Is there a way to give the appearance of volumetric haze in a stopmo scene? The straightforward way seems doomed to failure in a stopmo environment (unless there's some magical haze product out there that maintains a perfectly consistent density for hours on end...). Shooting in layers and applying progressively a progressively hazier color grade on the more distant layers digitally could work for some scenes, but that's assuming each layer is narrow enough to be approximated as 2D relative to the haze gradient, and the particular scene I'm considering involves a very large character visibly fading back into the fog.
Is it simply as impossible as it seems, or is there some genius trick to make it happen?
That's a tough one. The only thing that's coming to mind (well there are 2, but the second one is ridiculous - I'll get to it in a minute) is a sort of pepper's ghost effect. A sheet of glass standing between camera and puppet and something behind the camera that will reflect in the glass, and maybe you could slowly bring up the light on it so it obscures the puppet progressively. But you'd have to hide the camera and yourself well - a black screen with just a hole cut for the lens? Or maybe have the glass at a 45 degree angle so it isn't reflecting the camera. The scene it's reflecting could be painted to resemble mist, but not the swirling kind - just a long board painted with a misty whitish gradient dim at one end and much thicker at the other, and you slowly scroll it frame by frame so it gets stronger, or just animate the light getting brighter, and maybe at the same time animate your aperture slowly closing (if you have a clickless aperture ring).
Oh- almost forgot - the second idea is to have the camera securely bolted to the set, tilt it all upright so the camera is pointing straight down, and do your animation in a tank of water with something dissolved in it to resemble haze. I'm thinking of the Tool video with the silicone puppet where they tilted the entire set and camera on its side and had a stream of water so when the animation was tilted upright again the water seemed to be streaming horizontally across the set, and later they were animating the puppet as the water visibly rose around it until it was entirely submerged. Of course you'd need a full silicone puppet with no fabric clothes or anything, so it would need to be designed that way from the start.
There are massive problems with both of these approaches though - probably best to do it in post.
Hold on - I just realized the pepper's ghost effect would probably work a lot better if it's the puppet who's being reflected in the glass (at a 45 degree angle, so off to the side). You could animate a spotlight aimed at the puppet getting dimmer frame by frame as he walks.
Or possibly shoot a live action video of the set filled with smoke (I'll leave the hows and wheretofores up to you on that one) and composite in a shot of the puppet walking against a black backdrop. Then you can keyframe opacity of the puppet layer.
Or - probably the easiest way - shoot a background layer of the set, add a composited layer of the puppet walking away and a composited layer of stock mist.
Those are some fantastic ideas for a character moving through fog; I'm going to have to look up that video in the water tank, that sounds really cool. What I'm thinking of is a little different, though -- within a single frame, certain parts of the puppet/environment are less hazy, while others are far enough away to see them fading back into the mist; something like this:
I can't think of a way to do that without either a) animating in an actual, hazy environment (in which case, how can one maintain a consistent level of haze between frames?), or b) an *insane* amount of digital processing (pretty much matching the whole puppet digitally to carve out its shape within digital fog).
I'm open to the possibility that what I'm looking to do is actually impossible, but I figure if someone knows how to do it, this is the place to ask.
Just had a super-convoluted idea based on a very limited understanding of the actual physical properties of the materials involved... But mashing up one of your ideas with the genius UV-painted rig removal technique mentioned in the recent SIGGRAPH video yields something with a (pepper's) ghost of a chance of working.
So say I put up the half-silvered mirror like in a Pepper's Ghost setup, but flipped so it's working more like a teleprompter (light from the side goes on set, rather into the camera). In theory, if I set up a light firing right into the mirror (so hitting the puppet straight from the camera's perspective), and took a separate exposure lit only by that spot, the parts of the puppet closer to the camera would receive more light than the parts farther away (and the straight-on perspective would eliminate shadows). To get a good gradient from the near to the far end of the puppet, this would all have to be pretty close together, but maybe workable.
Now, using a normal light, that wouldn't be a super effective matte (as different colors on the puppet itself would interfere pretty hard), but what if I painted the whole puppet in a layer of clear UV paint, like Laika used for the electrocuted-box-troll shot, and fired a UV light into the mirror? In theory, that separate exposure would give a great matte to use to composite the puppet into a foggy environment.
There are a host of things that could be wrong with that idea, obviously, but hey... it's a start.
I like the thinking outside the box approach! That might not be THE idea, but it's a good lead toward it.
Took me a while to figure out which Tool video it was - the song titles are strange words that don't relate directly to anything in the song (except for a few like Sober or Prison Sex), so you just kinda have to start looking through their whole catalog of videos until you find the one you're looking for. In this case it's Aenema:
From the 90's - renaissance of amazing music and videos and when MTV still meant music television (or was it already becoming a wasteland of terrible reality shows?)
I don't know if that will work, or even if I understand it, but it sounds like it's worth trying.
I know a couple of ways to get a fog effect by splitting the set into layers, but all of the puppet would be on one layer so you couldn't get parts of it coming through the fog more than other parts.
The only way I know to get a true fog effect is what they did in Robocop 2, which is to shoot stopmotion in a sealed room with real fog. They had a sensor to read the fog density, and a computer to calculate how much fog to add to keep the same level. I tried it without a computer, in a closed room with fans to mix the fog, and giving the fog machine a squirt after taking each frame. Alas, it failed, I got horrible flicker from the fog level going up and down.
With layers, you can have the distance foggier than the foreground, and put the puppet anywhere in front of or behind other objects. The same as if you physically put screens of gauze in the set - the puppet can't walk through the gauze, and it can't walk forward from one layer into another. But you could shoot it clean, and put a dynamic pale foggy grade on it so it gradually became clearer from frame to frame, that would simulate the effect of it coming through the fog. You just wouldn't get parts of it more hidden in the fog than others. Unless you painted over it on other layers with transparent light grey, then erased bits around the edges frame by frame to keep up with the movements of the puppet.
If I ever shoot a Jack the Ripper mini-short I have in mind, I will have to deal with these issues. It won't be for this halloween, but maybe next year.
I can get a fog effect in Lightwave 3d with cgi objects, so a stopmotion puppet image mapped onto a flat object would get more obscured as it goes back further from camera, same with bits of the set mapped onto objects at various distances. I could even have them get clearer as the virtual camera moves towards them. But as it is like a flat image on a screen, each object will still have the same fogginess all over. With static bits of set, they could be shot in fog as a single still image to get that variation I guess.
This can be achieved digitally but you would need good compositing software - BMD Fusion is free.
Take your still images into Fusion and manually create your Z depth matte with polylines (giving each one a different shade of gray). Then, feed this Z Depth matte into a (can't remember right now) either 3D Fog or VolumeFog and pipe the Z depth to the 3dFog node. Then, create the fog material and pipe that in. It's better demonstrated than this. I'll try to dig up a video.
Here's video demonstrating a fog effect. However, it doesn't show manually creating Z depth matte - https://www.youtube.com/watch?v=O33Jnts2-BU
Other method I would think is better (or easier) is to film with stereo camera setup. Then, bring both left and right images into Fusion (for this you would need Fusion Stuido) and have the stereo images create your Z depth matte. Then, the 3DFog or Volume Fog would work.
This video demonstrates stereo camera work and depth masks - https://www.youtube.com/watch?v=Eo-x_eXRveo
At 1:14, Eric talks about creating depth matte from polyline masks. This is what you would manually create for the example above. You would then animate the polyline masks as your creature was moving through the fog.
Yeah, it's really the manual creation of the Z depth matte that seems to be the tricky part -- but if Fusion can generate that from stereo data then that's it! Stereoscopic 3D is so much simpler in stopmo than live-action... this would be a fantastic alternate application.
I'm still going to test out my theoretical setup using light fall-off to create the depth mask in-camera (testing with normal light and a monochromatic puppet at first), but I think Fusion + stereo shooting is ultimately the actual solution I'm looking for. Thanks Ernest!
A colleague of mine has been experimenting with the "stereo camera setup" that Ernest just mentioned. Here's a video explanation:
Basically, you'll need a stereo slider under your camera that will move it a few millimetres to the right for a second picture for every frame you take. Then there's software like this (http://www.yuvsoft.com/after-effects-plugins/depth-from-stereo-adobe/) that you can feed both streams of images into, and it will generate a depth matte based on parallax.
I wonder if this video (behind the scenes of Coraline) could be of any help to you:) If you're still working on a solution I recommend checking this out!
That stereo idea sounds interesting - might finally be a reason to shoot in stereo. (I thought I should try it, and shot my Harryhausen 90th birthday tribute in stereo, but really didn't get much out of it, and I usually go see the 2d version of movies now.)
You can use a 3d slider with Stopmotion Pro or Dragonframe, and it moves automatically when you take the frame. That takes most of the pain out of shooting each frame twice. But you can also do it for no money at all, with a home-made manual slider. That could be the way to try it out. I made one out of wood, with a screw to adjust the point where the slider stopped when sliding over from the left view to the right view. If I had wanted to do more 3d I would have bought the electronic slider.
So far the plug-ins for AE I have tested - Reel Smart Motion Blur, and Twixtor (for slowing down), and another slow-mo one I forget the name of, have been disappointing. The background smears and drags along, and in the end I couldn't get anything usable. I got a better results for some shots just using the built-in stuff in AE. For other shots, nothing quite worked well enough. I would expect that this would also be the case with this depth map generating software - they would naturally show demo scenes where it worked well, and not the ones where there were too many artifacts. But if there is a free demo version it would be worth checking out. I'm not naturally good with computer solutions, and especially not with After Effects, so others with more of an affinity might get better results.
I know that I would like to see a hansom cab coming down a cobbled street, emerging from the fog as it gets closer, and that's a very long object or collection of objects. The horse's head would emerge much sooner than it's rump, and that would come in well before the spoked wheels on the cab, or the cabby perched on the back. And it is not a series of flat planes, but a continuous gradation. If this could deal with that, it would be pretty amazing.