《騎馬與砍殺2：霸主》開發者日誌（4） ：動作捕捉

虛骨骼

Happy Friday to avid readers and passers by. It has been a little while since our last Bannerlord Blog. In this entry we are once more decreasing the number of unblogged rooms in the office by one. Although some animations are made and polished in various parts of the office, many are captured in our very own motion capture studio near the main door of our offices. The animation hub and thus the associated blog room for this episode.

怨念的讀者們以及只是路過的，週五快樂。距離上一篇騎砍2的部落格是有一段時日了。在這兒我們再一次把沒說到的房間一個個補完。雖然一些動畫在工作室的許多部門被製作和打磨出來，許多動畫都是在挨著我們工作室主門的我們自己的動捕棚裡捕捉的。動畫是本期部落格裡所主要介紹的。


The movements that characters and animals (neigh!) make in the game all have to be defined by animations. In most games, these animations have one requirement, which is to look good. In this sense, good normally means realistic, or at the very least, believable. If you don't notice anything weird, then the animations are normally pretty decent. Naturally, Mount&Blade also has that requirement, but on top of that, due to the mounted and melee combat systems, many of the animations are also inherently linked to gameplay. All of our hits are calculated precisely, so if, for example, a sword swing animation shifts the shoulder forward an inch, then swords will have an inch more range, (which will have an effect on the calculation). That is obviously something that will affect gameplay directly and requires extra care, as well as constant reworking, in order to get right. This isn't a factor that comes into play when making animations for firing guns or combat systems that don't use physics based hit detection.

遊戲中的人物和動物的動作全部由動畫定義。在多數遊戲裡，這些動畫都有一個要求，那就是好看。在這個意義上，“好”通常意味著真實，至少可信。如果你沒發現有什麼奇怪的東西混進去，那就基本挺好的了。 自然地，騎砍也有這種要求，但是在這種需求上，由於騎乘和肉搏戰鬥系統，許多動畫也固然要與可玩性掛鉤。 所有的擊打都是經過精確計算出來的，比如說如果揮劍時肩部前移了一寸，那劍的攻擊範圍也會增大一寸（這會影響計算）。這明顯是影響可玩性並需要格外注重的東西，精益求精，以求正確。這和那些不需要考慮物理接觸打擊的槍戰遊戲是兩回事。

We have largely aesthetic animations too, even in combat. The animation for reloading a crossbow isn't going to have a dramatic effect on gameplay - perhaps how low the character bends will affect what obstacles cover them while reloading. The speed of the animation is obviously a factor there but can easily be changed by swapping a few numbers around. What is crucial is that it looks like an actual crossbow being reloaded so that when you see it in the game, you hardly notice it. If, instead, it looks more like the character's bones are made of jelly and they are trying to eat the dirt on the ground, then it will probably have a negative effect on your experience when playing the game - after some initial amusement.

對動畫和戰鬥，我們也有審美方面的考察。給弩上矢的動畫並不會給遊戲性上帶來什麼激動人心的效果，也許角色彎腰到多低會影響他們裝填時候有什麼障礙物能遮擋他們。 動畫的速度是一個明顯的因素，但我們也可以通過幾個引數來改變它。關鍵是要讓這個動作看起來像是真的在裝填弩，好讓你在遊戲裡不會發現瑕疵。相反，如果這個動作讓人物像是果凍做的骨頭而且在啃地上的泥，你初玩遊戲的時候會有很負面的體驗。


To make these animations we use our motion capture studio as much as possible. The process of motion capture basically involves putting on a rather silly looking suit with little white bobbles ("markers", for the tech-head) on it and making the desired movements in the middle of a host of special cameras (18 in our set up). The cameras track the white bobbles and record the way they move into a computer file. Making animations with motion capture saves a lot of time and painstaking work moving the bones and joints of the virtual skeleton (this is more like a human-ish stick man than an undead in a fantasy game). Simply capturing this in realtime saves a lot of this work, perhaps leaving only some cleaning before the animation is ready for the game. It is also likely to produce a more realistic result as motion capture picks up on more subtle natural movements that are harder to recreate by hand. It is especially useful for idle animations (standing or sitting doing nothing!).

為了做這些動畫，我們儘可能的多用了動作捕捉。流程大體上是把一些白球綁在傻乎乎的衣服上（專業名詞是“標記”），然後在一組特殊的相機中（18個）做想要的動作。相機追蹤白球並記錄其運動軌跡。用動作捕捉做動畫節省了許多時間，和移動虛擬骨骼上的骨頭和關節的繁重工作（骨架更像人而不是像奇幻遊戲裡的不死族了）。實時捕捉節省了大量的工作，在動畫徹底完成之前只剩一些清理工作了。在拾取更微妙而自然的，無法手動調製出來的動作的時候，動捕會製作出一個更逼真的效果。在做空閒動畫（啥都不幹的坐著或者站著）的時候尤其有用。

Certain animations are more difficult to capture in our studio. For example our horses are animated by hand since, while it is possible to motion capture a horse, bringing it into the studio as an unexpected variable would disturb our more nervous programmers. Also motion capturing a horse in combat might present some animal rights issues. Charging a horse into a wall would have to be done by hand in any case.

某些動作在我們的工作室裡很難捕捉。比如馬的動作還是手動的，雖然捕捉馬的動作沒啥問題，但是把它弄進我們的工作室會增加不確定的變數，比如打擾到我們緊張兮兮的程式猿們。另外捕捉馬在戰鬥中的動作還涉及動物權益的問題。不管怎樣，騎馬衝進牆裡這種事情還是靠手工吧。


As you can see in the screenshot, the taverns in Bannerlord already have a lot more atmosphere than those in Warband. People are sitting, drinking, stretching and talking just like a real medieval tavern. Though this is still a work in progress and some way from where we want them to be. Our animators are constantly working on making Calradia come to life in order to give the player a vivid game experience while paying attention to detail and life-like movements.

如同你在截圖中所見，騎砍2的酒館比戰團更有範了，人們能坐下、喝酒、伸懶腰、說話，就像在真的中世紀酒館裡一樣。儘管這仍是一個工程圖，但這就是我們想要做成的樣子。為了讓玩家有更生動的遊戲體驗，通過集中精力製作細緻、逼近現實生活的動作，我們的動畫師努力讓卡拉迪亞躍然於遊戲上。

面部表情

We get messages from a lot of players saying they want to see more immersion and believability in the game. One feature we have planned for Bannerlord is facial animations.

我們從很多玩家得到資訊說，他們希望在遊戲中看到更多的真實感。我們計劃在Bannerlord新增一個面部特徵動畫。

For a very long time, faces in games were pretty much completely motionless. There are various studies about the importance of non-verbal communication (facial expressions, body language etc.) in communication as a whole, including the much debated and often quoted “7%, 38%, 55%” figure for words spoken, tone of voice and physical behaviour respectively. We aren't psychologists though and that isn't a debate we need to enter into. The general consensus is that they are pretty important.

在相當長的時間，臉在遊戲中幾乎完全不動。現在將非語言溝通（面部表情，肢體語言等）的交流作為一個重要的研究專案，其中廣泛討論“7％，38％，55％，”這些個經常引用的常用語，語音語調和身體的行為。雖然我們不是心理學家，但是我們必須進行辯論，以達成普遍的共識，這是非常重要的。


The results in game are already fairly astounding, bringing characters to life as they squirm and grin accordingly during conversations. They won't just be visible in conversations either; battles, town scenes, feasts, prisons and everywhere will all feature animated faces. The battles are an area of the game where we think the animations are going add a lot of fun and engross the player in the action. Not only hearing but seeing soldiers cry out in fury or pain, as the chaos of the fight ensues, contributes to an altogether more engaging and thrilling experience. This is a serious step forward from previous games in the Mount&Blade series which featured completely expressionless faces. It isn't the end of the world and it may matter to some players more than others but in a game where you have marriage, it's nice to see some emotion on the face of your betrothed, rather than that same dull-eyed look she's been giving you for the last 3 poems you read to her.

效果在遊戲中已經是相當驚人的，使人物的生活中，因為他們侷促不安，以及與談話相適應的笑容。他們將不只是有形的談話，戰鬥，城鎮場面，節日，監獄，無處不在的面部特徵動畫。戰鬥只是遊戲一個領域，我們認為動畫會增添了不少的樂趣，玩家在遊戲中不僅能聽到，還能看到戰士在混戰中大聲呼喊的憤怒和痛苦，這將更加引人入勝，帶來驚險刺激的體驗。這相對以前的騎馬與砍殺系列遊戲中完全沒有表情臉的，是非常重要的進步。這不是世界的盡頭，在遊戲中，你如果結婚，將會看到高興的表情在你的未婚妻的臉上，而不是同一個呆板的眼神從頭到尾，看著你給她讀詩。


On the technical side of things, all of the facial animations are made by hand. The face is bound to a skeleton (similar to the one used for the full body, as described in the previous blog) that allows different parts of the face to be stretched and squashed naturally into different forms. For each fixed form of the face we use blendshapes or morph targets to control the movements which are compatible with whatever kind of face you make. When it all comes together and a character has a believable, interesting reaction during a conversation it really creates some special moments that we think players are going to respond positively to and enjoy.

在相關的技術方面，所有的臉部動畫都是手工製作。臉部被繫結到一個骨架（類似於用於全身，如在以前的部落格中所述），它允許被拉伸和擠壓自然成不同臉的不同部分。對於臉部的每一個固定的形式，我們使用blendshapes或變形目標來控制它們，讓臉與你作出的任何型別的動作相容。當這一切都在一起，角色在對話過程中果然產生了一些真實的反應，我們認為玩家們會積極響應和享受。


Although this is something we have been asked about specifically, we know there are a lot of messages asking about gameplay mechanics and combat as well. We will proceed with talking about everything in the game in time and not mentioning them shouldn't be taken to mean that we don't care about them or aren't working on them! In fact quite the opposite. It's because they are receiving so much attention and are liable to change based on design decisions, through testing and iteration, that we are cautious of talking about something that ends up changing by the time our game is available to play.

Hopefully you enjoyed reading another instalment of the blog and finding out a little bit moreabout a little bit of our development on Mount&Blade II: Bannlerlord. Stay tuned for future updates by liking us on Facebook and following us on Twitter!

Join the conversation and comment on the forums! (314 comments)

翻譯：Iccgreatwall、QT
校對：QT
來源：騎馬與砍殺中文站