How ILM Made The Mandalorian's Razor Crest Motion-Controlled Miniature!

hey guys adam savage here in my cave and i want to  talk about the mandalorian yeah i i'm as obsessed   with you are at how great season one was and  i'm as excited as you are about season two but   one of the things i want to talk about today is  that season one utilized some genuine old-school   awesome ilm technology and it has motion control  shots that means the razor crest was built   by my friend john goodson as a physical model two  feet long and they had 17 i think motion control   shots there's a beautiful little featurette from  ilm up on youtube right now this is the technology   that they were using back when i worked at ilm but  no one's touched it in well over a decade the two   people most responsible for this process model  maker john goodson and vfx supervisor john knoll   both of them great friends of mine and we're going  to talk on zoom about how this came to be and how   they did it well john and john it is really good  to see you guys how how are you on this fine day   very well i do i i'm pretty good i'm  holding up i get to come to the shop a lot   um so i was super excited when the mandalorian was  announced i was excited when i started to see some   trailers but i got really excited when i found  out that there are some good honest-to-goodness   models in this show and motion control shots  and i guess the first thing i want to know is   how the heck did that come about what what  is the origin story of of these shots well uh   it started from um we built this really beautiful  cg model of the the razor crest but these you know   high high shine metallic finishes um you'd think  they would be easy to do in computer graphics but   there's a great deal of subtlety to them and  and trying to get them to really look right is   is challenging but i felt like we were in pretty  good shape on it we had a couple of shots that   we were doing uh that were space flybys of  the of the razor crest and um john favreau   was eyeing them a bit suspiciously saying you  know there's something i don't know about the   reflectivity just doesn't seem right and i i think  we should build a model i think we should build a   reference model uh shoot some stills of it um and  that'll tell us something about um you know what's   what's not right about this and uh um and we  thought that uh if we build a model that's that's   good enough for reference pictures uh hey wouldn't  it be cool if we actually used it to do some shots   um it's not too big a leap to go from a really  nice maquette to something that's suitable yeah   and wouldn't that be fun to do some motion control  work on it but the big problem for us was that   you know as as much of a prestige project as uh  as mandalorian is it's still relatively low budget   compared to the features that we work on and at  this point we had already had all of our budgets   uh approved for doing all the shot work and you  know one of the reasons why you don't see it   as much miniature and motion control work these  days is it's more expensive to execute than   computer graphics and not a lot of clients want to  pay for that and and so part of the challenge was   is there a way we can do this on the cheap because  if we turn this into a great big production   you know the costs are going to escalate out  of control and we won't be able to afford it   so the only way this is really going to be  practical if we is if we uh treat this as a total   garage operation you know it's it's it's got to  be done um using uh as much sort of off-the-shelf   commodity equipment and uh and kind of seat of the  pants um slab dash kind of but i mean you spent   you spent the first couple of decades of your  career like living eating eating sleeping   and breathing this process oh yeah but in the  interim since you since cg has become ascendant   the robot and diy and arduino marketplace  has become so much more uh uh expansive that   must have made it so i'm curious about the  process of like wrapping your head around an   old process and finding out that there's this new  way to do it well yeah i i started my career as a   model maker and then i transitioned into motion  control camera and i used some of the you know   partly how i got into motion control is i started  building motion control equipment um well that was   you know this this was back in the early 80s  when anybody who had a motion control system   had built it themselves um there were no companies  that were commercially selling them at the time   and i got fascinated by these cameras i thought  they were super cool and uh and i thought you know   i want to i want to build one and i was a college  student at the time i was going to usc and my   final project at usc was in an advanced animation  class where i built a four channel motion control   system that i i bolted onto an oxberry animation  stand at the school and i shot a two-minute slit   scan movie that was an art exploration of the  technique that was the last thing i did there   it worked it was great it was really fun the whole  thing kind of operated uh under computer control   and i showed that at my job interview at ilm and  partly on the strength of that work that's what   got me hired into the camera department so for  the first few years at ilm i was doing motion   control all day every day and then gradually uh  i started moving up into the computer graphics   department and then the supervision and and i  stopped doing that and about 10 years ago i was   working on a fun little project that uh where i  needed to do some electronics and started learning   the whole uh hobby electronics world which has  transformed uh you know radically over the the   decades it's super cool now um you know with the  the arduino community is really wonderful and   places like adafruit and sparkfun that that make  all these wonderful little breakout boards that   give access to hobbyists to you know tiny little  uh packages that you could never solder to   manually and you have all these libraries that  make them very easy to use it's it's fantastic   and it used to be that a lot of the motion control  equipment that we used was very expensive but   3d printers and cnc routers and those kinds of  things have driven the price of a lot of those   things down the linear bearings and the and the  steppers and all of that the economy of scale   has made it all much more affordable yeah exactly  so um when this project came up um you know that   the thought my first thought was could i put  together a very inexpensive system because we   couldn't afford anything particularly expensive  that could actually shoot these these elements   and you know first thought was um you know  i'm a big fan of the 80 20 aluminum extrusion   rapid prototyping system uh to see how much  of the structure could i build out of 80 20   and minimize the number of things that i actually  had to machine from scratch to build it and you   know probably the the the two biggest things i had  to machine to scrap from scratch where uh pan tilt   head that's what this is so i built this in my  garage this is a floating l pan tilt head uh and a   uh omega drive track and i'm assuming you sent it  out to have it anodized black i did yeah um sexy   well you know there's a reason why that  equipment's black is that you know especially you   have a silver spaceship any stray light bouncing  onto the equipment is gonna you know be visible   in the ship so so i'm curious if there's a moment  at which you got to tell favreau he visited the   set of mythbusters years ago and as we were  talking i said i love your use of practical   effects and he lit up like a christmas dream we  just started talking about practical effects so i   know how much he loves them i'm curious about his  reaction when you told him i think we can get this   across the line i think we can do it i didn't  talk to him in person about that what i did is   i put together a cad model of of what i was  proposing for this motion control system and   it was kind of the minimum deliverable like what's  the what's the least amount that uh we would need   to put together to be able to do the kinds of  shots we were talking about and i was looking   at our shooting space you know we have this uh uh  motion capture and element shooting stage at ilm   that's 50 feet by 50 feet with a 25 foot ceiling  and so i figured all right if i've got a if i put   the track diagonally i can have a 50-foot track or  so and that's probably about the practical limit   that i can do inside the space we've got and then  looking at the kinds of shots that they wanted to   you do you have to figure out what's the  scale of model that you want to build   and it's it's trade-off right because the uh the  bigger the model you build the closer you can get   to it um so you can get these nice detail shots  on it but then you're limited as to how small it   can be at the far end of the track right so um we  going back and forth a bit and i got um a 3d scene   file of one of the type of shots that they wanted  to be able to do i started playing around with   with different scales of the razor crest and it  looked like between 18 inches and two feet was   the sweet spot where you know you could put  enough detail into it yeah um that you could   get reasonably close to it but you could also do  dynamic shots where you get far enough away from   it and you as you know the uh on the first star  wars film the millennium falcon was this beautiful   six foot long model so beautiful that one and  the original dykstraflex had a 40-foot track oh   wow and you can see in the shots that the falcon  never gets far away from camera you know and so   then on empire strikes back um a two foot model  was built and they used both you know if you need   to get close to the use the six footer and if you  need to get far away from it use the two footer   and of course obviously uh you can add more  detail if you hire someone like john goodson   so john i'm curious what's it like to get a  call to build a model for a star wars movie   i i think you might have thought that that was  completely done well yeah and i think the little   doc that's running on you know on youtube right  now i'm saying at the beginning i'm like yeah   that sounds great and everything but it'll never  happen and then it did and because we've talked   about things like that in the past it comes  up and it's a nice idea and people love it   but it just never it usually never gets off the  ground so i thought this is another one of those   like yeah we want to do a practical model i'm  like great sure that'll just fall off the back and we actually were doing it and it was really  kind of a shock and the whole thing i've been   i've never stopped building models but the last 15  years i was at ilm i was doing computer graphics   and it's really interesting to go back to doing  what i was doing when i first started there the   first 15 years i was there that's what i was doing  we're doing you know ships on flips motion control   and it was just like you know like they say it's  like riding a bike it's like you go right back   to the same things and walking out to do stage  support it could have been yesterday it felt   exactly the same you know so i'm really curious  about this because i remember when fogler when   dave fogler we all know um shifted like you did  john from practical model making to cg model   making he said to me you reported you know it  feels like the same in my brain he said i'm still   doing the same kinds of problem solving but i'm  curious john goodson if like building a model from   a dungtang drawing for a star wars film if there's  some part of the process you had forgotten like   oh this is really neat or some well you know this  was different because you know landis fields 3d   printed the model so we started off with the  3d print but it was interesting because it came   in about 50 pieces it was like getting a deck of  cards oh my god like and it's got all the support   structure and everything i'm like i don't know  what the hell i'm looking at and i don't know how   this goes together so he made me a color-coded map  that showed all the pieces so i could figure this   thing out so it was a little different than you  know when you're building something from scratch   right you know and that that's one of the things  i say is really different between working in the   computer and working in a physical shop one of  the things i really like is in a real shop you   can go and you can pick up a formula one race car  body and maybe they're done detergent bottle and   you put these two things together you go wow  that's a really cool shape right well graphics   doesn't give you that you don't have that kind of  exploration there you're kind of stuck with here's   a drawing make that thing and so that that's  one of the things i think is really different   and sometimes you know if you look at all of the  concept models that were done at ilm physical   concept models so many of them were just kit parts  slammed together and that became the basement for   you look at the speeder bike the back end the  exhausts are actually the nose of a space shuttle   you know cut off and flipped upside down  and it doesn't have the windshield in it   if you look at it it's clearly obvious what it is  but that went from a concept model up to full size   right and that's what we've all been looking at  since returning to the jedi in 83 is you know   that sort of kit bash thing when we worked together at ilm i i think we  represented a couple of factions you used to   make fun of me for overuse of the laser cutter  and i was repeatedly astounded by what you   could render in styrene in incredibly short  periods of time so i'm curious as a modeler   what the process was like for you using a  3d model were there aspects in which you   like had to improve the model because the the cg  drawing didn't uh provide enough surface detail   uh well you know there were a couple issues with  that one of them was that we had a resolution of   3d print um there were build lines in the parts  and it was just because of the speed at which this   thing is being produced right and that's you know  one of the things about this is that there's there   are a lot of different steps involved a lot of  people involved from the initial concept through   development through the art department and then  if you go into execution of as a digital model   or then as a practical model and then all of them  have their challenges and so yeah the practical   thing it was you know getting getting all those  pieces putting them together and that was i've got   i the pattern how well we can see this it's  black but this is the pattern for the body   and this is about 40 something pieces wow comprise  this thing but when i got it all together i was   like there are a couple of things going on you got  to get rid of all those seams where you put the   pieces together and you got to clean up the build  lines because this is going to be covered in foil   right but then you also have to be able to  go through and cut out all the mount doors   and everything and make those as seamless as you  can so you can't see them the 3d print material   you try to drill a hole in it it just explodes  so and it's very delicate and you can see that   when i was taking this out of the mold all these  white patches of putty that's falling apart just   trying to get it out of the mold so it didn't seem  resilient enough to me to make a shooting model   out of it got it i took this and made a two-part  mold and split it right along this wing edge   and that allowed me to have a top and  a bottom and just recast an epoxy fiber   glass then i had the luxury be able to  cut out the doors and make everything fit   and i could work on it like a real model the 3d  the 3d print parts i start handling like a real   model and i break them all the pieces so and this  is also this was being done in your garage while   john knoll you're building the moco rig in your  garage is that correct yeah yeah i have a little   machine shop in my garage and um yeah the the fun  thing about uh about the motion control system is   that it's it exercises all three of the you  know the three pillars of engineering you   know there's the physical um mechanism building  uh motorized pieces that'll uh panic tilt the   camera that'll hold the the model and rotate it  and three axes uh move the camera down the track   so that's a machinist uh job and then there's  the electronics soldering up all the electronics   to to drive the thing and then the software  stack um writing the software that runs on the   the microcontrollers that uh that runs the whole  motion control system so it was yeah super fun um   you know cutting cutting metal uh soldering  wires and writing software well yeah what   could be better yeah um so what is it like on day  one there must have been a lot of excitement as   you're setting up to do a motion control shot  oh yeah yeah yeah we so we have this whole plan   um that you know when we got the green light to  uh i presented my proposal of uh all right this   is the budget i need to to build a this motion  control system and uh we figured if we start today   you know this was in i think january i  submitted the proposal we got an approval   that if we start today john can have the model  done in time for the shoot in in may that gives   me enough time to machine the parts i need  to make to build the electronics to write   the software so we were targeting uh i  think mid-may for uh for doing the shoot   and uh then every yes right go and uh and  we started started uh into the process   and uh you know for me the first thing is  all right well i need the hardware first so   let me start building stuff like that um  get it going uh and then um i was probably   oh two thirds of the way through getting all  the the hardware machine when i think john had   uh sent some pictures of uh of the miniature in  progress and favre got super excited about uh   about oh hey can we uh is there any way we  can we can show something at celebration oh no   which meant uh moving the times table up five  weeks and this was such a mad scramble that um   like the old days yeah you know that  immediately we started uh talking about   all right well uh what uh what kind  of shot could we do five weeks early   um and john what what part of the if we  only are going to show one side of the model   can you detail enough of it to get it to be  shootable for this one shot and that was where   they were like well we could do this one shot just  from the front and it's just gonna rock and roll   i'm like okay i didn't get the front done then  they had the meeting and they were like well now   it's two shots and then they sent me automatic  and i said well what do you what do you not see   and they sent me the shots and it it flies by and  you see the front the bottom the side and the back   and then i try to front the top the side and  the back and i said so the only thing you don't   see is the mirror on the left hand side that's  about it so you guys really narrowed this down excited part of me and thinking through this  is like we're all still 10 years old going   you know and it's like as soon as we're playing  around with the process that's close to this   everyone gets like i i in the in the featurette  there's like shots of people visiting the set   and there's that same grin on everybody's face  of like seeing the physical thing it had been   a while since we've done a motion control shoot  and um yeah a lot of the people who were working   at the company had you know weren't working there  when we were still shooting those elements so this   was something kind of exciting and exotic uh fun  retro thing and they were really excited to come   can i come by and just see it working yeah  we had a pretty steady parade of people that   that wanted to just see the system running i i  think that there's something i mean did you find   in the process of setting this up that it  allowed production to be more creative in   interesting and new ways that they than  they could have been with cg with cg models   no and it's it's sort of the opposite it's it's  um uh in that you have to make commitments you   have to make firm commitments to the design of  the shot because what's possible in the shot oh   well yeah there's there's some limitations um that  they're referred to in the the little documentary   piece about uh you have to think about how the  model is going to get mounted right and uh um and   you know one of the things i started to talk  about in that uh that documentary is that um   some of the shot designs were such that uh you  start with the ship flying towards cameron you pan   with it as it's flying away and it always has some  sort of role to it and the original model movers   that were done for new hope were two axis movers  it was a pylon with the motorized rotator so the   you know the post that the model is mounted on  can rotate and then that pylon is mounted on a   turntable so you have that sort of roll axis and  you have a yaw axis and you notice in new hope   there is not a single shot that is that design  where you start looking at the front of the ship   and pan with it as it goes away because if you  think about it there's how do you mount the model   you'd have you have to mount it on the off camera  side so you can't really do a nose mount and you   can't do a tail mount because the nose mount is  going to be visible at the front you know the   pylon is going to be over the top of the model and  then as you pan with it and it goes away the pylon   is going to be over the back of it so you have to  do a bottom mount or a side mount something that   that is going to be on the far side of the model  but then how do you do the roll on the model as   it goes away now you need a three axis mover to do  that you need that sort of cradle mechanism yeah   and at first i i had proposed all right well  we're going to embrace those limitations and   i'm going to just build a two-axis mover like the  the ones on new hope but then as soon as i saw the   wish list of the kinds of shots they wanted to do  half of them were these uh you know fly by with a   with a roll and i realized that uh all right well  there's no way around this i got to build a cradle   as well so we've got a full three axis movement  um john goodson i'm curious if working with the   model on set was all the same stuff extent were  you doing little paint touch-ups for the model   shot to shot and exactly the same nothing  that's the weird thing was walking out and   and starting to do that and it was there's been  a 15-year gap and it was like yesterday it was   yeah muscle memory kicks in it's just the  same thing and john was really patient   i come in john he was great in terms of giving me  time to work on stuff so i really appreciate that   you know something else went back to the garage  thing yeah one of the guy involved with this guy   named dan petrasky and dan has a machine shop  in his garage and he built the armature and has   helped me out with a variety of things on it  and uh we jokingly call ourselves three garage   productions because that's i i respectfully  submit my garage here for anything you might   need for the next season please call me send me  in coach i'm ready to come help i want to come   in bed with you for a day goodson i want to come  be your assistant on this next time no you don't um i was really curious to see uh that uh  the solution you came up with for the daytime   moco shots for the razor crest that blew my mind  tell me about that um yeah the razor crest is   especially tricky because it's that reflective  metallic surface and so it's oh and actually   sorry to derail but as a side tangent looking  at you covering the razor quest john reminded   me of that weekend we all spent covering the queen  ship in mylar to get that perfect shooting finish   but that thing i remember that thing would  it would wrinkle every night overnight it   would wrinkle because of the mylar and so we  would seam it and there'd be a panel line and   so you know it took a week for that thing to  stop deforming i just remember long days of   anyway sorry go ahead john yeah no  those metallic finishes are hard to do yeah so the razor crust reflects the whole  environment around it i mean that's basically   what you're seeing is just a reflection of the  environment and um and for space shots it was   relatively straightforward to light because you  really just need a key and you know a couple of   artfully placed bounce cards to to really get  it to look nice um and there's this level of   stylization that is perfectly fine with those  space shots but when you see it in atmosphere   um you know it's going to look wrong if we  do a slap dash shooting of uh we're trying   to emulate that lighting because you just won't  get the right thing it should be reflecting   all the colors of the environment and this very  subtle gradient of the the way the sky is brighter   at the horizon and uh and dimmer up uh above so  um you know i thought uh well hey this is the   mandalorian we're doing uh we're doing this uh  shooting stage down in southern california with   leds all surrounding they're they're doing  uh the equivalent of image-based lighting   back in the real world um you know i  could do a little miniature version   of that for these shots and so the thought was  well you want to surround the razor crest with   uh with imagery that is the right color the  right uh tones and with the right gradients   to try and get that same sort of look and so  we built this thing that we called the gazebo   that was uh basically uh enclosing the whole  razor crest in foam core boards and then we   arranged from below a couple of cinema projectors  that that could uh illuminate the inside of the   foam core and provide that reflective environment  for the razor crest so you built a mini volume   so that the reflections would work in the  environment you put it in i love that yeah had   that ever been done before back in the old days  um we well not not exactly like that uh but um   you know there's some some great old pictures  of the et mothership which was highly reflective   and they did a trick where they needed to  recreate the environment around it and you   know it's supposed to be in a clearing in the in  the forest at magic hour and so you would expect   that the horizon glow would be broken up by uh  by a tree line so if you look at those pictures   you can see that that folks are cutting out trees  and kind of placing them around illuminated foam   core around it so they're they're doing sort  of an analog version of that that technique   um you know john and john both of your careers  have have have spanned so many iterations of   the technologies um i'm and you talk about the  limitations of the equipment for star wars and i   know like for rogue one you actually utilize some  of those limitations to obtain extra realistic or   original trilogy looking model shots i'm curious  if they were in in going back to a motion control   shot if there were limitations you had forgotten  about that were like reinvigorated in your head   like oh i didn't i had i had yeah i wonder  if there's stuff that you forgot about that   you were reminded of in setting it up again yeah  you know one of the things that you see a lot uh   now are and i'm this is a pet peeve of mine  is impossible camera angles yeah or camera   moves where the the camera does something that  if all of this were real you could never shoot   and i'm a big fan of at least going through the  thought process of if this were all real how would   you be shooting it what kind of camera is this  is this helicopter is this a camera car is this a   rigid mount and then what what sort of aesthetic  things flow from those kinds of uh constraints   and uh you know there there were definitely uh  things to be thinking about um how big is the   camera isn't gonna clear the model right so as we  were working on shot design we had to make sure   that uh all right well the camera can't intersect  the model and where does the model mover go   um that so that it isn't going  to collide with with anything   and um is this a shot that actually can be done  with this miniature because as much detail as   was in it there's a limit to how close you can  get to it before it it yeah kind of looks like   a model now but do you feel like that anchors  the audience in the re in in a in a a a more uh   higher veracity reality well that's that's  the thought yeah is that you don't want to   uh be hitting the the audience over the head with  um with evidence just visual evidence that what   they're seeing isn't real um john goodson and i  know both of you johns i'm we're not to talk about   um further seasons of the mandalorian but um are  you excited about the idea of perhaps continuing   this process of doing more of more motion hill  shots motion control shots i know that directors   like favreau and guillermo and they love practical  model shots but like you said john they're they're   expensive do you think that um you're opening  up a new niche industry like returning to vinyl   well this work i think was very successful i'm i'm  pleased that um you're watching the show you know   the model shots don't stand out as not fitting in  you know they flow right into the rest of it and   and i'll say that the work that we did with the  miniature did improve the cg that as we were   working on other shots that were going to be done  with the cg model we were constantly referring   to the miniature and pointing it yeah but see  the way that this reflects here and the the   ping we're getting on these bright surfaces we  should be doing more of that on our cg model so   even though the the miniature was only in  what 16 shots or so they did have a pretty   broad effect that improved the look everywhere  and so there's no reason to not be doing that   predictions are you are you excited about  building more physical models john goodson   obviously you never stop yeah very much i think  this is great you know it just it kind of blew   me away that it actually happened yeah you know  it just like one of those things that we you know   people talk about it but it just generally  doesn't it doesn't come to pass but yeah it's   been fantastic and like i said it's kind of like  falling off you know it's like riding a bike it's   a 15-year gap but as soon as we're out on stage it  didn't feel like there was a gap at all it just we   picked up right where we left off it's just it's  it's a very scrappy it's a lovely scrappy thing   you know what i mean like in the same way that  original star wars was oh gosh what is that john   yeah well this is this is the system that i built  for season one okay this is uh so this is the and   it's a as you can see from the horrific rat's  nest nature this is a hand soldered prototype   and and yes it worked it absolutely worked but uh  um you know the next step that you do after you   prove out the uh the design  decisions in a prototype is   is you try to make a little bit better  version of it so this is uh this is taking   the the design with a few modifications from the  season season one setup and building designing   a printed circuit board that encapsulates all  that rat's nest of wiring is now in this board   so this potentially is going to be uh um a  little safer in that uh it's much less likely to   uh get caught on something and for a wire to break  loose or something so uh something like that i   think is um sort of a fun direction to be going  uh that's the this is the jog box this is the uh   this is the handheld component of the motion  control system and you you kind of need one   of these because you can't uh there's so many  things you need to do kind of out at the model   or looking through the viewfinder where you're  setting keyframes or or moving things around   and so this again you know this is a hand soldered  prototype that worked and we used it for doing all   that season one work but uh since then i've i've  started and this is still work in progress but   this is a printed circuit board that has uh all  of that in there and um i have yet to machine   the casing for this this new one but uh you know  it's well uh if you do gear up again if you do   gear up again and you need a model maker please  call me i'm ready i'm ready to jump in coach   um it's beautiful work that you guys did and it's  really exciting i i you know working in the model   shop was a a special a a super special time in my  life and it informs the adult that i am and uh i'm   so glad you guys got to get your hands dirty on  this production because the result is incredible   one of the things i've loved about this  is i get these texts from john at 10 30   11 o'clock at night like look at the anodized  parts there'll be a photograph of the parts dog box the fact he enjoys doing that so much it's  it's pretty thrilling to get these texts like hey   check it out i'm like oh yeah this is great  this is great yeah totally this okay so the uh   this new uh electronics package fits into a  very 80s retro uh front panel sort of like so i love it it is the most niche audience possible   that you are that you are  making these parts for john there are there are camera  operators going oh i wish i had that   that is so awesome guys thank you so much  for taking me on a tour through through   mando uh mandalorian motion control shots is super  exciting absolutely happy to happy to accommodate

Channel: Adam Savage’s Tested

Views: 360,046

Rating: 4.9592886 out of 5

Keywords: tested, testedcom, adam savage, modelmaking, ilm, vfx, miniatures, motion control, star wars, the mandalorian, razor crest, john knoll, john goodson, industrial light & magic (ilm), industrial light & magic the art of special effects, the mandalorian baby yoda, visual effects, motion control camera, motion control camera system, the mandalorian trailer, industrial light & magic, jon favreau, the razor crest, vfx breakdown, model making, mandalorian behind the scenes

Id: MtGAiEQEIXE

Length: 35min 48sec (2148 seconds)

Published: Mon Sep 28 2020