Yeah, TFA's point is that the basic/inexpensive camera in the hands of an unskilled user can be higher quality than an equivalent iPhone camera shot. In my opinion from the example shots used this is definitely the case. Camera phone distortion is pretty bad (you have to stand back further from your subject offset this, use a higher res setting, and crop in) and the processing has gotten out of hand in recent years to the point where it starts making photos look worse and worse.
I have a Blender account which I was able to use to log in to studio.blender. It still shows the banner saying I need to subscribe to view the content.
I don't use Blender (because i don't do 3d modelling even for fun) but that statement makes me think of "it's important not to conflate the Mozilla foundation's harebrained initiatives with Firefox".
That's how the blender open movie projects have been organized for a long time. If enshittifaction was coming it would have happened 10 years ago. In few of the earliest ones it was a bit different with more of the files being publicly downloadable. But even then they where pre-selling DVDs containing best quality videos, commentary from artists, tutorials and all the production assets. Almost 20 years ago when first open movies were released internet speed was orders of magnitude slower. So while you could in theory download various production files it was less practical. Blender studio subscription (previously called Blender cloud) has been a thing since ~2014, long before the recent increase in popularity and sponsors.
I am talking more about the the context of open movies, but the few game projects are done in similar manner. Although software licensing makes things a bit more messier.
Some people pay money to fund a team of professional artists and maybe even 1-2 software developers to work together with blender developers on the open movie project in return people who paid get access to all the production files and high quality training material. In the mean time everyone else still benefits from the new features and other software improvements made during the production of movie.
One of the big problems with many open source software is not enough dog-fooding, insufficient user testing and involvement of professional users for the final software. Most of the developers are programmers not professional artists, and most of the artists are not programmers making it hard directly contribute or even communicate the feedback in a way that's actionable. Many of the professional users also don't want to waste their time with half finished open source software resulting in chicken and egg problem. Blender open movie projects solve those problems.
Providing paid training materials while getting user studies on large size projects in the process of making them seems like one of best ways for more sustainable open source with less conflicts of interest compared to what most open core software does.
It's not like the the Blender foundation is diverting money from developers towards projects no one asked. People are getting exactly what they are paying for. Based on 2023 reports blender foundation gets 2-2.5 million € in yearly donations, out of which ~70% goes directly towards developer salaries, 10% other salaries and only 3% (72000) is labelled as "support studio for testing" in the previous years explaining that it's the money going towards "Blender studio" for specific work. In the mean time Blender Studio has 6500 monthly subscribers (~0.9m € yearly).
It's going to depend heavily on what you're doing. If you're doing common tasks in popular languages, and not using cutting edge library features, the tools are pretty good at automating a large amount of the code production. Just make sure the context/instruction file (i.e. claude.md) and codebase are set up to properly constrict the bot and you can get away with a lot.
If you're not doing tasks that are statistically common in the training data however you're not going to have a great experience. That being said, very little in software is "novel" anymore so you might be surprised.
Just because it's not strictly novel doesn't mean that the LLM is outputting the right thing
We used to caution people not to copy and paste from StackOverflow without understanding the code snippets, now we have people generating "vibe code" from nothing using AI, never reading it once and pushing it to master?
AI coding tools aren't equally effective across all software domains or languages. They're going to be the "best" (relative to their own ability distribution) in the "fat middle" of software engineering where they have the most training data. Popular tasks in popular languages and popular libraries (web dev in React, for example). You're probably out of luck if your task is writing netcode for a game engine, for instance.
I am retiring from tech after 20 years and embarking on a new career which requires intensive schooling. As a high school and college drop out, I will need to learn how to study again, and want to be as efficient as possible so that I may still spend ample time with my family. Any tips greatly appreciated!
And certainly people are capable of implementing algo-based clients. Heck, have them scrape your personal web and email and whatever activity, keeping it all local.
Laser printers use 600 dpi because most readers who appreciate typographic design and sharp fonts do not consider 169 dpi to be "perfectly sharp" even with all subpixel cleartype tricks enabled. Knuth's Computer Modern font that TeX/LaTeX use by default looks terrible below about 300dpi because its sharp serifs cannot be rendered correctly. I can't stand my 227 dpi MacBook Air M1 because the fonts on my Surface Pro (267 dpi) are so much sharper, especially because Big Sur eliminated subpixel cleartype tricks so fonts are now worse on Mac screens than they were when the 2012 Retina MBP came out. But a 338 dpi 4K display on a Dell XPS 13 is a noticeable improvement over the 267 dpi Surface Pro.
Coming from a retina display 1080p is a bit fuzzy. Definitely something I would need improved in future releases. But my biggest concern is whether Linux is ready for Mac users like me who expect things to "just work". If it just works, I can deal with 1080p. If it requires a mechanic, I don't care if it's an 8k display, I'm out.
reply