I don’t have a dog in this fight, but I’ve been screwing around with Suno for almost a month now. It’s fun, but the more you mess with it the more artificial it starts to sound. And the more you extend and build from one project, the worse it gets. Case in point, listen to how the guitars in these two songs just turn into a monotone ringing noise by the halfway point.
Ok so not talking about ethics, just putting that to the side… on just a listenability standpoint, do you think there’ll be that big of a change?
(not arguing @ u w/ hostility! I just think it’s an interesting angle you’ve mentioned that i hadn’t thought about before)
considering a band like Poidol, who i got no proof but come on, no one’s gonna say they’re NOT outsourcing their writing to whomever wrote for Pentagon & Arlequin…
I don’t usually imagine ghostwriting is absolute - unless the members aren’t playing the instruments live, i’d imagine there’s got to be some sort of accounting for playing styles at least. and at most a complete rewriting of each guy’s parts to individualize but stay within the shape of each song.
If what i’m imagining is true - i can’t guess AI replacing the ghostwriter here would be all that perceptible to us. It might even have happened already, that wouldn’t surprise me. I could easily see a future in which case it’s time for the next B-sides so the manager guy writes some adjectives on what style seems most profitable for his guys next and out comes a cohesive guitar/bass/drums/vocal melody tab. And that gets handed to the band and producer to work out to fit each guy playing.
But THAT, i’d imagine, is just about the current ghostwriter situation, i’d imagine. Only with a person instead of a program. So i could see a shift to this that wouldn’t tank quality, but instead be imperceptible to us unless we read interviews or hear forum chatter.
And to be real, i’m not listening much to bands like Poidol expecting novelty and songwriting quality. That “exact midpoint of vkei sound” is just nice background music and for me, doesn’t need to be any more than that.
Ofc all this is keeping ethics of this supposed practice aside - which i think would be a terrible, dystopian, anti-worker, anti-musician way of getting music started.
I just imagine it might start to happen a lot more subtly than say, tomorrow a new air music group like Golden Bomber showing up to pretend to perform an audio file some machine learning software generated.
Definitely,
I talk here too much so have this now as a visual representation of my explanation.
you can make an image of this kind in around 20 minutes in photoshop, having access to stock photo bank and some grunge textures (which are often available as a free download, and they’re less easy to clock than AI art.)
the licensing rules for stock images are a bit different for commercial use, but they’re on a legacy indie label that can afford nice production. it’s disappointing a.f.
if it gets recognized as art by the copyright law it will solidify the stigma against CG illustration which has always been there.
I’m not going to lie I would want to see one of yunger and more impressionable VK acts throw a tantrum similar to this one, tho:
In today’s episode of “Top Tier Usage of AI”, I bring you this:
I still don’t know what bug it is. Now I have two.
Messing around with visual kei prompts in Luma Dream Machine. Mainly just nightmare-fuel that reminds me of 2~3 y.o image generation ![]()
I’ve found most text-promts to conflate visual kei and kpop aesthetics, since the volume of kpop images dwarfs vk ones.
Yes, I tried to “Will Smith” Kyo, and it’s terrifying.
…
![]()
Posts like these make me wish this forum had reactions like MH did…
The Maria one will give me nightmares for the next whole year.
those are 3D backdrops, likely rendered in unreal engine - there’s dynamic lighting and particle FX which are part of background.
their PV studio is linked in the video description, and for ¥1,000,000 + costume, makeup and other expenses and a month turnaround time that sure is not AI
(I stan all of their recent videos done by that studio, needless to say… the snow one is an eye candy.)
You really don’t think it’s AI imagery? In especially the Shita MV it has that really typical AI render feeling.
I would say my answer depends on how it was used, which is something I don’t have enough information on. AI is just a tool, a tool that I think artists are in the best position to take advantage of. As long as AI was used to assist an artist, instead of outright replacing them, then I have no strong objections to it.
In this particular case, AI would be equivalent to CGI. If it’s good and convincing enough, I shouldn’t even notice. But I looked at a few seconds of it, and I could tell that something is not quite right, so we still have a ways to go.
I think Kaya used AI for the backdrops of Villain, and that MV is gorgeous and one of the best of last year.
it’s pretty difficult to say what is & isn’t AI! i mean, to draw one line and say on that side is, on that side isn’t. Like at what number of grains does a heap of rice become a pile, a mound?
I’ve been using photoshop since i was like 10; there’s a tool you can use where you click a thing like a speck of dirt obscuring the sky, and it’ll look around the image to determine how best to replace the area you click on. If your speck was surrounded by lots of sky-coloured blue, it’ll fill it in with some sky-coloured blue. is that AI?
Photoshop’s gotten a lot more complicated since then - two years ago, say i click on a speck of dust on a model’s argyle-patterned shirt; decently accurately it’ll read the pattern and the way the fabric curves around the body, and it’ll fill in the selected area with newly rendered pixels that (try their best to) agree with both these. is that AI?
Very recently, now i can take a picture of my studio, lasso an area, and say “there is a jaguar there” and then Photoshop will just shit out a whole-ass jaguar, matching lighting/shadow and all, wholecloth. Well, that’s certainly ‘AI’ lol
people talk about the “blu-ray” effect - that unappealing ‘hyperreal’ look to videos on modern TVs ca 2010. It actually had nothing to do w/ blu-rays or an increase in quality beyond what our eyes are used to - it’s this thing called ‘motion smoothing’ or a bunch of other differing buzzwords by manufacturers, but i call it frame interpolation. TV manufacturers somehow got it in their heads that people would like it (???) if all their movies and shows shot in 24 or 30 fps were displayed instead in 60, but there’s literally not that many frames in the source data to choose from. So the interpolation is the monitor drawing in real-time a new frame between every frame (or a few for 24fps), so a 30fps source would have one recorded frame, one generated, one recorded, generated, and so on. This gives that blu-ray effect, bc the motion blur doesn’t sync up to what our eyes expect from that speed of movement.
Well, that’s software / algorithmically generated content - literally half the movie is generated, and that’s been going on since the 2000s. AI upscaling but for time instead of size.
Anyway plz don’t call me a tech bro or AI apologist, i’m literally a musician & photographer who’d like to make money to live lol
but like my last reply - i think going hard on what we call AI (a tool / technique / style) is a bit of a distraction from a more fundamental concept of: is the studio / commissioner paying people to do work, or getting work done for free and withholding what should’ve been wages?
in that framework: i can’t at all mind a studio hiring a VFX team to make CGI whatever, unless it’s replacing what would’ve otherwise been practical effects by a more expensive IRL team.
there’s a making of showing how those kinds of music videos are done these days, it’s basically an unreal engine scene projected on a white backdrop + extra props and stuff if there’s budget for that
is there AI stuff there?
I mean some of that might be generated with a material generator.
I think they have an environment artist building those scenes from either stock assets, or making them from scratch - it’s not particularly advanced architecture; the ice sculpture in their new video looks much more interesting, but it’s 3D - if you look closely, there’s stuff like light doing reflections on it.
in case with shita, the background swirls look like acrylic pour art - literally just video clip art + the classic VK Ink Blot stock video overlay + liquid particle effects that have the acrylic effect mapped on them.
AI art tends to overdo particle effects because they always carry a certain wow factor, and prompters are generally not an inventive bunch - they either go for OTT in your face CG style with every condiment that they stole from artstation, or very photorealistic, shying away from generated fakeness, with barely any middle ground.
I had no idea the attitude was actually encouraged at the very top (AI is still not legally defined as fair use, despite openAI etc hiring lobbyist teams to talk their way out of copyright laws with the govmnt in the US) and trickled down onto the common internet prompt dunce crowd, but here we go:
I really hope miss thing never finds an aesthetician injecting her face and shaving off that nose who actually trained in art theory and aesthetic human anatomy because girl you’re getting bold in all figurative and literal senses.
I’d like for this Mira Murati jargon juggler to show us anything in the world that she’s ever created, and maybe she’ll take the hint and go away.
I think this is one of the best things ever said about AI:

I think it is okay if it is used as a tool to help with creative expression, but not to create whole pictures or songs just with prompts. And honestly if people make music only with AI, what’s the point of this music other than making money? Music and other art should convey feelings from the artist to the listener/viewer. At least that’s what I want from art. From music especially.
It’s pretty sad that Kyo of all people seems to like it. I bet he would never use it for his lyrics but I guess he has a diffrent point of view when it comes to music videos.
Totally agree with the tweet you shared. I wish the developers of these AI programs would focus more on AI being used to do stuff like laundry and cleaning rather than making artwork or writing stories. I’d rather do the creative stuff and leave the boring tasks to the AI, not the other way around.
The argument that it’s fair use is just an obvious one. I had never seen anyone else say it before I did.
And to @zeus (assuming you’re a mod) are we allowed to call people retards here?
I would advise against it



