Stop Motion with Live action Plates

[David Regan]

So I've got an project coming up with an interesting sequence involving a man fighting a large robotic ant. Fun stuff. We are going to do the sequence in two sections, one being the live action elements of the man, some in front of green screen, and the other being a stop motion of an ant puppet. It's not the most complex of sequences, as our budget/rescources are limited, we have confined much of the shots to singles, but there are a few shots which will combine both of the characters in the same frame, to really sell the effect. We are shooting on film, I'm looking at shooting 7217 and 7219, mostly the '17 for the spfx work.

 

Firstly, has anyone ever done much with matching live action with stop motion elements? I know I'll have things to watch out for, mostly matching camera angles, focus, lens, etc...to get the elements to match. And I know camera movement with both objects is pretty much restricted, since we don't have motion control capabilities.

 

Secondly, for the stop motion section, we are going to be shooting hi-res digital stills for it, as opposed to straight on film. Our output is HD, no print unfortunately. So, would it be worth taking our digital stop motion sequences, getting that scanned to film, then transferred back to HD? Given we are ending digitally, would it just make more sense to stay digital and try to futz with it a bit in post to try and match? Somehow I have a feeling even with digital grain and such it's going to stick out, which is why I'd like to have it get back on film, albeit it will be conformed to HD in the end.

 

Anyway, it's a sorta unique area of VFX I haven't delved into before, so if anyone has any tips or thoughts it would be much appreciated.

 

Cheers

[Scott Fritzshall]

Firstly, if you're doing greenscreen work, I'd really recommend that you try to shoot on 35mm or something higher than S16. You're going to end up with much more grain and it's going to be a pain in the ass to get a good key.

 

For shooting your elements, make sure that you are matching your lighting. That's one of the most important things.

 

If you get a filmout of your digital stills [i'm assuming that you were thinking of doing this in order to add grain to them, correct?], they're going to end up with the grain that the 35mm print stock has. That's not what you want, and it's not going to match the rest of your footage. You want to either shoot it on the same format, or add your own grain to it. You just need to be very careful about matching grain during compositing.

[Paul Bruening]

Hey Scott,

 

Won't they need to shoot one layer set then shoot matched action with a monitor for the second layer set? Either match the puppet to the live action or vice versa? I'm not second guessing them. I actually want to know this for myself.

 

If they are going to digitally composite them, why not do the ant in Maya? You can get an educational copy for pretty cheap if you throw a college student some drinking dough for doing the order. Wouldn't it be easier and more precise for them to do the MoTrac by shooting reference dots on the green screen then analyzing and matching up the Maya ant? They could shoot the green hand held if they cared to, then.

[Scott Fritzshall]

Won't they need to shoot one layer set then shoot matched action with a monitor for the second layer set? Either match the puppet to the live action or vice versa? I'm not second guessing them. I actually want to know this for myself.

Yeah, you need to record all of your camera and lighting information from one so you can match it in the other. Ideally, you typically want to shoot the background first so that you know what the light is like for the foreground. You need to take a lot of notes, and having screen captures on hand makes things a lot easier.

 

If they are going to digitally composite them, why not do the ant in Maya? You can get an educational copy for pretty cheap if you throw a college student some drinking dough for doing the order. Wouldn't it be easier and more precise for them to do the MoTrac by shooting reference dots on the green screen then analyzing and matching up the Maya ant? They could shoot the green hand held if they cared to, then.

Well they could do a CG ant. But A) I presume that they want to do stop-motion because it has a very distinctive aesthetic that's difficult to match in CG, and B) it's a lot easier said than done, especially for someone who isn't familiar with it. Matchmoving and modelling, rigging, animating, texturing, shading, rendering, and compositing a CG character is quite a lot of work, especially for someone who isn't experienced with doing all of that. There are enough no-budget films (and ones that do have a budget) that never get finished for whatever reason, and visual effects adds a huge layer of complexity on top of that. Unless you've got a very experienced VFX person (hard to find with no budget), I generally recommend that you keep the effects work as simple and manageable as possible.

[David Regan]

Thanks for the tips guys

 

Regarding matching lighting/movement and such, yeah I will definitely be taking detailed notes. We are planning to do a full blocking of the scene next week, so we can really lock down all the movements, framing, compositions etc...to plan for the best.

 

As far as keying, the way the shots work out, I was looking over our storyboards, we really only have to key stuff off the stop motion from the dSLR frames, which will be significantly easier than keying from S16 as you mentioned Scott. If this film had the time/budget/resources, there are many different things I would 'like' to do with this sequence, but due to our restrictions we have to limit the options. So for that reason, as I mentioned we won't be doing a lot of human/ant interaction, and as far as seeing the two layered in the frame, it will be easiest if the ant is in the FG.

For example we have a shot where the ant stands up and towers over the human character. So I'll shoot the wide down angle on the human first, then use a transparency to trace out the frame on the monitor. Then when we go shoot our stills, I'd like to bring up the image on the montior, overlay the transparency so the frame can be matched. This will be over greenscreen. Obviously lighting continuity will be huge as you guys mentioned, as well as camera angle, focal length, etc...

 

But I think it can sell.

 

Camera movement is my biggest regret, rather that we won't be able to do it. On the one hand, doing this CG would make for the possibility of being able to track the camera movement in the computer, which would make moving camera shots a bit more feasible. However, Paul, I would say my biggest reason for not doing CG, although I agree it does have a lot of perks, and we do have Maya, is just how it would ultimately look. I did smaller HD project last year, where we thought it would be a good idea to have a CG character, and even with about 10+ weeks of work, it really didn't mesh or sell at all, and just look reeeaaalllly CG. The amount of work required to animate, texture, render, light etc...I think would just be beyond our resources and manpower.

 

The one element I think will be CG, is the physical interaction between the ant and human. At the end the human stabs the ant with a long pole, so I think for that shot, where I see both and and human in the same shot, I will go with a CG pole.

 

Not that stop motion is going to be loads more realistic, but it's something the director and myself both want to try, and do like the aesthetic, this film is supposed to be in the campy style of old B-movies, with outrageous events, as I'm sure you can guess from the concept of an ant fighting a human. But regardless it should be fun and a learning experience.

[John Brawley]

We are shooting on film, I'm looking at shooting 7217 and 7219, mostly the '17 for the spfx work.

 

Firstly, has anyone ever done much with matching live action with stop motion elements? I know I'll have things to watch out for, mostly matching camera angles, focus, lens, etc...to get the elements to match. And I know camera movement with both objects is pretty much restricted, since we don't have motion control capabilities.

 

Secondly, for the stop motion section, we are going to be shooting hi-res digital stills for it, as opposed to straight on film.

 

Cheers

 

Hi david.

 

A few things. Firstly, the main issue will be matching not only your camera angles and lenses, but matching your depth of field. You'll find, especially if you're shooting a 35mm sensor DSLR that the DOF will be fairly shallow. Now your giant robotic ant which is really only 1' tall needs to ALL be in focus. Imagine shooting it at it's true scale, and then imagine what the DOF would be on a given lens setup.

 

Therefore it's not unusual to shoot miniatures at T22 or greater, so you can keep it all in focus and give the appearance of a larger DOF.

 

Essentially, it's a question of scale. What you need to do is work out how big the ant is MEANT to be and then how big it actually is and then you have the scale. Then you can work out speed of camera movement (tracks and moves) , lens height etc using this key scale amount to make sure everything matches up.

 

As mentioned, 16mm is not the most ideal VFX format. You might want to consider shooting the bluescreen elements on 35 or maybe even on something like RED. 16mm tends to be less stable than 35mm, especially when using the older cameras like the SR3's and below. Modern Aaton's and the 416 or the SR3 Advanced are better, but make sure you do a DX test before you shoot if you have to go 16mm. I've actually often found the Aaton A-minima is the most steady 16mm camera I've ever shot !

 

I'm a big fan of doing things in camera with CG augmentation rather than all CG, and I think it's great that you're attempting this. Make sure you let us see the end result.

 

jb

[Joe Giambrone]

Adding to what was just said...

 

You have the opportunity to build your model larger, and use less scaling. Do it! The bigger your ant model the easier it will be to sell the effect. There is a squareroot formula you can look up so as to match the framerate to the scaling, so the ant moves slower compared to the humans.

 

Hope to see your results.

[David Regan]

Thanks for the tips. I'll definitely be stopping down, thanks for reminding me John, I noticed in Dennis Muren's Article in the ASC manual he notes that for stop-mo 1/4 sec exposures are a good place to start.

 

Also in said article, and going off what you guys were saying regarding speed, I just want to make sure I understand the concept clearly. For moving miniature objects, you typically want to run the camera at a higher speed, again with Muren's example, a driving car going 60mph, obviously it can't move 60mph on a miniature set, so you film it moving slower at a higher frame right? Is this correct in essence?

 

However I assume this doesn't translate extremely well into my situation, which is stop motion, where the ant doesn't do a hell of a lot, mostly rearing up and swiping with its lets. So for this would I be safe in shooting it traditionally stop motion, or should I actually take more frames to slow down the action?

 

Finally, I was considering running the camera for the live action stuff at 90 degree shutter, add a bit more pop to the character's movements. Obviously the stop motion will have a different feel, but perhaps this will help it mesh a wee bit better. Just a thought.

 

Once again I appreciate the thoughts. This is all new territory for myself, so it's sort of learn-as-you-go experience.

 

Cheers

[David Regan]

Also just did a quick focal length conversion, I think this is right.

 

Say I shoot the stop-motion with a D200. The sensor size is 23.6mm wide, as opposed to the 12.52 width of a s16 frame. Hence the lens must be 1.88 times longer on the dSLR to have an equivalent lens length. Does this seem right?

[Scott Fritzshall]

Also in said article, and going off what you guys were saying regarding speed, I just want to make sure I understand the concept clearly. For moving miniature objects, you typically want to run the camera at a higher speed, again with Muren's example, a driving car going 60mph, obviously it can't move 60mph on a miniature set, so you film it moving slower at a higher frame right? Is this correct in essence?

 

However I assume this doesn't translate extremely well into my situation, which is stop motion, where the ant doesn't do a hell of a lot, mostly rearing up and swiping with its lets. So for this would I be safe in shooting it traditionally stop motion, or should I actually take more frames to slow down the action?

What Muren was talking about only applies to miniatures that are moving. It doesn't apply to stop motion at all. You do, however, need to pay a lot of attention to how you're animating your ant to make him move as though he was really huge.

 

Finally, I was considering running the camera for the live action stuff at 90 degree shutter, add a bit more pop to the character's movements. Obviously the stop motion will have a different feel, but perhaps this will help it mesh a wee bit better. Just a thought.

It's your aesthetic call. It's not going to make it look more similar to your live action, though, it's just going to make your live action look strange and stuttery.

Say I shoot the stop-motion with a D200. The sensor size is 23.6mm wide, as opposed to the 12.52 width of a s16 frame. Hence the lens must be 1.88 times longer on the dSLR to have an equivalent lens length. Does this seem right?

It seems about correct. I'm sure there's a chart or something that you could find.

[Paul Bruening]

Is all this really better than CG?

[David Regan]

Is all this really better than CG?

 

 

Haha, in an ideal world, probably not, given the creative possibilities it offers it would be nice but there are restrictions. The film is for a student project, meaning it has a pretty strict deadline, 10 weeks. If we had a full VFX team, and a bunch of people experienced enough to model/rig/animate/texture/track camera moves/light/render etc...in that time I would say go for it. While I don't doubt my peers abilities, I just don't think we have the manpower or time at this stage to set about that. Not that stop motion is any 'easier' but it's something we can get done, and fits more naturally into the aesthetic, think the stop motion of the robot in robo-cop, or the snake in beetlejuice.

 

So yeah I totally see where you coming from, but at the same time I'm excited to explore the possibilities of stop motion, and getting to light and shoot physical objects, as well as learn from all the challenges it presents.

[Chris Keth]

Think of it this way, Dave: In scale, this is the biggest lighting package you'll have for a long, long time, maybe ever. ;)

[John Brawley]

Is all this really better than CG?

 

 

Unequivocally yes.

 

The reality is that even the best CG can still fail to look *real*. Maybe that's OK if it suits your project, but usually people want photoreal CG has a role to play in augmentation and of course there are certain things that can only be done CG.

 

Its the reason why motion capture even exists. Because animators know that it's better to start in reality before going 100% CG.

 

GOOD CG, that is done well is incredibly expensive too.

 

jb

[David Regan]

Unequivocally yes.

 

The reality is that even the best CG can still fail to look *real*. Maybe that's OK if it suits your project, but usually people want photoreal CG has a role to play in augmentation and of course there are certain things that can only be done CG.

 

Its the reason why motion capture even exists. Because animators know that it's better to start in reality before going 100% CG.

 

GOOD CG, that is done well is incredibly expensive too.

 

jb

 

 

Haha, possibly true Chris, but I'm gonna think optimistically. 18ks.....somedaaay....lol (Not that tools have anything to do with good cinematography)

 

Yeah John, I guess you sort of hit the main crux of the issue, weather or not it looks *real.* The director and myself both recognize that stop motion won't mesh perfectly, it will have a different aesthetic, and given our both being newcomers to the technique, we will have our problems, and we realize that. But I think we both feel that in either case, have a real, photographed element, even if it isn't perfectly meshed, is still going to be better than a fully CG model where not only is it not going to be perfect, but it's going to be CG to boot.

 

So we will go ahead with stop-motion, learn along the way and have fun. I'll be sure to post results, I'm gonna keep researching and planning, thanks for the tips, I really appreciate it going into this unknown.

[Chris Keth]

Haha, possibly true Chris, but I'm gonna think optimistically. 18ks.....somedaaay....lol (Not that tools have anything to do with good cinematography)

 

Yeah John, I guess you sort of hit the main crux of the issue, weather or not it looks *real.* The director and myself both recognize that stop motion won't mesh perfectly, it will have a different aesthetic, and given our both being newcomers to the technique, we will have our problems, and we realize that. But I think we both feel that in either case, have a real, photographed element, even if it isn't perfectly meshed, is still going to be better than a fully CG model where not only is it not going to be perfect, but it's going to be CG to boot.

 

So we will go ahead with stop-motion, learn along the way and have fun. I'll be sure to post results, I'm gonna keep researching and planning, thanks for the tips, I really appreciate it going into this unknown.

 

Please do post results. I wish I had the chance to do more stuff like this when I was still in school. It's loads of fun.

 

I didn't necessarily mean that you'll never get into big budgets, Dave. You know I'm more optimistic than that. :P What you never will have, though, is the ability to toss up a light that takes 2 minutes to set and have it play for sun :-D