Sculptor Alberto Giacometti once said to his biographer James Lord, "The more you struggle to make [a work of art] lifelike the less like life it becomes. But since a work of art is an illusion anyway, if you heighten the illusory quality, then you come closer to the effect of life."
It's not a coincidence the massive hardbound bible of animation published by The Walt Disney Company is named The Illusion Of Life. Animation, like cartooning before it, is about leveraging the audience's imagination, letting their suspension of disbelief do the heavy lifting, using the illusion. Maybe that's why as computer-assisted animation techniques become all the more "realistic", they become less adept at the very thing animation seems best for: using the illusion.
Anime makes extensive use of CGI, partly as a cost-saving measure, but sometimes as a way to present an aesthetic. At its best, it shows how the illusion can be used well; at its worst, it's like a bad synth cover of a beloved tune. You never know which it'll be.
LSD (labor-saving device)
There's two general ways anime, and animation generally, can use CGI prominently. First is as a substitute for hand-drawn animation, as a labor- and cost-saving device. BLAME!, Kado: The Right Answer, Freedom, RWBY, and the most recent TV version of Berserk are all in that bucket. Anyone who's seen even any two of those projects knows how broadly dissimilar they are in their aims, and how wildly the quality of their CGI can vary. But all of them aim for different things, and so what would be a deficiency in one project would be a complement to the mode of expression in another.
With BLAME!, CGI works both as a cost-saving device and as a mood enhancer. The story's setting is a bleak dead-tech future where humanity is an endangered species in a world of viral machinery. CGI, even the CGI of a generation ago, lends itself nicely to depicting such an environment — all hard lines, jagged edges, cold surfaces, and industrial colors. Hand-drawing much of what's shown in it, including the action, would have been prohibitively expensive. It isn't a problem if the end result looks like it was milled by a computer, because the world depicted in it also looks that way.
RWBY couldn't be more dissimilar. Here, CGI was used as a rough-and-ready way to bring to life the adventures of a team of girls learning to hone their powers and defend the world from evil. The imagery itself is basic — simple colors, low-poly environments, character designs with just enough detail to get the point across. All that's just a means to an end, a way to close the gap between the physical screen and the show's spirited storytelling and lively characterization. What the show doesn't have in technical merit, it more than makes up for in other ways that stick to the ribs. The beauty of such a thing is, if at some point the show was reworked on a higher budget (or a live-action production), its existing story would provide a solid base to build from. It isn't good just because it has a scruffy, lo-fi charm, although that's one of its current winning points.
Kado: The Right Answer had at least as big a budget to burn as BLAME!, but to a different end. It's also SF, but more in the classic Arthur C. Clarke or Clifford Simak mold: Aliens from outside our dimension come to Earth and offer to accelerate the evolution of the human race, but by way of costs that may not be immediately apparent. Here the CGI is used for generally realistic, real-world, modern-day imagery — pretty, polished, if also slightly stiff, in the "marionette operated by one's elbows" way people still seem to be rendered in CGI. Where it works best is when it's used to render the extradimensional being, who manifests as a human of otherworldly handsomeness and whose vessel is apparently a giant four-dimensional Menger sponge. The whole thing doesn't look bad; it's just that I kept thinking how using CGI to depict just the alien presences, but using hand-drawn animation for the human characters, would have allowed each to throw the other into sharper relief.
There's almost nothing I can say about the new seasons of Berserk and its animation quality, or lack thereof, that hasn't already been said. The original late-1990s TV series, all hand-drawn and done on a TV budget, had a crudeness to it that at first seemed like a step down from the fearsomely detailed art of the source manga. When the project was brought back to the screen over the last few years, first as a movie trilogy and now as a new TV series, the production team opted for CGI all the way. Sometimes it worked; the designers labored to replicate the kinds of visuals associated with hand-drawn animation, like line hatching. But it never looked less than awkward, and the quality of the second season looked truly belabored at times.
The real source of the problems in Berserk wasn't the technical issues alone. It's how what was once aesthetic crudeness — the rough, but also powerful and direct hand-drawn look of the original series — has now been replaced by wholly technical crudeness. It doesn't look rough or spontaneous or unpolished in an endearing way; it just looks spastic and terrible and clumsy and cheap. It's not abstract enough from its original source to allow us to fill in the gaps with our own imaginations (and that would have been a bad idea with the visuals they were deriving from anyway), and it's not quite polished enough to get completely out of the Uncanny Valley. What a shame that one of the properties that most deserved to come back to the screen and finish what it started is only doing so in a way that makes its audience feel bad for it.
One more show that deserves mention in this context is Kemono Friends, a "cute anthropomorphic animals" story that's become a hotbed of fan theorizing but also a favorite on its own sweet-hearted merits. It's almost entirely concept-, story-, and character-driven. The imagery is practically an afterthought; most of the jokes about the show revolve around how bare-bones and unsophisticated the graphics are. But because the rest of what the show is has found a receptive audience, swipes at the graphics become superfluous. The imagery is just enough to get the message across, and with the audience that's receptive to the story, the low-rent look becomes part of the charm over time.
I think now also of The Fake and King of Pigs, Korean productions that are CGI but again where story and character are first and foremost. The animation is unspectacular, in the same way that a low-budget feature shot on 16mm (or on consumer-grade digital, as would be more likely these days). But they accomplish just enough to get the job done. Since they don't present us with anything spectacular, we don't go looking for it, and the seething, biting drama on display eventually takes the full share of our attention anyway.
The (sort of) real thing
The second way CGI can work is by trying to go all the way towards emulating reality. There, too, lies a spectrum of possibilities — not in the sense of "more to less realistic", since forensic realism is well within reach now. It's more about "more to less appropriate," since why you would want to make things look forensically real is the more appropriate question to ask than how.
Right now, the absolute high bar for forensic realism in CGI for me is Kingsglaive: Final Fantasy XV, the feature-length film that precedes the video game and sets up much of its backstory. It's staggering how convincing the movie looks the vast majority of the time, although that only makes the few times when it isn't 100% convincing all the more a thumb in the eye. Why it is like this is another story. It seems to be hewing for forensic realism less for the sake of being immersive than for the sake of aesthetic consistency both with the game itself and with much of the preceding Final Fantasy cinematic universe: the 2001 film, the Advent Children movie, etc. In other words, those things establish a pattern and set a bar to clear, and Kingsglaive is duty- and honor-bound to follow that pattern and clear that bar.
One recent production that arguably used CGI for hyper-immersivity is Gantz:O, the feature film derived from the long-running SF cyberpunk/adventure manga. The story involves a team of recently dead people brought back to life with technology, and recruited into a kind of murderous survival game where they have to fight off aliens that only they can stop. All this is set against the backdrop of a photorealistic urban environment — Tokyo and Osaka — where the level of realism matches and sometimes even exceeds Kingsglaive. But the character designs are deliberately cartoonish; one glance at their Kepwie-doll faces, and the illusion breaks. The only visible rationale for this seems to be to allow the CGI incarnations of the characters to hew all the more closely to their hand-drawn counterparts. Maybe that makes sense given that the film is aimed mainly at fans of the original material (it's too one-dimensional and arbitrary for most other audiences), who are intended to appreciate such a touch.
The same thinking seems to have been at work behind the CGI feature film made from Leiji Matsumoto's Captain Harlock. It's also hyper-realistic in its look, save for the characters themselves, who look like stylized 3D models of their original designs. And again, the effect is jarring; two incompatible design choices have been forced to live together in the same film. It doesn't help that the film itself is murky and lacking in proper dramatic momentum to boot.
Picture (less im-)perfect
In theory, at the end of the day, it's all CGI now. Hand-drawn animation is most often drawn directly into a computer or scanned in, and the resulting composite is assembled and processed digitally. But CGI that calls attention to itself as such is another story.
The first thing that's become clear to me after reflecting on all this is how using CGI to make things look hyper-real in the context of an "animated" film seems to miss the point slightly. Make things too real, and the pre-suspension of disbelief that animation provides as one of its inherent by-products threatens to disappear. It matters less that animation looks realistic than it looks evocative, and that it has the level of craftmanship to support that.
Second is how it helps to find the kind of imagery, and stories, that are best complemented by the way CGI is used and how it's limited by the current state of the art. We can get shockingly close to reality now, but the end result is that we become only all the more sensitive to the failures; any place we fall short only advertises the limitations all the more. Maybe, again, the point is not to try and reproduce the reality we know, only with a few selective tweaks. Maybe we're better off sticking to the original lesson provided by animation: creating a reality that begins and ends with its own rules, and inviting others to join us there.