Hello! My name is Dan, I am a professional animator,
and this is New Frame Plus. The remake of Final
Fantasy VII is here, and it is affording us all
a very rare opportunity: the chance to see a classic, beloved
game updated to modern AAA standards. To see it not just remastered,
but re-imagined from the ground up. Every single asset,
every game-play system, AND every story sequence
reconsidered and built from scratch! As an animator,
that last one makes me REAL excited. Consider the potential here! These animators and directors and
performers have the opportunity not just to make these scenes prettier, but
to flesh out their character performances and cinematography
with a level of nuance that the original game
simply was not capable of. To show you what I mean, I'm going to do a deep dive into the animation of just
the opening title sequence in both versions of the game,
starting with the 1997 original. It’s easy to forget,
looking at it now but, Final Fantasy VII was a
technological leap in its own right; the first game in this franchise to leave
those familiar 8 and 16-bit roots behind in favor of Sony’s new PlayStation and all the shiny new 3D
graphics it was capable of. But these big hardware transitions
always come with some growing pains, and Final Fantasy VII
is loaded with them. The title sequence begins as an FMV
cinematic; basically a pre-rendered video. The ability to store movie files on disc
was one of the more exciting new features of the PlayStation’s
CD-ROM format. We open on a field of stars, the camera
lazily drifting around in space for… probably 20 seconds
longer than necessary. And then we cross-fade to a medium
close-up shot of a woman’s face bathed in eerie green light. And, I should probably
just go ahead and say that the character animation in Final
Fantasy VII’s pre-rendered cut-scenes is just... terrible. They've got poor body mechanics, no sense of
weight, dead-eyed expressions everywhere… the craft of animation is fundamentally
about taking something not alive, like a drawing or a CG model, and creating
the illusion of life in that thing, and you will rarely see that
illusion work in these scenes. Now, this failing is
COMPLETELY understandable. CG animation was still in its
very early days at this point, And Squaresoft’s cinematic animators would quickly find their
feet over the next few years, so no disrespect towards them! Honestly, nobody in 1997 was
delivering great CG character animation in their pre-rendered cut-scenes. ...But yeah, this first effort
is pretty rough, though. So then we cut to a new angle, and we see that this woman is kneeling in
some sort of dark alley looking at something (although we can’t really see what
that glowing green something is) And it’s here that we can see that Aerith’s
character model is actually segmented, constructed of several separate pieces
of overlapping geometry. You can see the most obvious seams here
at her elbow joint and there at her waist. Now, this is actually how
all of Final Fantasy VII’s in-game character models
are constructed too; it was a pretty common character-modeling
approach in those early 3D days when deforming a character mesh with an
animated skeleton of joints in real time was just a lot for that
early hardware to handle. Of course,
animation in a pre-rendered movie isn’t bound by those
real-time processing concerns, so I’m guessing that this segmented
character model being used here is just an artifact of those real
early wild west days of CG animation where everyone was still trying
to just figure out the basics. And it’s also in this shot that we get our first taste of Final Fantasy
VII’s full body character animation, which,
like I said before: not great. You see the way she finishes standing up
and THEN starts straightening her back while her hips stay
completely locked in place? That sort of isolated motion is how
machines move, but not how humans do. All the parts of our bodies tend to
coordinate together when we move. It’s actually surprisingly difficult to move
just a single isolated part of your body without any other part
moving just a little bit. It what makes dancers like
these so darn impressive. Granted, I suspect that part of
the weirdness of her movement here might be due to the fact that
the animator's having to cheat to make her stand up AT ALL. Look at the lower half of her body for
the duration of this standing motion... You notice how her legs don’t
actually bend that much? Here, I’ll play it back in reverse...
Kinda looks like her knees bend slightly, and then her legs just stiffly
sink into the floor, right? I’m guessing that is
EXACTLY what’s happening. I would wager that this character
model (or at least her skirt geometry) wasn’t rigged to bend that way. Or maybe
there was some other technical complication preventing her from ACTUALLY being
shown kneeling on the ground, so the animator had to cheat
to make this shot work at all. And I say, hats off to you, sneaky animator
who made this impossible thing work. You weren’t the first
and will not be the last. ...oh, THAT’S why the camera
doesn’t tilt down far enough for us to see the glowing thing
she’s looking at, isn’t it? Because lowering the camera any further would
reveal she’s not actually crouched down and break the whole illusion?
Aw that's it, isn’t it! Sorry, I got distracted,
what was I talking about... RIGHT. Yes,
Aerith moving like a robot. Okay. And it really doesn’t help that her
eyes are 2D textures pasted to her face that are only capable of
looking straight ahead. Like, look at her head turn here. You see
how her eyes just stay locked in place, staring forward for the entire move?
That, again, is how robots move, not humans. I mean,
try it yourself right now. Try turning to look 90
degrees to your left while keeping your eyes focused straight
in front of your face the entire time…… is that not the most
robot-feeling thing to do? Now, I could keep nitpicking
this little stuff all day, but at this point you get the
idea, right? This is clearly a team
of artists and engineers learning their way around
a new animation medium, probably using the software equivalent of sticks
and rocks compared to modern animation tools. So, I think we can
cut them some slack. In fact,
let’s say some nice things! I really like the way this
extended shot is staged; the way it starts in close spending a quiet
intimate moment with this flower girl, and then slowly begins to pull back
as she walks out into the busy street. And then the bustle of this dark,
grimey city suddenly intrudes on it, breaking our line of sight and just taking over the
frame as the camera continues to pull back and the flower girl slowly
disappears into the cityscape. And, by the way: side benefit of having
all these nameless extras shrouded in moody shadow during this camera move?
Much harder to see animation flaws! Except on this guy. HEY. I see that hitch
in your walk cycle, random dude. Better get that looked at. But alright, we’re done with
character animation for a bit. The camera pulls back and back until we’re
looking over the entire city of Midgar. And just keep in mind: even though all of
this does look very rough and early CG, this one extended shot has been
a showcase of cinematography and character animation detail
(relatively) that, up to this point, Final Fantasy has never been
capable of doing before. Now it’s time for the second
half of this opening cinematic, and this is where stuff
gets really interesting. So, the camera begins its
flight back into town, inter-cut with some quick
shots of a decelerating train. We fly down, down, down until
we’re back at ground level. And the camera settles into place just as
this train pulls into the Sector 1 station, and the instant the train stops, the pre-rendered video portion of
this intro ends and we seamlessly(ish) transition from a movie
file to a static 2D image, the likes of which Final
Fantasy VII is going to have us running around in front
of for most of the game. But before we get
to the action here, I’d like to direct your
attention to these two guards. Because you probably noticed,
those guards appeared on screen BEFORE this pre-rendered video ended. And they were not PART
of the pre-rendered video. Those were in-game character
models, being rendered in real time, being layered OVER the FMV. This is a pretty cool little
trick they're pulling off here. I mean when you think about it,
almost every environment in this game is just a lavish pre-rendered 2D image
being layered behind the invisible terrain your 3D characters
are standing on, and there’s no real reason that that
background image CAN’T be a video. But to make that work, the animators had
to make sure that the game-play camera which is seeing and rendering
these two guards, imitated the pre-rendered cut-scene
camera’s movements perfectly to preserve the illusion of those 3D characters
being grounded in that moving 2D background. It’s a complicated trick, but quite effective at blurring the
boundary between movie file and game-play. Square would actually do
this a lot in the PS1 days and by Final Fantasy VIII,
they were getting real extra about it. ANYWAY. Now that the pre-rendered
video portion of this sequence is done, from here on, all of the characters
are going to be rendered in real-time. And real-time story scenes
in Final Fantasy VII have to deal with a handful
of very tricky limitations. Limitation No. 1: Because every animation in the game
takes up some amount of memory, the devs had to try to limit the number of
unique animations appearing in each scene. Because the more animations
needed in one single environment, the longer that environment
was going to take to load. Limitation No. 2: Because most of this game’s
environments are pre-rendered 2D images, their ability to move the
camera is extremely limited, and camera rotation is
completely locked down. So pretty much all of the action in any
given scene has to be played to one angle, kind of like a stage play. And Limitation No. 3: Final Fantasy VII’s characters....they
don’t really have faces. I mean, they do, but it’s near impossible to see them
at the games original resolution, and those faces weren’t
built to animate anyway. So, unlike the character
animation in the 16-bit games (which ran almost entirely on
exaggerated facial expressions), Final Fantasy VII’s characters
have to rely on gestures and heavily-exaggerated
body language. And that's not actually a
terrible trade-off in the end, but not being able to use facial expressions
is a pretty big acting limitation. All of that to say: for the duration of this
scene, the camera's barely going to move, the character performances are going to
be created by stringing together short, re-usable animations, and nobody is
going to be acting with their face. Ok? Ok. So first we see Biggs jump off the top of
the train and throw a guard to the ground, which reads SURPRISINGLY CLEARLY
considering how far away he is from camera and how low resolution
this game is. Next Jessie runs in and takes out
the second guard with a kick, which reads a little less clearly than the
throw, despite being closer to camera, largely because the guard's body
is kind of obscuring the move they probably could've tried to clear the silhouettes
a little better here, but I'm nitpicking. Notice how exaggerated all
these animations are, though! With render resolution this low and
character models this simplistic, there is no room for subtlety. The animators are having to push
everything to the extreme to read clearly. Wedge hops off the train next. And then
we see one of our main characters, Barret, who looks back toward the train and
gestures impatiently. The camera pedestals up a bit for the
grand entrance of our main character, who flips down for a showy,
absolutely cliche three-point landing. And there are a couple details
I really like in this moment. First, having Barret run out, look up
and gesture toward the train is great not only for directing our attention to
where Cloud’s entrance is about to happen, but it also signals to the new player that
Barret is in a position of authority here, and it sets the tone for the early antagonistic
dynamic between Barret and Cloud. And the second thing I really like is that
Cloud’s unnecessarily acrobatic entrance is a perfectly fine way to make your player character
feel like the coolest tough guy in the room (...eh, sort of), but in the LONG term, it also effectively sets up
Cloud as the kind of person who would try to present himself as the
coolest tough guy in the room, which is a surprising level of nuance for
a clump of untextured, faceless polygons hand-springing off a train. But yeah, then Barret tells you
to get moving and there you are, playing Final Fantasy VII. Overall, as clunky as these PS1
graphics and early 3D animations are, I’m surprised at how effective
all of this still manages to be. That’s kind of Final Fantasy
VII in a nutshell, really. This brings us to the
Final Fantasy VII Remake, a complete overhaul of the original game
with the full power of modern AAA budgets, talent and technology behind it. And of course, there are, TONS of
improvements you can just see at a glance. The high resolution graphics,
the high-fidelity character models, the lighting, the shading, the
effects... all of it gorgeous. But the two things that interest
me most are all those brand new motion-captured character performances
AND all the creative adjustments and changes designed to help this
opening achieve its story goals in ways that the
original couldn’t. Let’s see what 23 years of
technological advance can do. We begin, once more, drifting in a star fie-- ...hang on. ...yep.
This is different. Okay! It turns out the Remake diverges from
the original intro sequence immediately. This time we open with a bird
flying over a barren wasteland and we get our first look at the Midgar
skyline in broad daylight, almost completely
shrouded in smog. We get a good look at the actual city
itself during the day time rush, all of the densely-packed buildings,
construction work, some kids riding bikes… honestly, this new opening could probably
have stood to cut 20 seconds too but, they do establish some important
thematic imagery in these new opening minutes: dead landscape,
a bustling smog-coated city, dead flowers SPECIFICALLY, and, as the sun goes down, the bright green glow of Midgar’s
mako reactors illuminating the night. Now all of this is stuff that fans of Final
Fantasy VII are already well familiar with, but these two minutes sure do give the
newcomer a lot more useful information about this setting than
45 seconds of stars do. And that spray of green particles from
the reactor transitions us beautifully into the classic opening
sequence we know and love. I would say we are off to
a pretty strong start here. What’s more, craning the camera
down toward Aerith like this instead of simply cross-fading
affords us a much better look at what’s actually happening in
this scene; the Mako energy - the planet’s very life force,
we’re soon going to learn - is seeping from a broken
pipe in this dirty alley. And thanks to those two
extra minutes we just saw, the new player can probably start
connecting some dots on what’s happening here! Much more easily then they could
have in the original, anyway. And when we reach that
close up of her face, we can see that she’s
sort of lost in this glow, just basking in it, almost
meditative, possibly praying. And when she opens her eyes to look into
camera - just like she did 23 years before - ...man. What an incredible difference two
decades and 4 console generations can make. This Aerith actually looks alive. Even in this tiny moment, there is so much
more subtle fidelity to the performance. The way her eyelids stir
slightly before she opens them. The way she first looks down at the
pipes she was focused on previously BEFORE her gaze drifts
up toward camera. The way not just her eyelids but also her
eye brows move slightly as she looks up. And all of those
subtle eye darts... Especially when compared to
the original robo-Aerith, what incredible realistic
acting fidelity here. Then we cut to the second shot,
very similar to the original angle, but with some better framing. Except this
time, Aerith doesn’t stand up right away. She just... stays here, gazing at this leaking
pipe, lingering in the moment. I REALLY love this change. It’s not a radical departure from the original,
but the extra time spent lingering here gives us a chance to emotionally
invest in this moment WITH her, in a way that you really
couldn’t in the original. And immediately AFTER that,
another change that I love: Aerith stands up NOT just
because it’s time to stand up, but because she suddenly senses a threat
at the end of the alley. This is so good. For one, it gives her a motivated
reason to stand up and leave this alley she was just having
that moment in. Two, it further reinforces what the original
game was trying to drive home as well: that this city is
a hostile place. And best of all, three: if you already
know a little bit about Aerith’s story, then you know that she
is ALWAYS being watched. She is a Person of Interest in this town.
And so seeing her get spooked by… well, seemingly nothing
at the end of this dark alley helps to plant some seeds for
her story WAY in advance, and, again, the original version of
this cut-scene didn’t do that AT ALL. Also, by the way? THAT’S how you animate standing
up from a kneeling position right there. A+ Now here comes the biggest departures
from the original opening sequence: rather than maintaining one long shot from Aerith
standing to Aerith walking out of the alley to zooming out over the city to
zooming back into the train station, instead they cut to a new shot here and
add an entirely new series of events. Aerith hastily walks
out of the alley, nervously looking behind her
to see if she’s being followed, and accidentally bumps into a
passerby, dropping some of her flowers. She kneels to pick them up as the rest of
the city goes about its business around her (which is EXACTLY in line with
the tone of the original). But one of the flowers gets
trampled by a careless pedestrian. She slowly picks it up, cradles it in her
hands… and lingers in this moment too, just like she had with the
broken pipe in the alley. Now,
the symbolism of this addition combined with the shot of the dead
flowers earlier is Not At All Subtle. But for anybody experiencing Final
Fantasy VII for the first time (and a LOT of people
are going to be), drawing a direct visual parallel between
how Aerith treats the trampled flower and how she treats that broken
pipe leaking a weird green light, that I think is a worthwhile
bit of thematic foreshadowing. The flower girl is planting seeds.
Deal with it. But ok, time for that big flyover shot. The camera pulls back and back
and back… and keeps going because the city is huge this time and we get
that view of the entirety of Midgar, looking as gloriously
ominous as ever. Then we swoop back in to Sector 1, inter-cut
with those same shots of a decelerating train. Only this time, the camera flight ends
framing our main character atop the train. Again, arguably not a necessary change, but
it is a nice little hint to the newcomers that something is about to happen when
this train gets where it’s going. And can we just take a moment to appreciate
how far physics simulation has come? We already saw a ton of it before
with Aerith’s hair and clothing, but look at that wind whipping
through Cloud’s hair and shirt here. Awesome detail. Then we see the train pulling into
the station, and right... THERE, if I had to guess, is where we transition
from the pre-rendered cinematic to rendering this
scene in real time. Did I mention, by the way, that everything up until
now was almost definitely a pre-rendered movie, just like the original game did?
Because it probably is. And why shouldn’t it be? I mean, everything we’ve
seen so far in this sequence is meant to play out
exactly the same every time, AND the camera has to cover a ton of ground
and show all of Midgar at a couple of points. There is literally no reason to try to
make this all render smoothly in real time when they could just play a pre-rendered
movie and allow the actual level to load in the background so
we are not kept waiting. (A lot of modern games do that, by the way.) And, if my guess is correct,
they’re transitioning from movie file to in-game assets at almost the exact same
point as the original, so that’s fun! And if I’m wrong about
the exact hand-off moment, honestly, that's just a testament to
how good real-time rendering has gotten in modern AAA games that
it's this hard to tell. But here’s where the Remake diverges
heavily from the original again, because this game isn’t bound by any
of the original’s limitations. This is a fully 3D space
being rendered in real time. Now, the camera can move
wherever the director wants, so this time, they can actually shoot this scene using
standard cinematography techniques. So, instead of a static angle showing everybody jumping off
the train and fighting guards, this time, we get an uninterrupted moving Steadicam
shot of the attack from the guards’ perspective. They patrol the platform a bit, we hear
one of them get taken out off camera, the other hears this and turns to
see the platform behind him now empty so he gets suspicious
and starts looking… And watching this all play out
from the guards' perspective not only lends the scene more dramatic
tension, it also gives AVALANCHE a chance to look more
coordinated and stealthy. AND Jessie still gets to knock
a dude out with one kick. Then we get our introduction to
Barret (who is looking GOOD). Barret turns,
and just like in the old days he waves for Cloud to get his
butt off the train already. And Cloud obliges, just as before: in the
most unnecessarily dramatic way possible. And yes, this entrance
IS absolutely absurd, but, even ignoring the fact that it is a
direct callback to the original train jump, remember that performative coolness
and tough-guy posturing are CORE elements of Cloud’s character
at this point in his arc. WEDGE: "...but you know what I think?"
CLOUD: "Not interested." What we are seeing
right now is a facade which we are (hopefully) being
set up to eventually see through. And it does seems like Square’s narrative team
is leaning into that setup with some intent here JESSIE: "Ok THAT was pretty cool." I am excited to see
how they handle this. But, yeah, that’s it! Now we are playing
Final Fantasy VII, But Prettier! Only one cut-scene into this game and I am already really digging the
creative adjustments they’re making here. They’ve not only recreated the
original sequence with more detail, but actually added more
information and nuance, largely through the animated character
performances and cinematography. That’s what makes me so
excited for this Remake: because, good or bad, this return to Midgar
represents a HUGE animation opportunity. Final Fantasy VII’s story is PACKED
with some really juicy character scenes that the original game could only deliver through
text boxes and simplistic pantomime acting. There is so much untapped potential for
nuanced, animated acting in those scenes, and this Remake has the chance to
go back and make good on that potential! I am so jealous of the animators
in charge of re-imagining the scene where Cloud talks with
Aerith in the church. Or that scene where the party
is caught in Shinra tower and they’re all talking to
each other through the walls. OR the scene where Cloud is telling HIS
version of what happened five years ago, and Tifa’s just sitting in that
room, knowing his version is wrong, not knowing why, but
not SAYING anything? IMAGINE the blend of emotions happening
in her head throughout that scene. That is an animated
acting GOLDMINE. I’m calling it right now: whenever
that scene happens, watch Tifa’s face. That's one of the coolest character acting opportunities
in this whole dang game and they better not miss it-- SORRY.... I am sorry.
I got excited. I don’t know if this game is going
to be great, or a disappointment. All I know is that this kind of
opportunity for comparative animation study does not come along often and
I cannot wait to get started. I’m actually going to be playing
this Remake for the first time over on our other channel,
PlayFrame, starting Monday. And if you have enjoyed me nerding
out about Final Fantasy VII animation and want to hear HOURS more of
it? Well, I hope you’ll join me. And thank you very much to
CanvasWolfDoll for suggesting this topic! I had so much fun
digging into this. If you’d like to suggest a topic for
a future animation analysis video, then consider supporting the show like
all of these good people listed here. Thank you very much for watching,
and I will see you next time. [music]
yay
I could watch/listen to Dan explain stuff like this all day.