Not everything is bleak. The ending sequence for Frieren season 2 was entirely hand drawn with colored pencils[0] and the new Ghost in the Shell (coming out next month) is also hand drawn[1].
Mappa is imho the current heavyweight studio, doing everything from spectacular action setpieces with bold visual experimentation in mainstream stories like Jujutsu Kaitsen to the extremely labor-intensive stylization of Dorohedoro.
Some modern romcoms like Kaguya-sama (A-1 Pictures) and The Dangers in My Heart (Shin-Ei) were nicely done too.
Things are obviously not going in the right direction, but the current accelerated fall in quality the consumer sees is more influenced by general unwillingness to spend and/or take risks from bean counting production committees.
Not true. Madhouse isn't what it once was. Frieren had lots of freelancers.
Also, Kunihiko Hamada (whom worked on Frieren, is 1 of the lead animators, designers) left, citing disappointment.
Japan has to decide if they want to invest in their animation industry or if they'd rather let it die out by killing off the local talent while training animators in other countries to one day replace them.
Often the key frames or "important" parts are done in-house with the "mechanical" or boring parts being outsourced.
It's been being used for years. It saves time, but ultimately the results aren't good enough to replace humans. I don't expect that it ever will be. AI is incapable of creativity and sometimes it'll be faster and easier for an artist to just draw what they envision than it is to tell a computer what they want, have the artless software do it badly, and then have the artist tweak and fix the result until it matches what they wanted in the first place.
I can totally see where an animator might be dissatisfied with doing this on a hand-drawn production, however, and vastly prefer humans drawing every frame of the production.
Just the staggering amount of anime produced these days blows my mind.
One could die of overwork just binge watching them.
PS: so weird to watch the trailer while working in XR. I never imagined as a kid I'd be programming in a headset but now it's banal. I even buy 2nd hand HMDs for 100 bucks. Weird times.
I clarified a bit in a longer post. I do actually like the Vision Pro... but even though I own one I barely use it. It's a good device but it's just too many hoops for what I do. I find a much affordable Quest, even older generations, with adb and a proper WebXR support (including with AR support, which Apple refuses to implement) fits better my need. I have 0 problem popping a Quest 2 in my backpack without even thinking about it, only protection being a literally piece of cardboard on the lenses. I can work with it anywhere and give demos to total stranger with no worries about them dropping it or heck even running away with it.
FWIW I'm also not in the Apple ecosystem so all the wall garden hoops just tire me. I want the device I own to let me do my work.
I find that, probably counter-intuitively, super high resolution and refresh rate matter... but only up to a point. As I mentioned in the longer post if you are "just" looking for a monitor with another form factor, sure, maybe it's OK-ish, but to me it's very uninteresting and not worth all the challenges, including social, of having an HMD strapped to your face. Only when it's properly immersive does it shine, and then resolution and refresh rate don't have to be that high to be usable.
I program in VR for VR, and I help kids (and adults) learn programming thanks to VR.
What I do NOT do is writing my bash scripts or things unrelated to VR in VR. I technically can but I just have so much tooling and habits outside of VR it wouldn't make sense.
I'll answer your question as the end, you can skip to the last paragraph if you do not need the context.
Still do program in VR I can explain a bit more, or even show with a 1min video https://video.benetou.fr/w/iuJUP5oNnRAfH2xC3AW9s4 . Basically you can see ~20s that using the left pinch I can execute code in VR. Those specific code snippets are part of triad with querySelector() to get an element, setAttribute with it's attribute name then the last the value, e.g. here querySelector('a-sky').setAttribute('color', 'black') which sets the sky color to black. The idea here is to make a small change with a big impact.
If you are not a developer, it feels pretty cool (from the feedback I received) but if you are a developer there are 3 things you can notice :
- this is all done in a browser, as you can see from the beginning of the video, despite being properly immersive. I'm relying on eval() to provide a no-install no-build programming environment
- the black plane on the right is the browser console, literally. There is some visual feedback with green for console.log() yellow for console.warn() and red for console.error()
- the black plane on the left (hard to notice, ~34s in) is the code of the page
- the code snippets can be combined
- there is some syntax highlighting in order to hint that different attributes and there values are not compatible, e.g. scale black does not make sense or color 2 2 2 is also not correct
- the modifications are saved server side (but can also be done solely on device via local storage) so that others can build on the modifications you saved
and things you can't notice
- if you plug a keyboard, bluetooth or just USB-C, you can type code directly, it will become floating text snippet
- there is page reload code snippet at your feet, if you mess up you can recover from a proper state
- there is a non immersive console via eruda that is richer and expandable in case your changes break immersive mode, you still have a way to try to recover
- you can self-host that immersive Web on the device, it's just a Web server with HTTPS, if you want work properly offline, which is good when traveling, e.g. train or planes (that's how I do it on the go)
So... yes IMHO it's a good way to program in VR, today. It's also a good way to learn programming. If you program in VR chances are you quickly get frustrated by having to put the headset, work, hit a bug then remove the headset, try again, etc. That cycle can be frustrating so trying to find ways to do the spatial things in VR makes sense. The non spatial work though, e.g. backend, that can be properly tested outside of VR, IME should not be done in the headset, even when it can be done. Of course it depends on your workflow.
Finally which is a good headset depends on what kind of programming you want to do. If it's nothing spatial and you "just" want a 2D screen ... honestly I don't think it's worth it. I did try a lot of headsets. I worked with Varjo and Quest Pro, have a Vision Pro, Lynx XR1, Quest 3, Quest 2, Quest 1, Oculus Go, Valve Index, etc. I never spend 8hrs/day in any of those. It's short sessions or while on the move. I don't expect any headset to be "good enough" compared to a basic 4K 32" monitor, ever. What I also don't expect is any monitor, no matter how fancy (and same here, I worked with Looking Glass 8K, ~15k EUR monitor) to provide any proper sense of depth supporting manipulation. So if your programming work involve 3D work... then VR headsets are amazing. If so which one to buy? Depends on your programming language. If you are into JavaScript then any of the Quest will do but the latest generation (Quest 3) has color passthrough and higher resolution. It's also Android based so you can put termux to setup NodeJS, Python, etc and serve the content as notebooks. If you are into the Apple wall garden then obviously the Vision Pro but you'll have to go through all the usual hoops to publish anything.
TL;DR: high res with WebXR support and adb (or equivalent), e.g. Quest 3 (ideally rooted but that's hard to come by)
There are few languages that make it easier, e.g. JavaScript as in my case, or any other language with a VR library, e.g. Lua LÖVR or also tooling like Godot which does have an in-VR live reloading mode (sorry discussed in French only at https://video.benetou.fr/w/g111LSRGfj6HmNxa91au3q ).
If your favorite language isn't very convenient for VR you can get a WASM build via e.g. QEMU. If that's not practical you can get the heavy load in a container then communicate to it via ssh or a Web API.
My point being that you can program in any language in VR quite conveniently and the power of the device itself is typically not relevant. For example I started some simulation for swarm robotics from the HMD to explore the result in VR but the runs themselves where not done on the device.
TL;DR: your HMD can host everything or can be a thin terminal to your heavier workload.
Do AI platform companies actually pre-train networks to do the same for hand drawn artists?
Related question: If they do train them to do that, are there any that train people for the "reverse": learn how to draw with paper and pencil by showing techniques only i.e only the "what" but not the "how" ?
It seems Western AI platform companies generally don't prefer an architecture with multimodal non-literal inputs to closely follow intents of users, over ones based on pure literal descriptions. It was some Chinese guys that first did works in that direction. There appear to be psychological resistance to the idea of non-literal forms of thoughts among Western entities, as if there's some literal-text superiority theory deep down in people's minds. Others like researchers from Chinese labs probably don't have that.
Artists' responses to generative fill-ins are lukewarm at best, if the obvious responses were put aside. AIs tend to treat artists' intentions as deviation from the mean and tend to steer image into less interesting, more noisy directions. That negates potential productivity gains.
I don't think there's any AI trained to generate ideal strokes from prompts so to teach someone, or datasets that could be used for it, esp. with current climate regarding AI image generation - the bridge between AI and artists of many kinds are burning white hot, nothing is going through there.
Arco was an entirely hand-drawn movie w/ a 93% Rotten Tomato score released this year. Couldn't even break even. I liked it a lot.
Seems bleak to me.
I think ... a lot of people are going to be disappointed in that. I like it, but some of the characters designs are ... ugly. I believe intentionally so to draw contrast. But there's just no getting around the fact that some of the characters are ugly.
If you mean the American movie, no. It's trash. The 1995 anime, definitely watch.
E.g. In the movie the protagonist glances an office worker that looks like her. The manga has her explain that she looks like a cheap generic model to hide the fact she is full of expensive specialized government tech. I always think of that scene when I put an expensive work laptop in a ratty backpack.