Well I don't have any particular real expertise here either, I don't think anyway, and I'm always kinda nervous/over-enthusiastic, sometimes find it hard to know whether people are vibing or if I'm just off on my own trajectory lol. Then add to that I never want to Tav'splain, or be that dude ya know hehe. But I can try! (Oh goodness, this will be such a digression, but maybe it's interesting? heheh)
My whole theory on this stuff is that, when we have a solid range, or enough-of-a-range, such that people can effectively see themselves in the magic mirror or create their own face there (faces they'd recognize as familiar to their own/close enough or with that same sense of beauty or scope reflected back), then people are satisfied that they have 'lots of options.' When the range is more limited however, or we see notable gaps, then it becomes frustrating like 'why aren't there more options?' or 'where am I, and the things I like, in all this?'
I think a fruitful comparison would be how Zelda and the Hylians of Hyrule are visualized in TotK, compared to say Elves or Half-Elves in BG3. Elves could then serve as our baseline, cause you need that right. You need to know how big the eyes are normally say, to then key off that, to get our standard range.
Now if we just took princess Zelda and dropped her into Faerun in BG3, she's going to appear a lot younger here right? This is because there is overlap between the conventions. For example, Loomis notes that larger eyes and the unstressed jaw are markers for drawing the female head, and also the proportionality of the legs to the torso in the body keying off the 8 heads tall stuff. Now we switch to our Japanese drawing board, and we see the same really, but the standard conventions have even larger eyes, smaller noses, a jaw that's even less stressed etc for the baseline, and bodies that stand a full head taller in the highly stylized/idealized frame. I don't mean so much the super jacked up or extra lithe Fist of the North Star physiques, but say in instances where to American eyes, you might need to see these figures dressed up, so the apparel can do some of that work for us in translation.
So if it's a gal, is she rocking sailer moon bows and tights? or if it's a dude is he sporting a gee that's like bursting at the seams with pectoral muscles? In either case the heads might be smaller than what's familiar but you got a way to key all that compared to the adolescent versions of the same. It's actually kinda cool if these two games go head to head, since we get to see both visual traditions sort of meeting at the middle.
There are some major, what I guess I keep calling visual traditions or conventions, which existed before the modern camera was invented and which continue to be referenced even after we achieved mimesis in that way (via photography and a mass media-in-print capable of reproducing photographic imagery.) I think for example that we can compare etchings and woodcuts from England or America to Notan woodblocks from Japan, and see what sort of interplay was going on in those print-traditions prior to the adoption of the camera by the whole world. Representational or figurative drawing had different norms and ideals in each case, and the whole project of teaching one to draw (whether optically or sculpturally) in a way that would be broadly recognizable for the intended public is in constant communication with what came before. Those prior conventions come back into play when establishing what gets a nod vs a gasp on the receiving end. When caricature is involved, it builds off those earlier conventions in a very pronounced way.
Aside, so when Bridgman taught his students (Loomis draws a lot from Bridgman and Howard Pyle), who then went on to define the norms for American illustration in the next century, he suggested not thinking about the human figure as a human being, but instead thinking about simple forms. He actually suggests thinking about the human body as if we were trying to draw an insect.
Specifically a hornet or a wasp or an ant. You know, the sort of thing that has a "Head, Thorax, Abdomen and Limbs" arranged in a basic sort of way with a set proportionality and relative scale, with reference to itself via building blocks of that sort. Then he applies the same sort of framing to the human body using similar conceptual anchors. Head, Torso, Pelvis, Limbs, and trying to think about the spine that connects them to give us gesture and the sense of movement. The last mainly to help us see when things are disjointed and not connecting in a way that carries. Or with the head, how to compose that out of what are essentially a bunch of boxes and cylinders and spheres using relative distances, to make a face out of that, or more specifically, an appealing face that registers as human, rather than a monster or a goblin of whatever stripe.
The project of seeing, and reproducing what's seen, is to then simplify what's being looked at (in 'reality' or on the page) and to omit extraneous information so the anchor points are clear. This gives the 'reality' which underpins the image, and this can then be exaggerated for particular effects like mood or emotional resonance, or callbacks to an earlier tradition/visual milieu.
Since these ideas were exported via things like book covers and film posters, later animated features and cartoons or other popular media, there was a real preponderance of the one tradition over all others for a long while. As a counterpoint, beginning in the second half of the 20th century, and massively in the 1980s with the rise of arcade games and then video games and manga, the currents shifted direction a bit and we begin to see how some of those ideas were re-interpreted out of another set of visual conventions. This time one with its source more in the Indo-Pacific than Western Europe.
This style we tend to call manga for shorthand. In Japanese ?? gives a term that means something like 'comic' or 'cartoon' and sort of basic in that way - the symbol for 'Man' meaning something like 'spontaneous' and Ga meaning 'image' - so quick, full of verve, meant for popular distribution. An American comic book with images by Kirby with the crackling super heroes going "Zip Bang Pow Kablammo!" in the hands of a Japanese school kid, would also be manga. They might say it's American manga, but anyway, like that - cartooning at its core - but then pushed towards figurative realism in various ways, sometimes more extreme or less extreme, to achieve something sympatico in the style.
Now add into the mix an additional complicating factor of the gendered gaze imposing a set of beauty conventions on top of all that. You will note this I'm sure, how even though the issue has been raised many times, that the conventions for females are highly charged and often less flexible. This would be the oval face effect, which we seen in the balloon aesthetic in 3d animation. At its most crass and banal, men get like a dozen buckets in many fun shapes, whereas women only get 1 very round bucket.
You can extend it beyond gender into other arenas, but that one is easy, because it's so clear when watching any popular toon these days.
So how does any of this relate to the sort of Motion Captured and 3d modelled figures and faces we see in BG3?
Well for starters we're missing some stuff at the base, and the faces are gated by fantasy race. Even among the more expansive options available to Elves (or Humans), even if you can change colors or hairstyles, we don't have an option to say lower the bridge of the nose to see that same face with an epicanthic fold to the eyes, or a restructured jawline. I'd say out of a couple dozen options, we only maybe get what, one or two like that? And they are hidden as an Elf exclusive.
That's a bit of an issue, not just for the Halfling models, since doing that sort of ignores fully one third of humanity on planet earth. But going specifically to the Halfling male head models... just look at them together and tell me those dudes don't all look like they could be from the same family hamlet in middle England hehe. Now if we get one of those mod unlockers that make all faces available to all races, we start to get a real spread, sure. But then you also see where, absent the caricature, the halflings start to become indistinguishable from humans outside of the body phenotypes, and their headshots have to rely on a different posing or you'd never be able to tell who's who from the portraits.
This goes to the earlier concept we kicked around about maintaining the illusion of choice and variety, instead of giving us the actual variety that would be needed, which as I've said time and again, is right around 100 preset heads for each gender archetype. Not 6, not 8, but 100!
Are we gonna get that? Well, no, I can probably say with some certainty we won't, but still they could do a little bit to get us further along I think.
I don't know if that answered a question or if I'm just rambling, but I think they need to start back at square one on the Halflings, and that's going to be tough. They're not going to want to do that right, since it's a lot of work, but I think the payoff would be worth it and they don't need to remove what already exists, they just need to keep piling on until we get a spread that's large enough, with sufficient variety, that some of those issues recede into the background a bit more. Right now they are totally front and center, cause we've only got half a dozen, if that makes sense.
ps. In my perfect Character creator there are a total of 256 'faces.' This is so it can mirror the full color Hex palette in terms of scope. That would give say 100 for each gendered archetype, and another 56 for a blend at the median. Also I imagine people saying things like "I want sliders! Why can't there be sliders here!!!?' But it's like ok, sliding from what to what, exactly? I mean right? hehe. Cause really all sliders amount to are a series of presets that we can cycle through quickly, but they can result in many mismatches, or the rubik's cube fails and that just time sinks it into frustration. Anyway we can get almost the same thing going on with a series of completed head/face presets, provided there are enough of them. And those can be more readily animated, since they'd be real faces as the basis. For casting you'd probably need many more models, and not all might be selected for the full spread. The idea is to build up that library though, find the keynote swatches that would give us the requisite full palette of faces. Then morph those faces into the various fantasy races, in all the various ways - with each Fantasy race having all 256 options.
The rest we handle through makeup and styling. I've never seen it done before. Not in a D&D cRPG anyway. I'm convinced the first studio to really pull it off, would just live forever as the gold standard high bar in a Char creator. But start small I guess, some more halfling heads for now hehe. Also to the 7 heads thing, that would be accurate these days. In life drawing and also sort of as a response to the earlier dominant conventions which started at 7.5. Most people are closer to 7 heads tall in reality, and so when you draw at 7 heads these days it suggests a character more down-to-Earth, as it were. It's good advice for drawing in the current times I think, the 7.
pps. A Hexadecimal code for the Human Face?
I think that's it in a nutshell, I was trying to summarize. Obviously we'd get a somewhat idealized version of that first, cause it's actors right, but where we (the player) are like the casting director of our character/party. But like the most egalitarian broad ranging take on that possible, so we can really cover a lot of ground. They have studios everywhere now right? Once the face palette is in place, and the headshots are all in rows and columns, then they can iterate off that by sending us along to the next phase: make-up and wardrobe.
This where we get to play out of own mini Fangoria Ren Fair with the dress up. If we get sliders they can come in there too right, basically copying a theatrical production for the step by step on that. What they are already doing behind the scenes more or less, but making that into part of the gameplay. That's the idea I think, that I can't ever help returning too, cause I've been after it since I was like 5 years old heheh. If they could make that, they'd have it on lock for ages and ages - so much replay!
3ps. Here are the defaults for the stronghearts, just to go along with the male halfling heads on the previous page. They are the same as we'd see in char creation on the edit appearance zoom, just set 'em all to bard to see the necks. The topnot is alright cause you can see where the crown is basically. This is the idealized 3/4 view in the nice lighting, we could swing em around but basically gives the gist. The layout in the current UI seems like it would slot nicely with 8 heads rather than 6 to complete the bottom row on that. This is just the six screens pasted one on top of the next and reduced a little. Sometimes it's helpful to see them all together. I think they could actually do a lot from this point in the hair/make-up department. Once you're at this level of caricature, things like sliders for brows and noses or lips would probably work well, or whatever other morphs a fantasy race might have going on like the longer ears for elves and gnomes, the horns and scales for tiefs and dragonborn, that sort of deal. Unique hairstyles and jewels based on the fantasy phenotype could be cool there as well. I don't know if much changed compared to EA halfling options, probably just my graphics card is slightly better now and the ultra cleaned up on the hair a bit, but I mean for the core of what's there, still pretty much the same right? That's the 6 swatch version, we need like 128.