So IMHO it's one of the best executed desktop programs while also being in a massively difficult field, and they deserve the praise. In many places there are outliers like this; Stripe for Documentation, Wikipedia for encyclopedic knowledge, Google for search, etc.
The two main criticisms that I've found is that first the interface is very confusing (but this has been improved constantly in the last 5-10 years!). As a personal opinion, I believe this video was a big inflection point (Blender seemed stubborn before the video, but this video at the very least softened the opinion of Blender): https://www.youtube.com/watch?v=UWacQrEcMHk
The second one is that it's not "industry standard", but as more and more people are using it for hobbies and indie projects, this will def change eventually, so I'm not too worried about that as well.
If you want a critical take on Blender therefore, for one you ought to look outside a community of business-minded devs. Look where artists hang out and you'll find a less rational take that is based on individual likes and dislikes. Where logic doesn't matter so much--it's how you feel. That's where you'll get into it. And if you look at other FOSS graphics software communities where the project lead is someone who just does whatever they want, even better, you'll find a group of users who either adapt or leave, and that's quite different from Blender's psychological development ecosystem in a lot of ways.
Alternately you could go more-business and go ask a SolidWorks user group to tell you how great Blender is. They will probably tell you how business is done properly and why the Blender way is clueless in various ways.
It so happens I ran an online community for another FOSS 3D package for over a decade. We got some people who were anti-Blender, and some who were pro-Blender. Often the anti-position was based in prefers-underdog reasons, which I understand. Even in FOSS there are some clear underdogs and people get upset when they like the underdog and continually see another project winning attention instead.
We also got some people who were pro-Blender, who thought that our community must be anti-Blender, because we were obviously "competitors in the space." In fact there was never a competitive mindset, it's more that people who lean into that business psychology of "here's how you win" think more often in terms of competition.
BTW Blender is definitely an awesome project, I've used it since 2002 and on various commercial projects in addition to hobby stuff.
That sounds more like FLOSS done right than most "doin' business" available these days. If most "doin' business" worked as you describe, FLOSS wouldn't have eat so much market from them.
It'd be nice if they made some sort of "stable" API that plugins can rely on in the long term, even if they had another one that is "better" (whatever their current definition of "better" might be).
Blender is indeed great but Maya is still extremely popular and many professionals happen to prefer Maya for one reason or another. I don't often see that fine grained of a discussion of media tools here.
My bet is Blender will become more standard in the next 10 years. It’s becoming big in gaming (I’ve heard)
They also break the API accidentally frequently. My favourite is that they have accidentally inverted the direction of rotation of the rotation operator 4 times since 2.8 and randomly removed one of the modes that the UV Smart project operator worked in.
The documentation is also lacking for the API and none existent for the main code base.
Despite this Blender is amazing but I do wish they were more professional. From a developer point of view the two things that would make life vastly better is if they
a) hired a technical writer to produce decent docs
b) stopped breaking the API every 5 minutes and when they did documented it properly (and in advance)
So, not such a bad decision. Building with MSVC used to be a bad decision, less so now.
Having to have divergent code paths per version is hell.
Blender has a Python API. Which is arguably one of the easiest languages to keep a stable API in, since you're not dealing with an ABI.
Yet that breaks regularly.
I'm okay if I had to recompile for every version of maya I support. I just don't want to have to have to maintain tons of ifdefs per version
The technology & the creative development are part of an intertwined process.
Also its competitors are buggy-ass insanely overpriced monthly subscription proprietary editors, so no surprise there.
There are still things to complain about, like the undo speed, CSG modifier unreliability, lack of certain features, but none of them are strict deal breakers.
But it is minor... and i can still use 2.79 anyway :-P