CGWorld Vol. 296 Interview with Creatures

From Lewtwo




Pokémon Scarlet and Violet are the kanzen shinsaku of Core Series Pokémon, and are the first open-world games in the series. We've done a deep-dive exploration into the work of animators at Creatures, Inc., who have helped breathe life into many Pokémon inhabiting the Paldea region, the setting of Scarlet and Violet.

[translator's note] Kanzen shinsaku refers to the main games of a new Generation.
In the original interview, CGWorld abbreviates Scarlet and Violet as "S • V", using the uncommon middle dot.


Animation considerations based on the Ecology and Temperament of each Pokémon


Wakasugi: First of all, what was your role on Scarlet and Violet?
For Scarlet and Violet, The Pokémon Company took the role of the production and branding; GAME FREAK handled overall game development including the specifications of the Pokémon's 3D assets; and our company, Creatures, were in charge of 3D asset creation for most Pokémon.
Under Fukaya, I was a manager and director of the motion team, much like he was.

[translator's note] Yoneya is the sub-manager of Creatures' art team under Fukaya, but is credited as the motion lead in Scarlet and Violet, which is a higher position than Fukaya's on those games.

I served as the modelling lead. The teamwork between modellers and animators was essential to this game's development; so I'm joining this interview, too.

Wakasugi: Could you tell us about your career, too?
It's my fourteenth year, mostly in the gaming industry. My debut work at our company was Sword and Shield, and I've been among the Core Series Pokémon developers since.
I began my career in CG animation and joined the company in 2017. I have about eleven years of experience in total, six in the gaming industry.
Much like Yoneya, I worked for CG animation teams until I joined the company in 2018; I have about twelve years of experience in total, five in the gaming industry.

Wakasugi: What are the first steps for 3D asset production in Scarlet and Violet?
Scarlet and Violet has more than 400 species of Pokémon, about 120 of which are new. GAME FREAK prepares the settei, motion instructions, and the like. Within [those documents], there are san-men zu, idling and attacking poses, expression sheets, etc., which our modelling team uses to work on 3D modelling. The Pokémon Company and GAME FREAK then supervise our modelling, and their approval brings us to a meeting between the art directors, the modelling and motion leads, the assigned animators in-charge, and so on.

[translator's note] San-men zu, literally "three-aspect figure", means views from the front, from above, to the side of an object.

We call it kyara uchi; setting the course for the specifications and motion of a Pokémon. For example, we settle the species-specific characteristics of that Pokémon and the range of motion for its joints. Basically, we follow GAME FREAK's settei and so on; however, they mainly document attack motion, which leaves [it up to] us to provide various emotional expressions, for the most part. Last but not least, our modelers' jobs go as far as joint setups and skinning, while the rigging controller setups are done by our riggers.

[translator's note] Kyara uchi seems to be the shortened form of "character uchi-awase", literally "character meeting".


Wakasugi: Are there cases where animators' ideas lead to further adjustments of the joints, too?
Yes, [the modelling team] listens to the motion team and adds joints to the parts that the motion team want to move, before passing [the data] on to the riggers. There are sometimes requests after the actual motion implementation [tests], like "Could you add another joint here, since I want this part to sway, too?" Implementing [test] motion often reveals issues that we hadn't thought of before; leading to frequent communication between our modellers and animators.

Wakasugi: What information is important for animators on the sheets with all the instructions?
The ecology. What lives do they live, and in what environments? Do they live in groups; do they have companionship; are they aggressive; or aggressive but also cool-headed when in battle? Even a gentle Pokémon has to attack during battles; will the Pokémon attack so the opponent won't get hurt? We deeply consider [aspects] like that.

Wakasugi: Fascinating, but it sounds hard to grant all those requests. (laughs)
That's right. (laughs) The assigned animators collect [additional] reference material based on the ecological data and poses [provided by GAME FREAK]; with that, as well as our brainstorming, we can solidify how their motion will turn out. After the kyara uchi, we [prepare] a prototype model, with which we create some motion within Maya as [a sort of] taste test. Once the supervisors' approve, we finalize the remainder of the animation assets.
Prototype animation models have the general shape and joints completed, but not the textures or lighting.
Once we finalize the models, we replace the prototype models with the final ones to further improve motion. Implementing motion into the actual games requires animation-state setups using a tool, which is another job done by our animators.

Wakasugi: Is that an in-house tool?
It's a tool developed by GAME FREAK. For animation-state setups, each animator consults GAME FREAK's templates prepared [in the tool] to configure animation transition endpoints. This sequence of work makes [the assets] ready for delivery, but that isn't the end; fine-tuning the motion while ensuring the transitions will continue right up until the development deadline.



For new Pokémon, their taste test motion is the cornerstone of their characterization, so we spend a lot of time checking it over. For older Pokémon, we emphasize the personality of a Pokémon across its poses, and its consistency of characterization.

Collecting reference material, not only for living things, but also for inanimate objects


Wakasugi: I guess quite a few Pokémon are difficult to showcase properly from certain angles. For example, in the case of Quaxly, were there some instructions like, "The pointy part of its head must be kept at this angle"?
There were no instructions on angles, but — Quaxly's head has a bouncy texture, you know. There was a request for maintaining that across all of its motion, which was tough for us to fine-tune.
The settei has some requests noted down, such as "This detailed [part] should always be visible"; so during the modelling process, we'll shape so that they consistently look good, and add more joints for fine-tuning, to use an example.

Wakasugi: What references were collected in the kyara uchi phase?
For Pokémon based on a real-life living thing, we collect videos about the ecology of that creature. As an example, for a Pokémon based on a dog, we seek out videos of similar breeds of dog. You know, a Shiba-Inu wags its tail like a "whoosh-whoosh!" when excited, while a Golden Retriever does it more gently. Our discussion can go that far as to which is the better for our interpretation.
Similarly, for a Pokémon with a frog motif, we think as much as "What species of frog?" to [comprehend] the essential [elements of that frog's] motion to incorporate it [into that Pokémon].
Many Pokémon are based on multiple living things and inanimate objects rather than one single thing, so we discuss what elements to incorporate, and which to part with. For example, Fidough is the shared concept of bread and a dog. Whilst its outer body feels like sticky and moist dough, it also showcases the animation and behavior of a small dog, which is also important for us to consider.

Wakasugi: That's tough! (laughs)


We observed dough and added the bread-like bouncy texture at key points. I believe the willingness to observe is essential, both in CG animation and gaming jobs.
For Pokémon based merely on inanimate objects, our development focuses on how to make them feel alive. Sandy Shocks is a Pokémon with its legs resembling U-shaped magnets, and is designed to collect ironsand unintentionally with its legs on [each] step. The assigned animator(s) watched videos of ironsand, noting "This is how ironsand is supposed to move..."

Wakasugi: Indeed, you can't model the flow of ironsand with only your imagination.
In an animator's career, you don't get to observe ironsand moving very often. (laughs)
Right, that was fun.

Wakasugi: So, apparently you're required to cover a wide variety of movements and come up with a range of [your own] ideas there.
I'm sure it's rare to have a game with such a wide range. There are a lot of things you can't imagine just by trying to act it out yourself.



Wakasugi: For Pokémon that have appeared in previous titles, do you use their existing animations?
We do that where possible, but there are moments where we tune them to meet new requirements, add new motion, and so on.
Scarlet and Violet had a change in the joint specifications, which overhauled the naming of each joint. As a result, we could hardly use motion from previous titles as they were, so we had to re-organize the data while consulting riggers and technical artists about what script would [best] migrate the motion at a low cost.
That was a change based on GAME FREAK's suggestion for easier data handling for future-proofing. The specifications of facial assets have been changed as well; almost all facial assets were newly created. Past Pokémon games used texture-based scroll animation to render facial assets, but it's been changed to joint-based animation since Legends: Arceus. It's cuter when the eyes and mouth move nicely, and that also offers a wider variety of expression; so we focused heavily on these facial assets in Scarlet and Violet.

[translator's note] The "texture-based scroll animation" appears to refer to how all Pokémon games pre-Legends: Arceus had their eye textures on a single image, which would flip-between different points on the texture whenever an expression was called.



The UI has buttons for each facial expression preset, which allows the assigned animators to simply click a button at the right point during a motion sequence [to apply a preset] to provide facial assets [in that sequence]. This not only streamlines the process, but also prevents inconsistencies across animators and issues with the designs.
With the change from the texture-based [scroll animation] to the joint-based [animation] for facial textures, our technical artists listened to the requests of the modelers and animators, and developed in-house tool(s). Portraying each Pokémon's characteristics consistently is key to maintaining the Pokémon brand. We've been trying our best as a team every day to achieve that.

Rebuilding Texture-based facial expressions with joints


Wakasugi: I'm sure the way facial assets are built is quite different between the texture-based ones and the joint-based ones. Did you have any difficulties?
Since there are more than 400 species of Pokémon, [we must] ensure consistency across the assigned animators; to achieve that, modellers created presets of facial expressions while looking up the expression sheets, and asked [the animators] to use them.
We decided that if each animator had manipulated the joints directly, the degree of freedom would've been too high, and it would've been difficult to standardize the style.

Wakasugi: Given that Pokémon have very simple designs, just a tiny bit of change can cause something somewhat... sakuga hokai.

[translator's note] Sakuga hokai is a term that can refer to severely bad animation.

Yes, raising the corner of the mouth just a little more could result in a different impression. (laughs)
What's more, we needed to assess how far we would rebuild the texture-based facial expressions with joints. For example, Goomy is a Pokémon characterized with a body that stretches and contracts vertically — its facial expressions with textures would hardly be able to contort, but some of those expressions were difficult to rebuild with meshes, for which we [had to] make many adjustments.
It was difficult to rebuild the Ku [く]-shaped eyes, common within manga and anime, only with joints; so we had to think of other ways.
Since the Core Series Pokémon games attach high importance of individuality for each Pokémon, we wanted to maintain familiar facial expressions as much as possible, and we discussed [this topic] many times with The Pokémon Company and GAME FREAK. Also, animation with joyful-looking eyes in the texture-based style would very often look strange when those eyes were gone [during the rebuilding process], which required us to rebuild the motion as well.





The Ku [く]-shaped eyes occurr as a result of the combination of joints and UV scrolls. Texture-based expressions could only switch between each other in stages, but the jump to joint-based animation now enables more natural eye blinks and mouth movements.
The workflow of joint-based [animation] for facial assets is as follows, for each Pokémon:
  1. The animation team tells the modelling team what animation the Pokémon should have. Their ideas [must be] shared in detail: ex. whether [the animation] involves whole or partial body scaling; or whether [the Pokémon] also flails when opening its mouth wide.
  2. The animation team set up the joints and the [rigging] weight, using this shared information. They then set the Root Joints, which allow the entire eye to move as parent nodes, and configure [each child node] for the intended facial expressions and shape, depending on each Pokémon.
  3. As for older Pokémon, the motion team tries using the existing facial asset data, checks for errors or clipping glitches, and debugs them.
  4. The final assets for the facial expressions are formed into a preset and registered to their in-house tool.
The look of the Pokémon was yet another example of overhaul in Scarlet and Violet, and so were their shader settings as a result. This has also drastically changed the mechanics of, for example, glowing [body parts] and fluid movement of surface patterns; so [the motion team] remodelled the motion of those. I believe it was a project where the motion team had a larger workload than in previous titles.

[translator's note] The "surface patterns" appear to refer to the normal maps that allow for a degree of texture on a Pokémon's body in Scarlet and Violet, such as Seviper's scales.


Wakasugi: I felt the movements of quadruped Pokémon in Scarlet and Violet were more realistic than ever, and I was impressed that they ran in such a way that showcased their skeletal movement. Have you enhanced the running motion of Pokémon that appeared in previous titles, too?
I'm happy to hear you felt that way. This is an open-world game where Pokémon are symbol encounters. Players can observe overworld Pokémon at a distance; so we've enhanced [the motion], considered the life-likeness and natural motion, whilst also meeting the necessary [settei, etc.] requirements. There was even some data from the 30 FPS era among the motion assets for previous titles, so we've rebuilt at 60 FPS as well.

[translator's note] A "symbol encounter (shinboru enkaunto)" is a wasei-eigo that refers to a fully visible encounter, either spawning or stationary.
Additionally— Fukuya's use of "the 30 FPS era" is vague. It wouldn't make sense for him to be referring to their development setup, as ever since the debut of these models in Pokémon X and Y, the Pokémon have been animated at 60 FPS, albeit this can only be seen exclusively in specific menus within the Generation 6 games.



When we [first] tried motion blending on the transition between idling and running, the berry clipped into its body, so we've adjusted the running motion. We could've opted to prepare a dedicated transition animation [between idling and running], but [we decided against it]; [inserting] a transition animation between idling and running would make the gameplay feel sluggish, and avoiding devising transition animations as much as possible [has become] our principle. I believe that this [new] sequence of idling to running has made the transition feel far smoother, and the [whole] animation look way more suited to the Pokémon's personality.

Meticulously developed post-attack animation featuring weight transfer


Wakasugi: When a Pokémon attacks an opponent in battle, it returns to the initial position and then performs weight transfer — I really love it. For example, [a Pokémon] taking a step forward and throwing a punch will [not only] step back to the initial position [but also] shift body weight with its other leg before going back to idling. In my opinion, given that Pokémon's designs are [often] simple, more stylized animation would hardly look incongruent. Why did you [opt to] attempt weight transfer [in an attacking sequence]?

[translator's note] Weight transfer is an athletic technique of shifting body weight from one part of the body to another during action. In-game motion for that is a level of animation follow-through after a Pokémon makes a sharp move, which makes it look more natural.

There are two points we're wary of.

The first is a balance with the time scale [of the attack]. An attacking animation [has to] have quite a short duration between its launch and the [final] hit, so there's only so much you can portray. [Our rules] are, therefore, that the animation before the hit should emphasize the coolness of the movement, and the one after the hit, which is the longer [part], should be crafted with great[er] attention to detail.

The second is how to transition back to the idle animation. [To make this transition more seamless,] we've adjusted the animation curve, overall movement flow, etc. so that the gap between attacking and idling doesn't stand out.

Wakasugi: So, the reason for the short duration of an attack is that you must attach an importance on the sense of speed [for the player] after they press a button, right?

[translator's note] There was a misprint on the printed version where the highlight for the interviewer's text was missing.

Exactly. It's not a universal rule because each project has its own guidelines, but Scarlet and Violet does have a short duration.
Those who have moved here from CG animation companies are often surprised to learn that [a Pokémon] must return to an idle animation. They're likely to wonder, "Why does it have to go back? Isn't it fine to transition into another pose?"
I thought so too, right after I joined this company. (laughs)


[translator's note] The tool appears to be fully in English, with the only exception being "保存" the Save button.


Wakasugi: You know, when a battle begins, players often look at their Pokémon from behind at an angle. I always thought this may have been a tough job, as it seems likely to result in uninteresting scenes, [or confusing situations] where you can hardly tell what Pokémon they're supposed to be, and so on.
In the motion team, we routinely have [technical] talks like, "So, it looks good from the front view, but the silhouette doesn't look as good when viewed from behind at an angle, so let's lower the arms a little". We've been focusing on maintaining some really cool motion from all angles, and spend a lot of time adjusting the idle animations in particular, which players often watch.

Wakasugi: In the filming industry, it's often fine to prioritize the rear view of a character, regardless of the front-view design, when you only need to see it from that view. When it comes to games, it's necessary to look consistently good at all angles, right?
Yes, we've been keen to keep the motion consistent from all angles, with the in-game camera switches in mind. Since Pokémon have simple designs, one angle often carries a different interpretation than another, which is a tough job.
I too had trouble [getting used to] this rule, right after I moved from a CG animation company. The [motion] lead pointed out my careless work every time they checked it over, like "It's no good from this angle".

Wakasugi: That sounds like a Rubik's Cube. (laughs) Like, you've solved the front side and found the back side messed up — that must be very hard.
You're right. (laughs) If we have the game's mechanics and the camerawork locked in at the time we create a Pokémon's 3D assets, it may just be enough to adjust their appearance using the in-game camera. In reality, that's not always the case, so it's essential to check from all angles. That being said, we can estimate most angles, and visual checks from those angles in particular are done very carefully. An example of these adjustments could be ensuring that a [Pokémon's] signature horn is visible, where its animations are clearly seen.

Wakasugi: What Pokémon were particularly difficult to fine-tune?
Pokémon with a lot of swaying parts. For instance, Quaquaval, the final evolution of Quaxly, has decorative wings and is prone to a messy silhouette, which was tough for us.

Wakasugi: Since its front-view design is [already] complicated, tuning up its rear view must've been quite hard.
I squinted at it constantly to figure out its silhouette.

Wakasugi: Were the swaying parts added manually, too?
Yes, manually. We have some scripts for automation, but we seldom use them. We tried to automate this, but manual adjustments on the visuals were required in the final steps; so we decided against it.




Some animators challenge themselves with animation they consider themselves weak at


Wakasugi: Is the same animator in charge of facial assets, body animation, and adjusting the flow of movement?
Basically, the same person works on the entire process. That being said, Pokémon that have appeared in previous titles may have a different member assigned to them in Scarlet and Violet. There are about 80 animators in total, both inside and outside our company.

Wakasugi: That's a lot! Pokémon come in a wide variety of sizes and designs, with a wide range of movement styles. Do you consider suitability when assigning an animator to a Pokémon? For example, is this person good at quadrupedal motion, or is this person better with cuter motion?
Their work obviously shows what they're best at; so we take that into consideration. On top of that, we'll respect their requests and preferences as much as we can. For instance, if someone says "I love insects", we'll assign them a Bug-type Pokémon.

Wakasugi: It sounds fun that staff can ask what [kind of] Pokémon they want to work on. Do some people say, "I definitely wanna work on this!" too?
(laughs) We want staff to work on their favorite Pokémon if the timing is right for the assignment. In many cases this isn't possible, so these requests won't always be granted.

Wakasugi: Did you assign the same person to a whole evolution line, like Quaxly, Quaxwell, and Quaquaval?
We want to assign the same person to similar[-looking] Pokémon, but the timing doesn't always line up. If we assign a different person, we'll give them completed motion [assets] for them to look up. As an example, we handed Quaxly's motion [assets] to the animator(s) in charge of Quaxwell and Quaquaval for reference, and asked them to reproduce the bouncy texture of the head. At the same time, some staff tell us, "I'm not great with [animating] birds, so I'd like to challenge myself". In these cases, we may assign them a less gimmicky and easier Flying-type Pokémon.

Wakasugi: The more gimmicky, the more difficult, right?
Many gimmicks and swaying parts can increase difficulty, yes. In particular, Legendary Pokémon are likely to be more difficult, as they have detailed settei, which we must stick to throughout development.

[translator's note] The staff appear to use the word "gimmick" to describe an appendage or an otherwise unique aspect of that Pokémon's design. This is different to the technical term of “shader gimmick” that is used elsewhere in the article.

Portraying life-likeness despite various limitations

For Miraidon and a few other Pokémon, the modelling team created shader gimmicks for the animation team to handle, such as the way a surface emits light, the motion of moving particles, etc. The modelling team has put in effort to properly realize the gimmicks shown on the settei, and depending on the Pokémon, there can be a huge number of parameters. If we simplify that to a single parameter, the motion would be too monotonous; so we intentionally leave all of them open so that animators can adjust them. This results in frequent communication such as, "How do we use this [parameter]?" and the like; [their jobs] feel tough to me.
If [animators] neglect this, they may end up using [the parameters] in a way the modelling team doesn't expect, only to be pointed out, "that's completely different from the settei".
Let's take Chi-Yu for example, a Pokémon based on a goldfish with magatama-like objects around its eyes. We made a gimmick where its eyes and magatamas shone gradually, but we needed to give the radiance of its eyes and the glowing of its magatamas separate timing; so we added [other] gimmicks with independently adjustable parameters. Things like those are prone to timing errors and therefore require close communication.
It's so complicated that there's even a design like "This facial expression must have no shine".


Spidops, an early-game [Pokémon], is another one with both a complicated body structure and movement, and developing it [required] a lot of discussions with modellers. It has eight legs like a spider, so its knee joints can only bend in a single direction.
Spidops' eight legs are all cylindrical, and it needs to bend its joints whilst maintaining a consistent [leg] shape; so we needed to coordinate this in detail with riggers and animators. There are even some occasions where it moves whilst two of its front-left, front-right, back-left and back-right legs are attached to each other, so it has an extremely limited range of motion. Moreover, it has no ankles — so the animator(s) in-charge had a hard time.
We humans have thighs, shanks, and feet; Spidops has thighs and shanks, but no feet on or parallel to the ground, which caused us a lot of trial and error to get its motion working properly.



Wakasugi: So, Pokémon's animation causes trouble that a human's doesn't, such as a narrower range of motion and less joints.
We often assign Pokémon full of restrictions to veterans among the animators.
Some Pokémon, such as Iron Thorns and Iron Bundle, have robotic or mechanical traits and therefore a limited range of motion as does Spidops, which gave us a lot of trouble.
We adjusted their range of interference and motion whilst following the settei; [those Pokémon] must've been difficult to operate. Still, the assigned animators [have achieved] good-looking movement after much trial and error.

[translator's note] A "range of interference" of an object appears to refer to the space occupied by the object that disallows clipping.

With those restrictions in place, we made progress while consulting modelers about how to portray [a Pokémon's] life-likeness. Iron Thorns and Iron Bundle have ankle joints with a narrow range of motion, which was troublesome; but above all, the fact that they have no knees was [the] hard[est].

Wakasugi: What do you mean by "no knees"?
They only have hip joints and ankle joints; there are no knee joints to begin with.

Wakasugi: That's a pain.
They must walk within their design; so it was a tough job to give them a viable range of motion.

Wakasugi: [Pokémon] purely [based on] living creatures keep the reference material as the source of your ideas, but those with an inorganic motif must depend on how inventive the animators are.
Varoom was another one of our tough jobs. It's based on a single-cylinder engine, and it has a lot of inorganic elements in its design. We needed to balance [its design] with its cartoonish, exaggerated movements as shown [on the settei]; so we found it difficult to share common ground about interpreting its movement.

Wakasugi: How many motions does a Pokémon have, and how long does it take to create them?
23 animations on average; and 60 was the most. Developing [animations of] a new Pokémon takes roughly one-to-two and a half months, including the supervision period. [The duration] quite varies, depending on the level of difficulty and the amount of animations.



We've [managed to] prevent its legs from clipping into its torso or the ground by tediously and repeatedly fine-tuning its joints across the body.


Carelessly moving the head downwards would make its swords clip into the ground, so there was a lot we had to watch out for. There was a lot of trial and error on how to give it cool-looking motion.

Developing animation not only on land, but also in air and water

Scarlet and Violet needs to portray Pokémon living in an open world with a wide range of topography, so there are more motions per Pokémon than in Sword and Shield.
Let's take jumping for instance: previously, [a Pokémon] only had the animation of jumping out of a Poké Ball in battle; this time, we needed jumping animations for the overworld. A simple copy-paste of an in-battle animation wouldn't suffice at all; detailed, dedicated animations were required, resulting in a big increase of the number [of animations].
Walking and running aren't all of the movement animations; some new types have been added, such as jumping and the like, or hovering. This made it more complex for us to set up the animation-state transition endpoints.
Some Pokémon not only travel on land but also swim or fly. For example, Glaceon requires [its motion assets] both on land and in water, and Bombirdier both on land and in the air. More and more motions have become mandatory in this way.

Wakasugi: So, some Pokémon having water or air animations is why the number of animations per Pokémon varies. In those cases, do they require not only two sets of movement animations but also two sets of attacking ones?
Yes. On top of that, we prepare two sets of emoting animations as well. For instance, since a flying [Pokémon] isn't grounded with its feet, dedicated adjustments [on these animations] are essential, even if they may partly be shared.

Wakasugi: Producing 3D assets for an open world seems arduous!
We really felt it was. (laughs)

Pokémon inhabiting both sea and sky, in addition to land, required a greater number of animations.




Wakasugi: Were there some cases where the amount of animations changed during development?
We estimated the amount of animations for each Pokémon in the early stage of development, but there were rises and falls as in-game events and the like came up. We really wanted as many Pokémon and as many animations as possible; so we tried to find common ground with GAME FREAK for feasible [development], with a comprehensive consideration of data capacity, man-hours, etc.

Wakasugi: Could you tell us what trouble a newcomer tends to have during Pokémon motion development?
This isn't specific to newcomers, but [animators] sometimes make a bipedal Pokémon, in particular, look too human-like in its movements and behavior. Even if its animation is nice, we'll ask them to correct it when it isn't faithful to the personality and the design of that Pokémon. Examples include a Pokémon showing dexterity when it has nothing like "It's dexterous" on its settei, or a Pokémon with its attacking animation showing great speed when it shouldn't be capable of moving that swiftly. These points seem like a hurdle for those unfamiliar with Pokémon.

Wakasugi: When you have as many as about 80 animators, you'll definitely have a lot of difficulties in communication. What did you pay attention to the most?
We didn't have direct communication with each animator of the partner companies, but we asked the lead [of the team] in each company to supervise their staff. We tried our best to keep in close touch with the leads.
We had a [private] project wiki dedicated to sharing the specifications. If we had used incorrect naming conventions on the wiki, for example, each company would've delivered the wrong data [to us]; so we were extremely careful.
The development team at our company [Creatures] is more than twice as big as that in Sword and Shield. We never had this scale of development before, and there were a lot of uncertain, experimental [workflows]; we know there's still considerable room for improvement. This time, [our workflow] greatly relied on the communication efforts of each member, which wasn't the optimal way for sure.
I also think that communication may have become more troublesome as more members worked from home, due to COVID.
To lower the barrier of communication, the leads of both the modelling and motion teams always tried to be like, "Feel free to ask questions at any time, no matter how trivial they may be". Actually, I was [very] busy with a lot of jobs, but still tried my best to empathetically answer questions from members in-charge.
I tried hard to always give an air of "I'm not busy at all". (laughs)
We also had a lot of information unsuitable for merely text-based communication, so we were flexible about holding some brief video conferences as well.

Creatures, Inc.'s Role: Expanding how Pokémon are expressed


Wakasugi: To conclude the interview, could you tell us what you find interesting about your jobs and what you consider a great [work] attitude?
More than 1,000 species of Pokémon have been discovered so far, all with their own unique personalities, which I'll never get bored of. Also, over the last few years, I've come to feel strongly about the importance of communication and the sharing of information with other departments [in the company]. When I was in a CG animation company, I'd [only] focus on thorough improvements for my own shots; nowadays, I consider things more often like "I wanna do this, then who should I ask about it?" and "If I do this, who'll have a harder time?" I believe an attitude like that can prevent mistakes and lengthy delays.
This isn't specific to the gaming industry, but it's quite important to communicate with [members assigned to] each process. You'll be able to create more accurate animations by, for example, listening better to modellers' [explanations] during kyara uchi [to understand] "Why are we setting a joint here?", "What's this [rigging] weight for?", etc.
I've been focusing on meticulous modelling [work] since my days at a CG animation company, and I find this to be an important attitude [that's relevant to] working in the gaming industry, too. I want to uphold my meticulous work, [all the more so] because Pokémon's assets in particular will be used for a long time across the series.

Wakasugi: Any advice to readers learning game animation?
We've been making 3D assets for Pokémon, but you shouldn't stick with [just] them alone; you should observe humans, animals, and various objects, and practice animating their movements. Through observation and production, you can improve fundamentals, such as [understanding] how the center of gravity shifts.
When watching demo reels, I/we pay attention to how the center of gravity shifts. I believe only when you're successful [in animating it] will your personality and commitment begin to shine through. Demo reels with the energy of "I love this!" bring me/us joy, and show me/us [the creators'] desire to grow.

Wakasugi: Indeed, the shift of the center of gravity is important in Pokémon animation as well.

[translator's note] Wakasugi appears to have comparisons with 3D animation films in mind.

This may come as a surprise, but most of those who only love Pokémon or video games won't suit Creatures' culture. As we take charge of creating Pokémon's 3D assets, we're aware our role is to expand in-game expressions of the Pokémon. To achieve that, you should take an interest in various topics, not limited to just the fields of entertainment.

Wakasugi: Now that you mention it, there's bread, there's ironsand, there's a single-cylinder engine, etc. the concepts for Pokémon are limitless.
Let's say you have a meeting with GAME FREAK. [You'll] glean information from their settei and instructions— share interpretations by drawing comparisons to something else that you have some background knowledge on; that'll be an instant done-deal. Then, when coming back to your desk, communicate the meeting to the animators in charge in a way that they can best understand, and share the interpretations again without discrepancies; this'll make your jobs more fascinating.

Wakasugi: Your stories have made me more aware of how interesting Pokémon animation can be. Thank you so much!

References