2012-02-29, 週三 16:50
上層分類: 動畫製作
分類: 動畫與 Rigging
作者 Yoyo 師堯

專訪Rigger Josh Carey - 專案篇

感謝Way推薦給我資深rigger Josh Carey專訪文章，後來間接又找到作者另一篇專訪，由於篇幅很長，但是內容又都很豐富，所以我就擅自把它主題式的拆成兩篇：一篇是關於短片Looney Tunes - Coyote Falls訪談內容，另一篇是Josh Carey談他個人職涯與展望。由於這支短片是華納公司第一次把它的經典卡通由2D轉為3D的嘗試，有其重要意義，撇開成本、故事、編導來看，我覺得技術方面是非常成功的，後來片子也得到了第38屆安妮獎的提名。主要角色是逼逼鳥RoadRunner和大野狼Coyote，先來回顧一下正片吧！

精彩的Making of

I saw some of the impressive rigs on your website. How do you usually start out with these character?

我在你的網站上看到一些令人印象深刻的rig，你是怎麼開始rig這些角色的？

A lot of what we've done in the past at Reel FX is take other companies IP and make commercials or shorts with them. This means that we need to adapt and be able to imitate/replicate everything they've done with their rigs (and animation, rendering, etc). In these instances, we generally absorb as much footage of their movies as we can, trying to pinpoint joint locations, volume preservation tough-spots, key things that need to move during animation, etc.

我們在ReelFX曾經做過許多直接承接客戶IP或是廣告和電影短片的案子，這意謂著我們要做得和他們本來的一模一樣，包括：rig、animation、rendering所有部分。這時候，我們就會儘可能地觀察原始素材，像是客戶製作的電影，試著去精確地指出角色的骨骼位置，保持身體特定部位的體積變化，找到動作時需要移動的關鍵部位。

As for the Looney Tunes shorts, we didn't have anyone to imitate other than the classic cartoons. Animators did an awesome job at grabbing frames from the cartoons and showing us specifically what poses they wanted to be able to recreate. We used the official WB character sheets to figure out the on-model poses that these 2D characters needed to hit in 3D. Often times when starting out, the first thing i'll do is take a pencil to a printout and draw my skeleton over the character and indicate any notes, special cases, or animator requests.

至於「兔巴哥」(Looney Tunes)這個案子，我們就是想辦法重現經典卡通就對了。Animator做了很多功課，像是從舊卡通中抓出各種希望能夠做到的pose。我們用華納官方的角色設定稿(character sheets)去想像如何把2D的角色擺進3D裡。每當一個案子要開始時，我會做的第一件事就是拿出一枝鉛筆，然後在印出來的角色圖上畫出骨骼位置，並且記下所有心得、特殊需求和animator的各種需要。

Chuck Jones had a big influence on the animation history. How did you feel about renewing the old Looney Toons?

Chuck Jones對動畫史有很大的影響，你對於重新詮釋「兔巴哥」(Looney Tunes)這樣的經典有什麼看法？

First honored and excited, then a little scared. We did not want to have any part of something that did not pay homage to Chuck. We said at the start that we would go all out to make sure we nailed the classic chuck style, while still going for the semi-real look that WB was asking for. Even the models themselves went through many rounds of approvals until we were happy that they were close enough to the original designs, and that the rig could take the model the rest of the way based on the poses in animation. We did not want tiny toons models or anything remotely off-model from the classics, and we fought hard to push that through.

一開始覺得很興奮與榮幸，接著有一點恐懼。我們完全不想失去Chuck動畫的精隨，所以我們一開始就設定好要極盡可能的追求Chuck的經典風格，不過我們同時也要兼顧華納的要求，一種半寫實(semi-real)風格。因此在model部分經過了很多輪的修正改進才達到我們貼近原始設計的要求，我們依據動畫裡的pose作為model過關進入rig的標準，我們很努力地想要達到這樣的目標，不照抄原本的設計，但也不希望有任何一點偏離經典。

One effect that I really liked is when the Coyote moves really fast, and you can see three arms at once. How was that done?

影片裡有一個我很喜歡的效果，就是當Coyote跑很快時，你會同時看到三隻手，那是怎麼做到的？

Multiples! We basically had parts of the original rigs, with transparency controls, available to the animators to bring in multiple times per shot. Our animation TD wrote a tool to help animators deal with the multiples that we included from these rigs. Animator patience obviously comes into full effect here again as they were basically animating on 1's at this point, as well as a major props to the Lighters that had to deal with the crazy amount of multiples that the animators were adding to their shots. We gave the control to the animators to specify how much of the limb needed to be shown - whether it was a Road Runner head where only the head (no neck) was shown, or a fully faded coyote arm - animators could dial that in. That info was then passed down the pipe into lighting/rendering.

很多組的rig！基本上我們有原本做好的rig，然後加上透明度(transparency)的控制，就可以讓animator在一個shot裡用上很多組同樣的rig。我們的animation TD寫了一個工具可以幫助animator處理這樣的問題。animator的耐心顯然再次發揮巨大功效，不過他們只要做一次就夠了，但是可苦了lighter，他們變成一個鏡頭要處裡很大量的rig。我們將控制權交到animator手上，他們可以設定手臂部位的顯示程度，或者是Road Runner只要出現頭而已，然後把脖子以下都隱藏起來，或是一整隻半透明的coyote手臂，animator可以綜合使用這些功能，然後將這些資訊傳遞到下個工作流程：打燈和算圖(lighting/rendering)。

How long did it take you to create a character for the Looney Tunes shorts?

在「兔巴哥」短片裡，你們花多久時間做一隻角色的rig？

We had an initial rig out pretty quickly (couple weeks), but it went through many many iterations of fixes and changes. The key was actually getting the model as close to 'on model' as possible (according to the 2D character sheets), and that took time. We were making tweaks to the rigs all the way to the end, and we split up the rigging between body and face to 2 people, which helps the process go faster. I can't pinpoint an exact number for you, but I would approximate that we spent somewhere between 4 and 8 weeks on them. Possibly more :)

我們很快在幾周內就先做出最初版本的rig，之後就是很多很多的改進跟修正，關鍵是我們必須讓3D model更像2D設定稿上的樣子，這花了我們很多時間，導致一直到案子快結束時我們都還在做微調。因此為了加快工作效率，我們還將角色的rigging工作拆成兩部分：身體和臉部，分別給兩個rigger負責。我沒辦法確切告訴你我們花了多少時間，但我估計大概是4~8周或更多。

Can you tell us what was your role and and duties you had during the Looney Tunes production?

可以告訴我們你在「兔八哥」案子裡擔任的角色跟責任嗎？

During projects as well as between projects, I'm the Department Supervisor for Rigging. This means that I globally supervise how the rigging dept runs, our workflow, our future R&D efforts, etc. It also means that I Supervise rigging on projects, which is what I did for the Looney Tunes shorts. We had a team of about 7 riggers that I supervised. Basically, I had to make sure that our rigs could do all the crazy things that our animation director wanted them to do. For the coyote and road runner shorts, I mostly handled the body rigs and figuring out how we would do our 'multiples' system. I say 'mostly', because a lot of our team touches a lot of different rigs at any point during a production. On the latest round of shorts, I again mostly focused on the bodies of Elmer and Daffy, but then focused on the faces for Sylvester, Tweety, and Granny.

我在這個案子期間擔任rigging部門的主管，負責整個rigging部門的營運，包括：制定工作流程、未來研發方向...etc，同時這也表示我主管所有案子的rigging，就像我在兔八哥短片裡擔任的角色一樣。我們團隊有七名rigger，基本上我必須確保我們的rig可以做到動畫導演想做到所有瘋狂動作，以這部的角色coyote和road runner為例，我大部分時間都在負責身體的rig，並且去規劃如何處理多樣化的rig系統。我說大部分是因為我們團隊在案子進行時常常會接觸到很多不同的rig，所以當短片進行到最後時，我又去負責Elmer和Daffy身體的rig，但之後又馬上接手Sylvester、Tweety和Granny臉部的rig。

At ReelFX are they using more joints or blend shape based face rigs?

在ReelFX以joint為基礎做臉部rig比較多，還是用blend shape為基礎比較多？

I would like to say that our face rigs are pretty much your standard rig these days. They have a pretty defined set of blend shapes that go on top of a pretty complex joint-based setup. The animators can use either set, but most of them use a little bit (or a lot) of both. The joint-based controls are generally more broad-based influences, with the ability to turn on 'micro movers' to get super refined deformations.

我必須說我們的臉部rig比一般絕大部分的rig要複雜許多。我們在相當複雜的joint基礎上再加上一組定義明確的blend shape。animator可以運用任一組合來做，但他們多數會兩種混著使用。以joint為基礎的控制器用來產生比較明顯的變形，至於要做非常精緻的變形就可以打開微控制器(micro movers)來完成。

Do the riggers also model the character themselves? Are riggers ever involved in modeling blend shapes (if the pipeline requires it?)

Rigger也需要做角色model嗎？Rigger需要參與model blend shape嗎？或是說工作流程上有這樣的需求嗎？

Personally, I've always made blend shapes myself (and enjoy doing it!), but at ReelFX, the modeling team handles both model and blend shape creation. We do have a few riggers that would be completely comfortable making blend shapes though. I would like to say that our modeling team and rigging team works very closely together, so by the time we get approved models, edge loops are how we need them and models are good to go.

以我個人來講，我總是自己做blend shape並且樂在其中，不過在ReelFX，我們的Model團隊會同時負責model和blend shape。我們有些rigger有時候也很樂意建立需要的blend shape。我想說的是，我們的model團隊和rigging團隊合作非常緊密，所以只要我們確認model的佈線和我們rig上的需求相符之後，model就等於過關了。

Did you work closely with animators when achieving such flexible rigs and was there a requirement on your behalf to understand concepts of animation?

為了要完成那些活靈活現的rig你是不是要跟animator保持密切的合作呢？還有你認為自己需要瞭解那些動畫原理嗎？

Absolutely! A lot of us had the original model sheets up at our desks as both inspiration and for goals of what we need to hit. The animators did a GREAT job going through footage of the old cartoons and pulling screen grabs of poses that the characters would have to hit. Bryan (the animation director) would always walk over and show me something insane and say "So... think we can do this?" It was a great collaboration between our teams. On this latest round of shorts, deadlines were a bit tighter because of other ongoing projects, and we still wanted to push the quality to be even better than the first set. Sometimes it was a bit painful, but in the end we had some great work to show.

當然！我們很多人桌上都會放著角色設定圖(model sheet)，這不但是為了讓我們在工作時激發靈感，另外就是提醒我們要達到圖上目標。Animator做了很多很好的功課，透過研究過去的卡通畫面，然後把角色的pose截圖下來當作目標。動畫指導Bryan總是會走過來然後給我看一些瘋狂的圖，然後問我「所以…這個我們辦得到嗎？」我們之間有很棒的合作。因為同時還在進行其他案子，使得這支短片在後來變得有點趕，但我們還是希望能把品質再提高，超過原本預設的標準。這麼做有時候會有一點辛苦，但是最後我們才會有精彩的成果得以呈現。

As far as a requirement to understand concepts of animation - yes! Every rigger must know those concepts. Most of our team animated at some point, whether it was in school or on the job. Either way, we all have a love for animation, so we do want to make the best and easiest to use rigs for the animators.

回到正題，就rigger需不需要瞭解動畫原理這件事來說，答案是，需要！每一個rigger都需要瞭解那些原理。我們團隊多數人都有動畫的經驗，不管是在學校或是在工作上遇到。無論如何，我們都很熱愛動畫，所以我們想要做對animator來說最好最簡單使用的rig。

How many controls do rigs like the Road Runner and Coyote have?

請問像是Road Runner和Coyote這樣的角色有多少控制器？

The originals (coyote and road runner) had approximately 500 controls, roughly 150 of those were in the face. On the more recent rigs like Elmer and Granny, they were closer to 700 controls, mainly because they had outfits on.

原本像是coyote和road runner這樣的角色大概有500個控制器左右，裡頭大約有150個是臉部控制器，最新的幾個角色像是：Elmer 和Granny，他們就有多達700個控制器，主要是因為他們身上還有穿衣服的關係。

We finally had the chance to write a data mining tool that tells us how much of those rigs actually get used, but we didn't have that tool in place until we were mostly through the rigs on those shows. Only about 5% of those rigs were never touched, but on any particular shot, the rig usage could be anywhere from 50 to 75%. Obviously this means that not all controls were used in every shot, or even the same controls from shot to shot. Anyway, interesting statistics to go through, which means we can optimize rigs based on previous projects data.

後來我們終於有機會開發資料探勘(Data mining)的工具來提供我們更精確的資訊，用來說明我們的rig是如何被animator使用，一開始沒有這樣的工具，直到我們的rig跑過了幾個案子才有。我們發現，只有5%的控制器是animator從來不用的，一般鏡頭裡大約都會有50~75%的使用率。很顯然這告訴我們：不是所有的控制器都會被用到，還有不同鏡頭之間所使用的控制器會不同。總之，有了這樣有趣的統計數據，表示我們有機會根據之前案子所累積的資料把rig再做的更好。

I’m very curious as to how you were able to push for such great flexibility and exaggeration in the rigs that were used in the Looney Tunes shorts, it really seems like CG has come along way. Was there any need for research and development for the Looney Tunes short, and what were some of the issues, if any, that were encountered during production?

我很好奇你們如何做到「兔八哥」短片裡那樣有彈性與誇張的rig？這似乎又把CG動畫向前推了一步。你們有為了「兔八哥」的案子做任何R&D嗎？有的話，你們做了哪些？

R&D happened on the fly for our Looney shorts. The schedules were so tight, and our insistency on getting the models/design to perfection (to hit the right Chuck Jones designs) just made things tighter. We had to develop a few new rigging techniques, but nothing that you would see on a feature. No new plugins, but some general scripts/tools and rigging pipeline updates were made. Generally, our rigs are pretty flexible with the standard templates. Kyle Wood, one of our lead riggers, spent a lot of time developing some new facial rig modules, which certainly helped the animators achieve the squashing, stretching, and insane shape-pushing that was needed. The animators did an awesome job at pushing and pulling those rigs to match the poses of the classics, and a lot of what you’re seeing on screen is because of their patience with nailing each pose. They were sculpting on each frame at times, whether that was using the rig, using special deformers or tools, or sculpting with in-shot PSDs.

「兔八哥」的R&D其實是很倉促的，時程非常趕，而且因為我們很堅持要在model上達到Chuck Jones原作的設計，所以時程就變得更緊。我們開發了幾個rig的工具，但是你在片子上看不出來，沒有開發新的外掛(plugins)，主要是一些程式工具(scripts/tools)和rig工作流程的改進。老實說我們的模板對建構普通角色來講已經很足夠了。我們當中一個資深的rigger Kyle Wood，還花了很多時間幫我們建構了一套新的臉部rig系統，這套系統對於幫助animator做到臉部的擠壓伸縮和一些誇張變形很有幫助。Animator做得很棒，他們花了很多時間去把這些rig擺得跟過去經典卡通裡的pose一模一樣，很多你在影片裡看到得很棒的東西，都是因為他們很有耐心的去精雕細琢每個pose，當他們在使用rig時，他們會用上變形器跟一些工具，或是針對鏡頭使用PSDs(編按：Pose Space Deformation)，他們會一張張的雕琢畫面。