I’m as surprised as you are that it’s taken me so long to talk about a subject near and dear to my heart in this column. No, not dinosaurs—although I’m surprised they haven’t been brought up yet—but human mammary glands. Everybody’s favorite source of fatty adipose tissue has been a staple interest of mine since, according to my parents, the tender age of six. I have watched with great interest, and occasional disappointment, how these organs have been portrayed in video games from the NES to the modern day. It really is fascinating to look at the history. In terms of the portrayal of breast physics, we’ve more or less been in a stasis period that originated on the PS2, but a game released late last year took an important step forward. But before we can identify the progress, we should probably discuss the problems.

As long as there have been busty women in video games, developers have tried to realistically rig those busts. It would be unrealistic for a woman as chesty as Lara Croft to scamper and leap around ancient ruins without her considerable assets bouncing and swaying with her. Of course, the PSOne couldn’t pull this off: Lara’s hilariously primitive character model simply wasn’t able to feature that kind of movement. Sprite-based games fared a little better, though. Sprite artists could depict individual frames of animation however they wanted, thus we’re left with Mai Shiranui’s infamous bouncy idle animation, which was brought back in HD for The King of Fighters XIII (a game I somehow don’t own). Aside from Mai, however, I can’t think of any other sprite-based female characters with quite so much jiggle. The real strides were being made in 3D modeling.

To be clear, the underlying problems remained the same: while breasts were now modeled separately, they were still a core piece of the underlying character model. Tricks to create jiggle or “breast physics” were little more than selecting a point on the breast (usually and unfortunately the theoretical placement of the nipple) and timing its movement a few frames behind the rest of the character model. This gave the illusion of bounce from a practical standpoint, but you were just watching a piece of the texture map move around. Examples of this are in King of Fighters: Maximum Impact, its sequel, Soul Calibur 2, Metal Gear Solid 3, and Dead or Alive 2: Hardcore. In that last game, increasing the player’s age in the Options menu causes further delay of the rigged movement. Nonetheless, 3D models still weren’t at the level of 2D Mai Shiranui. In a few of these games, you can actually watch the overlying texture map stretch with the delay. It’s fairly painful to watch now.

It's pretty clear that Mai's "sideboob" here is just a texture map applied to a flat surface.

Even a game as “advanced” as Dead or Alive: Extreme Beach Volleyball had the same problem, though it revealed two other limitations of the technique that, upon retrospect, became obvious in all games of that generation. First, the breasts could only be moved so much, because you’re essentially stretching one aspect of the character model. Stretching too much would lead to collision with another part of the body, such as the arm (this actually does happen in a few games). Additionally, both breasts followed the same stagger path. The second limitation is much more damning and has been much tougher to shake: costumes applied to the characters were merely different texture maps. Now this may sound obvious, but think about it: the same character model does not change whether Tina Armstrong is wearing her cowboy-themed bikini or a string bikini. Clothing has no effect on the body or the breasts, which is patently ridiculous, but is a limitation of the technology. For example, changing Lara’s outfit in Tomb Raider: Legend has no effect on her character model and any cleavage you might see (e.g. when she wears her cocktail dress) is a part of the new texture map. Lara should be wearing a restrictive bra when in her more traditional outfit, so her breasts should be shoved together and possibly up, but they’re not.

Lara's showing a lot of painted-on skin here, and you can pretty much tell, based on the color differences between her dress straps.

The technique of only changing a model’s texture to simulate a change of clothing is also the reason modders are able to create “nude patches”: they just apply a skin-tone texture map to the character model. Character models are already “nude.” What you see is analogous to the Sports Illustrated “body paint” bikinis. Kate Upton’s not wearing any clothes, just paint. It’s the same with the characters in Tomb Raider or Dead or Alive Xtreme.

This was the standard for the PS2/GameCube/Xbox generation.

With a new generation of technology, one would hope that this situation would change. It sort of did, but not for the better. The early examples were both on the Xbox 360: Dead or Alive 4 and Dead or Alive Xtreme 2. In the former, the “same character model, different skins” approach remained largely the same. However, some progress was made in terms of realistic clothing. Christie’s open jacket, for example, did not adhere to her character model below her breasts, although it didn’t move independently. Exaggerated cleavage on some characters, however, was clearly a texture map. The clothing was not a separate element and did not affect the characters’ bosoms. Dead or Alive Xtreme 2 didn’t try to solve this problem, instead went completely in the other direction. Developers still relied on applying different skins to the same model, but thanks to improved technology, they went all-out on stagger animations, giving the ladies’ breasts individual stagger physics and allowing more stretch. However, this resulted in a new problem: breasts bounced, swayed and rotated on a central point even in a calm breeze. It’s like these people have never seen how large breasts work. Large breasts are heavy and don’t actually move all that much. Even characters with smaller breasts like Kokoro or Hitomi bounce to the same ridiculous degree and that looks even MORE unnatural.

Poor Kokoro's breasts are having a disagreement about which way to go. And she's just sitting there!

I mean, I could go into a whole other diatribe about breasts as physical objects and how different anatomical modifiers would affect their movement, but that’s probably not fit for this particular website. I’m probably going off the rails already.

Crystal Dynamics took a weird approach to Lara Croft in Tomb Raider: Underworld in that they removed all pretenses of bounce and basically went back to the unified character model of old. This is nice in a way, because it’s not distracting and is fairly realistic if Lara is wearing a sufficiently supportive bra. It also implies that the developer just didn’t want to bother with this physics problem. I suspect that’s the reason, because they didn’t fix the physics when they later released bikini DLC. It is weird to see a fully unsupported, bikini-clad Lara jumping and climbing without any motion to the girls whatsoever.

My argument falls apart if we assume that Lara has fake breasts, which really do just kind of adhere to the chest and don't move much.

For the most part, developers have tended to stick with texture maps as clothing (and cleavage). One interesting exception is Soul Calibur 4. Ivy and Taki’s standard outfits are customized character models, and the character model switches to a generic female model when you flip to a different costume or try to modify the costume in the character creator.

Now there is a game that has upended the tea table. It comes from those Dead or Alive guys—again acting as pioneers in this field—and it came out last September. Dead or Alive 5’s character models are affected by clothing. Now, there are some instances of different texture maps applied to the same character model, but only in instances where the clothing is essentially similar. You can really see the advances in “breast physics” technology when you go from bikinis to, for want of a better term, dominatrix or leather-based outfits. It’s obvious in some other costumes, too. Under the traditional texture-swap model, Kasumi in a bikini would have the same character model as Kasumi in any other outfit. Here, Kasumi in a bikini has breasts that are separated. This is especially clear when you lose a fight and she’s resting on the ground; she doesn’t have cleavage. However, when you put Kasumi in her leather-based ninja outfit, her breasts are squished together by the outfit. Suddenly, she has cleavage again. Similarly, Helena’s swimsuit shows off the unsupported hang of her breasts and the space between them, while her opera outfit shoves her breasts up and together. They even move differently in these two outfits!

In an unrestricted top, Helena's breasts fall realistically and are apart from each other.

But when you put our favorite French opera singer in a bodice, she's got heaving cleavage. Wonderful!

The most impressive realism in terms of movement is displayed by Tina. If you put Tina in her cowgirl bikini costume, start a training fight and just let her go into an idle animation, she starts hopping back and forth from one foot to the other and her breasts jump up and down, threatening to spill out of her bikini top. They don’t follow her character model. There is no rigging delay. They’re doing their own thing—realistically, I might add—wholly apart from what her body is doing. And this bouncing changes if she’s wearing a more restrictive outfit, such as her yellow top. This is a HUGE step forward in breast animation. As somebody who’s always had an eye on this critical aspect of animation, I’m impressed that Tecmo has managed to pull this off and hope other developers follow suit. 

You win, Tina... although you should probably be worried about a wardrobe malfunction.

There are still advances I'd like to see, though. It would be nice if smaller breasts moved LESS than large breasts on a consistent basis. They would also move differently. Small breasts have a tendency not to sag against the chest, remaining more firmly anchored to the pectoralis major muscle. Large breasts, on the other hand, do sag or droop against the same muscle, so their base of movement is different and their motion more pendulous. For all the advancement DoA5 has made, Tina's breasts—which should be sagging against her chest—are attached all the way around, which looks unnatural and influences their animation. So that's the next step, developers! Go forth and INNOVATE. On my behalf.