Amarna Forum

I'm making this thread to talk about the real human inspirations behind Anime characters. Unlike this post by Erika I'm going to focus specifically on the visual aspect, in particular the face. At least for this first post. Hopefully this discussion furthers my objective of having people see video games as something parallel with reality rather than entirely separate, as well as a push to take further interest in things Japanese artists generally find interesting. Okay, let's get started:

---------------------------------------------------------------------------------------------------------------------

River Phoenix/Squall Leonhart




This is a very easy example, it's well know, his designer said so publicly, and they look very similar. Still interesting to talk about though. First thing that comes to mind for me is how thuggish River Phoenix looked. Is it because he's Jewish? I don't know. It's not like Squall doesn't look angry.




It's probably Phoenix's nose that gives me this impression. I find it disgusting. Despite him being Jewish like I just said, his nose actually points in the reverse direction yet angled at the same intensity as you'd expect. This is a very White trait but not a good one. Something I associate with the mass of short White American males, thankfully Nomura fixes this by giving Squall a nice straight nose.



Maybe it's the weird smile. Something just feels slightly off. Squall is cooler in every way though so whatever, the issue has been resolved permanently.



---------------------------------------------------------------------------------------------------------------------

Reiko Nagase/Kei Yoshimizu



The Ridge Racer girl. "For display purposes only." Her history is a lot more interesting than just 3D human to 3D model.







At first she was a nameless sprite in a 3D world based on one of these "Race Queen"/"Grid Girl" things. Something that apparently exists in Japan because of a single advertisement depicting a European-inspired woman (Heil). You can see her in this form on the right side of this image:







Lol.





This image is so fucking cool. Hey, ! Let's hear about the Japanese idea of what the Xbox was, huh?





When Namco was bringing the series to the Playstation, Kei Yoshimizu decided to have the mascot girl from 'Rave Racer' stop being an ugly whore and to get nice and pretty instead. She got a name and started looking like all the 3D models in that collage image I linked above. She's perfect for R4 being a game about the the isolation and elevation of the finer aesthetic components one can find in 'racing fast cars'.





But here's the part that relates to the point of this thread: When designing Reiko, Kei Yoshimizu didn't have the ability to use a pretty Japanese model in creating her. So... he shaved, plucked his eyebrows, and used his own face as reference.





(Comparing these images she kind of looks more like someone else on the team...)


Anyway, she's basically his daughter. Awesome.


(Source)

Some more CG by him:










Unrelated Ridge Racer R4 fan art:





---------------------------------------------------------------------------------------------------------------------

Kaori Kawamura/???



Half Russian half Japanese Jpop/Jrock singer. Apparently she broke both of her arms to avoid going to school and this got her into the music industry somehow, but I can't find a source for these claims. This picture already looks like a Final Fantasy character but it's another picture that specifically caught my eye.



I'm absolutely sure that I've seen an Anime character that looks exactly like this doing this exact pose. My first thought was Yuna from FF10. Same hair, nearly the same face and skull, both are not exactly Japanese but still look like a very specific kind of "pretty Japanese girl" (though Yuna is obviously far prettier). Problem is Yuna would never do that pose or face. Yuffie from FF7 also came to mind. She looks kind of similar and shows her teeth unlike Yuna, but she also doesn't do that pose or winking thing, and the hair covering the entire forehead puts her in a completely different class of anime girl than the one Yuna is in.



Back to Yuna, they actually changed her face a lot between FF10 and FF10-2.



Her hair is more like the image of Kaori Kawamura in the 2nd. Makes sense because they started creation on 10-2 by turning her into a Jpop idol both in what she wears and physically.



They ended up remaking FF10 and 10-2 for the PS3, which uses significantly worse models than the original strangely. It also looks fucked up in a countless number of other ways which doesn't even make sense, it's not like they outsourced it to China.



Yeah, Yuna looks terrible now and so does everything else. The careful lighting is gone, the soft fuzzy PS2 anime glow everything was built around is not replicated or compensated for. But the worst thing here takes up a very small percentage of the screen despite is importance. The eyes. Terrible thing to mess up considering anime makes them huge a lot of the time to reflect their expressiveness (importance). In real life your eyes have a lot of moisture on them and are pressing firmly against your eyelids. Maybe this is some kind of delusional theory (I'm confident in it though) and it might be kind of hard to see in these screenshots but to me it looks like the distance between the eye and eyelid is slightly exaggerated by Final Fantasy 10. Maybe they were excited about the technology allowing them to model eyes, regardless the result is quite extreme given how expressive eyes are even if you can barely see the extra distance.



I think it's more noticeable here. The exaggeration combined with the soft edges on everything creates a great result. Meanwhile the remaster makes everyone's eyelids cast these extremely harsh dark shadows. The distance between eye and eyelid might be exactly the same in the PS3 version but it wouldn't matter given the changes. The expressive three dimensionality is lost. Yet it somehow looks less like 2D drawing than the original. Neither the advantages of 2D or 3D are being used effectively here.





Compare this to a similar image from Final Fantasy 12. Basically no gap at all between the eyelids and the eye, less 3-Dimensional than reality even. But it's taking advantage of the 2-Dimensionality, they're painted on doll eyes. I like that they tried both ends of the spectrum. Later efforts at 3D anime people seem to rest somewhere between the two:










FF10-2 actually has a fully Japanified Yuna character who I find really weird looking. Maybe instead of saying "fully" Japanified I should say "non ideally" Japanified. When Japanese artists make someone "more Asian" according to them, it turns Sylvie Testud into Ashe from FF12.





"More Asian" maybe, but not exactly. She's only more Asian in Anime's definition of that word, which carries nothing negative along with it (as it should be).

"Her hairstyle is based on photos of Avril Lavigne taken by paparazzi."
https://web.archive.org/web/201904020146...characters

You guys ever see this page on hardcoregaming101? This whole series of pages is interesting, but this one has great lined up galleries full of examples. Not all of which are Japanese.

Most amusing examples to me are Hideo Kojima using western actors likenesses way, way before Metal Gear Solid.



Since Metal Gear is such a pastiche of cool Western things I find this funny and appropriate. But the re-release redrew everybody into original faces, lining up with the later games where possible. Lame. Kojima drawing more directly from Hollywood and realism is rather essential to the formation of the unique character of Metal Gear. Through this we can also see specifically what he was into. Big Boss being James Bond is very, very cool.

Anyway, in MGS1 everyone is a low-poly hentai protagonist shadow-face in-game while 2D depictions are Yoji Shinkawa's illustrations. Which are interestingly halfway between Western and Japanese concept art, rather than explicitly being manga or anime styled. My point in raising this is that there's basically no canon realist depiction of Solid Snake's face until Metal Gear Solid 2, so they're basically free to create one from scratch. And as far as I can tell that's what they did. He's not a scan of a model. He's a constructed 3D man.



Our own Pigsaw has raised in discussions before how the Japanese perhaps have an illustrators eye rather than a sculptors when it comes to 3D modelling, perhaps informing their general style of results and sensibilities as creators.



Snake has clear visual inspirations. And clear thematic ones. Kojima's not stupid. He fawns over Hollywood to some extent but understands it better than most. With Metal Gear Solid he correctly read the intentions and natural trajectory behind John Carpenter's character Snake Pliskin. Rather notably, after Metal Gear Solid 1 came out Carpenter made a sequel to Escape From New York in which Pliskin basically acts like a Metal Gear character and reaches Metal Gear like conclusions about humanity. Naturally this film filtered absolutely everyone. My point being that Kojima can get what these guys are about rather than just dumb copying by rote. And that's very much how we get the cast of Metal Gear. The looks of the different members of the Snake family are all thematically informed variations on Snake Pliskin. Their father, Big Boss, looks the most like Pliskin. He's a similar character, a rough human outlier who goes his own way. Solidus is an older, clipped and professional infiltrator of the civilian world. Liquid is an aryanised globe-threatening terrorist, Pliskin elevated to a heroic status. While sad silly Solid Snake is Pliskin beaten down into a helplessly loyal shaggy dog. MGS1 opens with him cutting his hair down short to serve the government again (before that he looked just like Liquid), in MGS2 he's let it grow out rough, but still bound by a bandana, which to some extent brings to mind a collar. It's a taming element to his look. Visually he's torn between being wild and free and professional and limited.

To create Solid Snake's 3D incarnation they didn't just make a model of a guy's head in a computer. It's more like they drew a character's face based on a conception of who he was, as one would when making a manga, and then modelled that face. Scroll back up to him. He doesn't really look like a man. In this early PS2 period especially games still had video game people. They were representations of people. Somewhere between stylised and inherently limited.



Maybe there are people out there fortunate enough to look like Raiden. But none of them were brought in for scanning to make him. Again, concept and realisation.

In the two main examples of the OP, Ridge Racer and Final Fantasy 8, we at most have humans used as visual reference material. Squall in the finished game is absolutely not River Phoenix. Were the designers looking at Phoenix at some point? Maybe. But Squall doesn't look different because they fucked up depicting his face. Squall looks different because he's Squall. Squall was meant to look a certain way, to be a certain kind of person, and they succeeded in capturing that.

I think that what went wrong in 3D video games, why they started looking bad, was this shift from attempting to realise concepts in three dimensional characters. Constructing idealised humans. To attempting to put real people inside video game space. Scanning tools themselves obviously facilitate this to a massive degree, but the sensibilities I believe are the key difference. Even in newer games you can tell pretty readily which ones are trying to put a person in your computer and which ones are trying to create an idealised idea of a human in your game (anime).



"Looks like a Final Fantasy character" became another way of saying "inhumanly beautiful" for a reason.



While you don't hire Debra Wilson because she brings to mind any particular thing (or do you...), you hire her because you specifically want Debra Wilson. You want the face of a real middle aged black lady. I don't actually recognise bottom right here though, it's Forspoken (guessed and looked it up). Interestingly she's almost unrecognisable there. The Japanese, when given Debra Wilson tried to turn her into something else. That's very funny to me. I really don't think they get using celebrities for the hell of it. Closest thing to an exception is Hideo Kojima, but he's not an idiot. He doesn't fetishise Hollywood. he uses these people for brilliant effect. All of these chosen actors in Death Stranding fit their given roles to the point that I couldn't really imagine them being otherwise. And Death Stranding being a game about human connections in a strained age makes the use of photorealistic actors feel relevant and purposeful for possibly the first time ever in my opinion.



Death Stranding is not so much about specific faces as being able to recognise the humanity in a face. It's more significant I believe that performances are motion captured in this game than faces. Several characters are actually scans of one person played by another (quite well, I liked the guys doing Refn and Del Toro). Nobody in Death Stranding is anime in their visual construction. They're people. They are then elevated to anime by the fact they are in a Kojima game. They are more specifically Kojima people. We're back to concept and realisation. Every other part of the above character is some kind of considered artifice to convey a certain sense of who he is to express certain ideas in this story. If this was a hard realist work trying to simply emulate reality his name would not be Sam Porter Bridges.

So what makes a game anime? As I said many times already, I believe that anime is media constructed from concept to realisation. Many things of course were like this, but the japanese are now the last men standing. If you think a Western game feels constructed along comparable lines, please show us all what you have in mind.

Found a Japanese man doing these on twitter.



They don't really look like their subjects, do they? The idea he has in mind is taking the wheel.

anthony Wrote:[...]

Your reply reminds me that I really have a tendency to get ahead of myself in my posts, there's no need to skip over things that are only redundant to a tiny percentage of those reading. Would have been much nicer to begin with something along the lines of "Of course, all the characters I'm going to talk about here were 2D drawings first, often many 2D drawings, before they were ever 3D models. This isn't because the developers needed references to use in the modeling software or even that they needed them to be slightly idealized before doing so. Beginning in 2D drawing isn't a contrivance devised in absence of being able to scan in entire humans. They start there because the goal all along was to draw Anime in 3D." Anyway, on the topic of drawing Anime in 3D, watch this video everyone's already seen before:



Xrd is trying as much as possible to "draw Anime in 3D". This counterintuitively requires less literal 'illustration' after the 3D construction has begun than would be required for a traditional 3D character.




There is almost no drawing done on the texture files for these characters. Really the only thing drawn on here at all is the eye (makes sense). Everything else is done through the careful process that might be compared mechanically to sculpting rather than painting, it's work requiring alterations to small parts of the geometry individually to form the desired result. In this case, the desired result is an impression that 3D software wasn't used at all. Shading is done by giving vertices properties and tinkering with variables. The "pen lines" on the model are created by moving vertices on the UV into the areas off of their "piece".






Not like the more "old fashioned" method of 3D drawing is less strange. You've got to be cruel to drag 2D into 3D. The final step towards creating the character, after the model is done even, is to fit your drawings onto the flayed and spread out hide of the character you're working on. If you don't hang them up to dry like this, you can't finish the job.

anthony Wrote:Maybe there are people out there fortunate enough to look like Raiden. But none of them were brought in for scanning to make him. Again, concept and realisation.

In the two main examples of the OP, Ridge Racer and Final Fantasy 8, we at most have humans used as visual reference material. Squall in the finished game is absolutely not River Phoenix. Were the designers looking at Phoenix at some point? Maybe. But Squall doesn't look different because they fucked up depicting his face. Squall looks different because he's Squall. Squall was meant to look a certain way, to be a certain kind of person, and they succeeded in capturing that.

Here's another video everyone's seen that demonstrates this point nicely. They started out drawing girls with their direction being the desire to create an appearance that evokes something innocent, moderately damaged but not irreparably damaged. Sophie Marceau and Charlotte Gainsbourg were used as guides when things started looking not damaged enough. I guess to them innocent, lost, and somewhat hurt looking is French, I don't disagree. Anyway the other person used as a guide is the actress Heather who mocapped, voiced, and inadvertently named Heather. She was literally a guide, like a pose reference they were drawing over. Obviously motion capture of the time looking really weird is fitting for Silent Hill. Working with the available tools rather than pushing them uphill towards the misguided end goal of putting existing humans in your screen.

Excellent thread. I have nothing more to add at the moment, but I have thoroughly enjoyed it so far and am excited to see more discussion on this topic.