blender.org/add-ons/mpfb/ for low-poly character mesh generation tools. =3
Why is this being down voted?
If you are pushing vertices around to make a model, you eventually have to think of topology issues if you ever want to make a good animation.
And then when you have a good topology... Then You need to rework the UV Maps, Normals and other such data so that it's all consistent.
LLMs are not the right tool for this. I'm sure some AI retopology tool can come around but there is a HUGE difference between a model designed for stills, a model designed for animation and finally, a model designed for close up facial animations.
The topology differences alone are insane.
---------
Now maybe in the future an Ubermesh (like the Uber shader / principled shader) can be made to make these processes consistent. And then tooling can be made over the hypothetical Ubermesh. But today it's a lot of manual retopology that requires deep thinking about animation details.
Ex: an anime style mouth doesn't usually move the jawline (Think Sonics mouth: its basically a moving hole that jumps around his face). Meanwhile, western 3D animation (Baldurs Gate and the like) obviously have fully made lip sync including phhfff, eeee and ahhh sounds.
The 'Kind' of animation you want leads to different mesh topologies. It's just the nature of this work.
And people get ANGRY when you ex: animate Sonic with traditional teeth and jawline (see 'Ugly Sonic'). You can't just willy nilly pick one style randomly. You need purposeful decisions.
--------
I guess if you are a pure sculptor / stills then you don't have to worry about this. But many 3d projects are animated, if not most of them. The retopology steps are well known.
LowPoly look will almost certainly require manual effort (these topology problems tend to disappear with more triangles).
Blender is often used to project the high-granularity sculpted detail onto a low-poly (rigged and clean topo) Normal map. i.e. you can get great looking assets that won't hit your poly budget as hard. When it works its great...
People have feelings, and sometimes may misinterpret others intent. I am happy we've met if we both learned something. =3
Oh yeah, I was agreeing with you. Just providing context for others if they didn't understand what was going on.
Huh, TIL https://github.com/ahujasid/blender-mcp - do you use this?
I tried it and it's really not very good. It "works" but models can't seem to do anything complicated. Claude 4 Opus btw
Thanks for pointing that out.
I have not tried it. Instead I have been asking Claude (etc.) "How do I create a repeating triangular truss..." or what-have-you. And then I follow the steps they list.
I don't know if it supports HDR on MacOS but, AFAIK, it doesn't on windows and in Linux it is only supported with Wayland.
Though I don't like the Wayland x X11 flamewar, I'm happy to see some modern features are only supported on Wayland.
That may please the crowd that will be able to say "sorry, I can't use x11 because it doesn't support a feature I need" bringing some symmetry to the discussion.
Edit: correction: it is about the development 5.0 version: https://devtalk.blender.org/t/vulkan-wayland-hdr-support/412...
> I'm happy to see some modern features are only supported on Wayland.
Why?
If anything that's a reason to why I wouldn't fully jump to blender.
I have been working on my own hobby game engine for the past 15 years and have been excited to introduce Blender to the workflow. If this is the case I don't like it. Wayland has never work for me the same way as X has.
What else are you going to use on Linux?
If they spent 15 years on the engine as is, what's another few more years rolling a proprietary modeling system?
On a serious note, I do wonder if this Wayland only limitation is something a fork could work around.
I don't think there's an X11 HDR standard, one would need to be created and implemented.
Nuke, Maya?
If the starting point is that Wayland is missing features that X has, the good outcome is not getting to a point where neither option has all the features, the good outcome is that either one has all the features.
That's at the cost of lots of duplicated work by the already sparse number of people capable of implementing a graphics server.
There's also a third option where Wayland is foundational and the X11 network protocol is implemented on top of that for people who need it. Why should a network GUI service implement a driver to talk to a specific model of video card?
> That's at the cost of lots of duplicated work by the already sparse number of people capable of implementing a graphics server.
Yeah, it kinda sucks but this is where we are.
> There's also a third option where Wayland is foundational and the X11 network protocol is implemented on top of that for people who need it. Why should a network GUI service implement a driver to talk to a specific model of video card?
Agreed; I have long argued that it would have been far better to transition to everything on the same backend with effectively rootful XWayland being the only (bare) X server, and then after that try to deal with the rest of the stack (if you really must). And maybe in 2026 we'll finally start to see movement in that direction with https://gitlab.freedesktop.org/wayback/wayback
It definitely does on macOS and I think also on Windows. You have to set the color management for the viewport to Display P3. In older versions this precluded you from using AGX or Filmic, but I think you can actually use AGX with Display P3 now.
I really wish to could use Wayland, but there is too much problems or bugs related to the software I use for work and also play. I will test it again with this new version Blender (that was one of that software with problems).
In general, most repositories use old stable versions of Blender. And often folks are reduced to using snap to maintain version specific Compatibility with add-ons etc.
Also, getting the most out of Intel+RTX CUDA card render machines sometimes means booting into windows for proper driver support. Sad but true... =3
The reality is most commercial software and users are on Windows machines. It is fundamentally a Blender interoperability, and 3rd party platform license compatibility issue. We all wish it wasn't so, as many artists find the Windows file systems and color-calibration concepts bewildering.
Making a feature platform specific to a negligible fraction of the users is inefficient, as many applications will never be ported to Linux platforms.
Blender should be looking at its add-on ecosystem data, and evaluate where users find value. It is a very versatile program, but probably should be focused on media and game workflow interfaces rather than gimmicks.
Best of luck =3
I agree with you, but I think this limitation is for much simpler reasons, like "the contributor only knew how to make this feature in Linux, and only in Wayland". cross compatibility for stuff a base as color grading can be a thorny issue.
If nothing else, it's better to have some implementation to reference for future platforms than none.
We've all seen too many plugins become version specific or indeterminately broken
https://www.youtube.com/watch?v=WFZUB1eJb34
Someone needs to write a Blender color calibration solution next vacation =3
The significant majority of the film and animation industry uses Linux.
Linux RTX CUDA drivers are getting better, but really depends on the use-case. For a Flamenco render farm it makes sense for sure.
Creatives on wacom tablets and Adobe products etc. will exclude the Linux Desktop option. =3
Not just for the farm, the large majority of the movie and tv vfx and animation you see is done by artists using Linux workstations.
Not the artists I meet, they love their wacom tablets and pressure responsive painting programs... i.e. most of the other software is windows only.
I like Linux (use it everyday), but many CAD, Media, Animation, and Mocap application vendors simply don't publish ports outside Windows.
Most studios have proprietary workflows with custom software. =3
The applications I agree, Wacom tablets though have great driver support on Linux (in my experience more stable than Windows).
Indeed but this is a discussion about Blender and you posted originally:
> Making a feature platform specific to a negligible fraction of the users is inefficient, as many applications will never be ported to Linux platforms.
All the large studios use Linux, that's why all the third party software that is used in feature animation and vfx is supported on Linux. So I'm just saying 'negligible fraction of users' in the case of Blender (which as a project would like to increase adoption in professional feature animation and vfx) isn't really true.
I am sure Studios account for a small portion of the 4.5 million unique downloads each release. Note that less the 20% of users ever touch film or animation projects, 73% are single users, and most related user applications are Adobe products.
Stats are available from the published 2024 feedback data:
https://survey.blender.org/feedback/2024/
Best of luck, =3
Recommended reading:
https://www.poetry.com/poem/101535/the-blind-men-and-the-ele...
I'm not sure download stats are hugely relevant because that would imply the needs of every person that downloads Blender are weighted equivalently which would make little sense.
Or are you suggesting the Blender foundation has no interest in getting wider adoption among film and animation studios?
I think the foundation projects hold a lot of potential, but what they release as "stable" is rarely ready for a production setting. People do use Blender for small side tasks commercially, but would you honestly bet your company reputation/job on their 31 years of shenanigans?
Updates still permute the core to break parts of the program, and brick countless add-ons or custom code. People turn users into Beta testers, partners into IT support, and hide workflow details under layers of feature-creep kludges.
When Blender updates for feature X, they will usually brick feature Y. YMMV
The Foundation may intend to improve user adoption, but they can't even cover there own unit-tests on internal add-on code. =3
If only Blender had an unreal engine tier real time capability!
Blender used to have a game engine. Which version was it removed?
Just use unreal ?
[dead]
[dead]
For all their talk about performance, that webpage is incredibly slow/stuttery.
Luckily Blender is not a webpage
imagine blender as electron app... ^^