Poor Man’s Process

The WidthScribe promotional video I recently completed for Astute Graphics involved the actress driving a car – except we ended up casting an actress who can’t drive. We got around this in a few different ways, including the obvious substitution of a qualified driver in the wide shots, complete with appropriate wig.

Perhaps the most interesting technique we used, and one which I might well have used even if she could drive, was Poor Man’s Process. Nowadays, most fake driving shots in films and TV shows are achieved by shooting against a greenscreen and replacing that screen in post with a moving background plate. A more traditional technique is to film against a rear projection screen – a screen onto which previously-shot footage of a moving background is projected in real time behind the actors. This was known as Process Photography.

Poor Man’s Process leaves out the screen altogether, shooting against a plain, ambiguous background that doesn’t reveal the lack of movement – typically empty sky. Careful use of camera movement and dynamic lighting create the illusion of movement.

Here is the set-up we used on the WidthScribe promo.

Making the magic
Making the magic

The car is parked on Nick’s drive, which is conveniently sloped so that – from the camera’s point of view – only sky and a bit of a distant tree are visible in the background.

A light behind the car represents the sun, and Nick chops a piece of cardboard up and down in front of it to represent the shadows of passing trees.

Low budget wind machine
Low budget wind machine

Sophie operates a hairdryer to blow Laura’s hair around.

Col shines a reporter light into the lens, moving it around to create the impression of the sun changing position relative to the camera.

And I dolly the camera side-to-side while vibrating it ever so slightly.

When intercut with wide shots of Nick’s wife driving the car for real, you’d never know the close-ups were cheated. (An additional trick we employed was to sit Laura in the passenger seat of the moving car then flop the image in post, for the over-the-shoulder shot of the pylon passing by.)

The drapes are to cut out the reflections in the windscreen.
The drapes are to cut out the reflections in the windscreen.

Poor Man’s Process works best at night, but with the shallow depth of field provided by DSLRs it’s now possible to get away with it in daylight too, so long as the shot is kept fairly tight and the road you’re meant to be driving on is fairly open.

You’ll want to vary the lighting effects you use according to the surroundings the car is supposed to be in. You can use spinning mirrors to sweep “headlights” or “streetlights” over your actors, or move a keylight representing the sun or moon slowly side-to-side, or even place two out-of-focus bulbs in the background of your shot to represent another car behind.

I’ll leave you with an example of Poor Man’s Process in use on a big-budget Hollywood film, Michael Bay’s 1997 Alcatraz actioner, The Rock. All the close-ups in the cars were shot static in a car park.

Poor Man’s Process

Double Vision

When I first started making amateur films with my grandad’s Video8 camcorder, the only “actors” I had available were me, my friend David Abbott and occasionally my sister. Even a little later on, when I managed to rope in a few extra friends, there were still far more characters than actors. The solution? Each person – with the addition or subtraction of a hat, jacket or pair of glasses – portrayed multiple characters.

While this massively confused the audience (which fortunately consisted of only my parents), it did teach me a thing or two about how to shoot scenes in which one actor plays two characters. Below is a run-down of the various techniques you can use next time you shoot a script featuring time travel, parallel universes or uncanny doppelgängers. Back in the nineties, when I was making amateur films, there was no way of combining two separate video images on screen at once (without a hugely expensive video mixer), so only the first two of these techniques were available to me. Nowadays they can all be done with a green sheet, a decent home computer and the right software.

1. Ordinary cuts

The simplest method is to block your shot so that one doppelgänger’s face is not seen, be it out of focus, turned away from camera, in shadow or whatever. Simply dress a stand-in in the right clothes and make sure their hair matches. Two over-the-shoulder shots, each with a stand-in providing the foreground shoulder, can be edited into a very natural conversation.

Katie Lake (foreground) doubles for Georgina Sherrington (background) as the character of Kate steps into her own past in Stop/Eject.
Katie Lake (foreground) doubles for Georgina Sherrington (background) as the character of Kate steps into her own past in Stop/Eject.

2. Hidden cuts

It’s possible to pan from one doppelgänger to another, without any post-production effects, so long as the pan is very fast – a “whip” pan. There is so much motion in a whip pan that the eye will not detect a cut in the middle of it. Alternatively, track the camera so it passes close behind a foreground object or character; use the moment of darkness as this object wipes frame to hide a cut.

A quick camera move behind a foreground wall masks a cut from Johnny Cartwright to Johnny Cartwright in The Picnic.
A quick camera move behind a foreground wall masks a cut from Johnny Cartwright to Johnny Cartwright in The Picnic.

3. Split screen

As long as your camera is locked off, it’s the work of seconds in post to create a simple split screen effect using your editing software’s crop tool. If a straight vertical line doesn’t suit your shot, more unusual matte lines can be created with a garbage matte filter. Watch out for changes in lighting when you shoot the two elements, particularly when filming outdoors, and beware of shadows crossing the matte line. If you can afford to hire a motion control rig for accurately repeatable moves, your camera doesn’t even need to be locked down.

Elizabeth Shue plays both old and young Jennifer in a split screen shot from Back to the Future Part II (1989).
Elizabeth Shue plays both old and young Jennifer in a split screen shot from Back to the Future: Part II (1989, dir. Robert Zemeckis).

4. Green screen

If you try to use technique 3 for a shot where one doppelgänger passes in front of the other, you’ll quickly find it a post-production nightmare of painstaking matte animation. Instead, shoot the foreground doppelgänger against a green screen. As always with green screen work, light carefully to reduce spill and match the background plate.

Peter Kay is green-screened over an element of himself and Patrick McGuinness in the 2005 Comic Relief music video "(Is This the Way to) Amarillo".
Peter Kay is green-screened over an element of himself and Patrick McGuinness in the 2005 Comic Relief music video (Is This the Way to) Amarillo.

5. Face replacement

This is the only suitable technique when your doppelgängers are in close physical contact and both faces are visible. The action is performed by the actor and a stand-in, who may wear a green hood with tracking marks on it. Later, the actor performs the second character against a green screen, wearing a green body suit, with the angle and lighting carefully matched to the earlier shot. This isolated face, or the entire head, can then be tracked onto the stand-in. (Alternatively, on a big budget, the actor’s head may be cyber-scanned and a CG version of it tracked onto the stand-in.) Beware that only experienced VFX artists will be able to pull this off convincingly.

One of the many face replacement shots in the climactic fight of the dimension-hopping Jet Li vehicle The One (2001, dir. James Wong)
One of the many face replacement shots in the climactic fight of the dimension-hopping Jet Li vehicle The One (2001, dir. James Wong)

The best approach is to mix as many techniques as possible, relying mostly on the simpler ones but hitting the audience with a more effectsy one every now and then to sell the doubles. Happy sci-fi shooting!

Double Vision

Top Five Low Tech Effects

Hollywood is known for its wallet-busting excess. James Cameron, Ridley Scott, Michael Bay, George Lucas and J. J. Abrams probably spend close to a billion dollars a year between them making their blockbuster movies. But all of these directors (or at least their visual effects supervisors) know that sometimes the cheapest, simplest trick is the most effective. Here’s my run-down of the top five best low-tech effects in massive movies.

5. Jedi camera tricks

George Lucas is a filmmaker who was at his best when constrained by clunky technology, and that was certainly the case when he made the first Star Wars film. The motion control work produced by the nascent ILM may have raised the bar for every movie to follow, but Lucas frequently had to resort to much simpler techniques to put the world he imagined on the screen.

Star Wars
Star Wars

To make Luke’s landspeeder appear to hover, a mirror was attached in front of the wheels, reflecting empty desert. Some optical work was required to complete the effect, but Lucas got 90% of the way there in-camera. Sadly his approach to filmmaking is now exactly the opposite.

Space Jockey
Space Jockey

4. Space Jockey kids

When Diddly Squat came on board Alien, his vision was to take the B movie script and turn it into an A picture. This was embodied by the huge Space Jockey which the Nostromo’s crew find in the alien derelict. The studio didn’t want to pay for the massive sculpture, but Squat insisted it would raise the production values of the whole film and he got his way. When it was built, however, it still wasn’t big enough for him. The solution? If you can’t make the Space Jockey bigger, make the people smaller. He drafted in his kids and the DP’s, had the costume department whip up some small-scale spacesuits, over-cranked the camera and – bingo! – the Jockey looks twice as big.

3. Sie haben meine green screen gestohlen

The cinema of Michael Bay is like a twelve-year-old boy’s wet dream: explosions, car crashes, toy robots and Megan Fox. His best work, I suggest, is his 1996 film The Rock, featuring a destructive car chase around the streets of San Francisco. Amongst all the money shots of falling telegraph poles, coin-spraying parking meters and exploding trams (“Where’s that son of a bitch at? I’m gonna hunt him down! That mother fucker ain’t safe nowhere!”) are close-ups of the characters driving which were executed in a way a twelve-year-old would be proud of.

Did Bay put the car on a low-loader and tow it through the streets? No. Did he put up a green screen and comp in the backgrounds? No. Did he go old-school and use rear projection à la 24? No. He simply parked the car in a place where only empty sky could be seen behind, rigged a couple of out-of-focus lights in the background on dollies, ordered a bunch of grips to rock the car and shook the camera for all he was worth.

This technique is known as Poor Man’s Process, and is most commonly used for night driving scenes. But Bay had the genius to see that in a fast-paced action scene, as long as he kept the camera moving (note all the zooming) he could get away with it. To my mind it actually looks better than if it had been done for real, because the technique forced him to make the camerawork frenetic, which adds to the energy of the sequence.

2. Masking an edit

Arguably the highlight of J. J. Abrams’ masterful third instalment in the Mission: Impossible franchise is the scene in which we finally get to see the whole process of making and applying one of those miraculous masks. In a single shot we see Tom Cruise put on the mask and Ving Rhames blending the edges until Cruise looks like an utterly convincing Philip Seymour Hoffman. There are many ways to transform one actor into another with digital technology, and while such technology was doubtless used to smooth out the transition, the shot has a very simple trick at its heart: as the camera tracks behind Rhames’ back, the momentary darkness is used to hide a cut.

Abrams employs the same technique in his equally brilliant Star Trek reboot, when Scotty accidentally transports himself into a water pipe in the engine room. Released from the pipe, he falls painfully to the ground and then gets up and dusts himself off in the same shot. A foreground pillar wiping frame conceals a cut from the stuntman to Simon Pegg.

Titanic
Titanic

1. Cardboard Titanic

Before he fell in love with performance capture, presumably as a result of getting the bends while shooting Ghosts of the Abyss, James “King of the World” Cameron was very canny with his visual effects. His films invariably ran the gamut of techniques, from the most cutting edge technology of the time to the oldest tricks in the book. So while Terminator 2 pushed the envelope with its computer-generated T-1000, Cameron was not averse to dressing Robert Patrick up in bacofoil for quick shots of the liquid metal villain in motion. And Aliens’ climactic nuke cloud was a cotton wool sculpture with a light bulb in the middle.

But perhaps Cameron’s most remarkable low tech effect comes in the second reel of his 1997 smash hit Titanic. The crew famously built a near-complete full-scale replica of the titular liner on the Mexican coast, while shots of the vessel at sea utilised a 45ft miniature surrounded by CG water and populated with motion-captured passengers (then a brand new technology). But when FX geniuses Robert and Dennis Skotak needed to put the Titanic into a green-screen shot of Jack and company playing poker in a dockside pub, they simply stuck a photo blow-up of the model to a piece of cardboard and dressed a Hornby train set in front of it. They called it the ship of dreams. And it was. It really was.

Top Five Low Tech Effects

Stop/Eject Trailer VFX Breakdown

The following post has been created and released because you lovely people out there have between you contributed over £900 to the post-production funding of my fantasy-drama Stop/Eject. Visit stopejectmovie.com to become part of the project if you haven’t already.

Since Stop/Eject is still being edited, work has not yet begun on the visual effects for the film itself, but the trailer we released in May features three representative FX shots and it’s these that I’m now going to take you through. First, watch the trailer if you haven’t already.

And here’s the breakdown video. A full explanation of the steps involved can be found below.

Dan’s Death

This is a key shot and was carefully planned. I wanted it to be in slow motion, but the Canon 600D we used can only over-crank to 50 frames per second if the resolution is lowered from 1080P (full HD) to 720P. I built this limitation into the VFX design.

Firstly a wide shot of the shop facade was recorded at 1080P25. This was deliberately done in the morning, when the shop was in the sun and the opposite side of the street was in the shade, in order to minimise the genuine reflections. Then Kate (Georgina Sherrington) was shot emerging from the shop at 720P50, with the camera mounted on its side and framed solely on the doorway. This slow motion element could then be placed within the larger 1080P25 frame recorded earlier, maintaining the image resolution.

Dan (Oliver Park) and the car were shot in a car park with a locked-off camera, the former at 720P50 to ensure his movements matched the slow motion of Kate’s, and the latter at 1080P25 with the car driving at half the speed it should have been. These two elements were combined with a simple feathered crop. The collision will never be seen, so he simply vanishes at the critical moment. I figured the lower resolution of Dan would not be noticeable once this element was composited as a reflection.

In Photoshop, I created an alpha matte of the shop’s windows and door. I applied this to the Dan/car composite, then layered it on top of the Georgina/shop composite with an opacity of about 50% to give the impression of a reflection.

The door’s alpha matte was key-framed to distort as Kate opens it. The final step was to animate this part of the reflection to pan sideways with motion blur and fade out as the door opens.

Rehearsing the "ghostly Kate" scene. Photo: Paul Bednall
Rehearsing the “ghostly Kate” scene. Photo: Paul Bednall

Ghostly Kate

Kate and Dan were shot separately for this, with the camera locked off. For both elements I created a difference matte in Final Cut, whereby the computer compares the element with a base image, in this case the same locked-off shot without either character present. I could then important these mattes into Shake and clean them up using rotoshapes and Quickpaint nodes.

Further use of these tools was made to animate the intersection of the two characters, revealing Kate little by little as she passes through Dan, all the time striving to give the impression that both bodies are three-dimensional. (Here I drew on many über-geeky teenage hours spent watching VFX shots in Quantum Leap and Red Dwarf frame by frame on my VCR, admiring the artistry and trying to figure out the technical trickery.)

After subtracting one matte from the other, I imported the result back into Final Cut and applied it to the original Dan element, placing the Kate element beneath to generate the final composite. A separate, static matte for the foreground records was used to layer them back on top, and an artificial camera move was applied to the whole thing to take the curse off the locked-off look. This move was possible without loss of image quality since we had shot 16:9 but were masking to 2.35:1, so we had surplus material at the top and bottom of frame.

Tape Archive

Not a tape in sight. Photo: Paul Bednall
Not a tape in sight. Photo: Paul Bednall

Hopefully you didn’t even spot that this was a VFX shot. Sophie spent many sleepless nights labelling hundreds of cassette cases for the basement scene, but even these represented only a small fraction of the number required to fill the master shot. Therefore it was always planned to lock off the camera and fill in the missing tapes digitally.

I started by exporting a single frame to Photoshop, where I used primarily good old copy-and-paste, plus a bit of airbrushing and a lot of distorting and resizing, to clone the real tapes many times over. I then imported this layered file into Final Cut.

Next came the most time-consuming part. Hold-out mattes had to be generated for Kate and Alice (Therese Collins) to keep them in front of the cloned tapes. These were created in Shake as rotoshapes and key-framed every few frames to follow the characters’ movements. Once applied back in Final Cut, the foreground characters and falling tapes appear to occlude the digital tapes in the background as you would expect.

That’s all, folks. Please keep the donations coming. We’re just £43 away from the £1,000 mark and the next public reward – the podcast covering day four of the shoot.

Stop/Eject Trailer VFX Breakdown

The Power of Post

Ray Bullock Jnr. as Joe
Ray Bullock Jnr. as Joe, preparing to fight the marauding demon

As we search for an editor for Stop/Eject, here’s a demonstration of the power an editor wields.

A few weeks before the premiere of Soul Searcher, my fantasy-action feature about a trainee Grim Reaper, I showed the film to my flatmates of the time. One comment concerned a scene in which the trainee reaper, Joe (Ray Bullock Jnr.), fights a demon (Shane Styen) while drunken revellers cheer him on.

As scripted, the scene covers only the start of the fight – enough to show how cocky Joe has become in his new role. (On the day of shooting we decided to extend the scene to show Joe killing the demon, but after a couple of shots Shane injured himself, forcing us to revert to the shorter version.) My flatmate wanted to see Joe kill the demon, and I decided he was right.

I certainly wasn’t going to do a reshoot at that stage in the game, but nonetheless I was able to alter the scene to have the demon die. I did it in three steps, and these conveniently illustrate the three prongs of attack you can use in post-production to change and improve your story.

  1. Re-purpose existing footage. The demon knocks Joe’s scythe from his hand early in the scene, so my first challenge was to get the hero his blade back so he could strike the fatal blow. Fortunately I had left the camera running while shooting a series of takes of Joe’s scythe hitting the ground. Therefore I had also caught the scythe being picked up on camera. This was never intended to be used, but it worked a treat.
  2. Visual effects. I’m no fan of digital fixes, but there’s no denying they can get you out of a tight spot. The existing cut of the scene had a shot of Joe walking up to the demon, but now I needed to put the scythe into his hand, so I cut out the scythe from a freeze-frame of another shot and motion-tracked it to Joe’s movements.
  3. Sound. This is the most commonly-used tool for changing things in post. Any time in a movie that a line of exposition is delivered without the speaker’s mouth being clearly seen, chances are that it’s Additional Dialogue Recording (ADR). It’s far easier to get the actor into a recording studio to perform a new line than to go back to a location or a long-struck set and reshoot. But in this case all I needed to do was put in some slicing, crunching sounds and a grunt, which I ran over a handy shot of the drunken revellers.

Here is the original scene followed by the bits I changed and then the final version:

Remember that with great power comes great responsibility. I’ve heard of actors who’ve found themselves edited into scenes they’d never shot. If you’ve substantially changed the character with your tinkerings, or placed an actor into a sensitive or controversial scene, be sure to discuss it with them just as you would have done beforehand if it had been shot conventionally. The same goes for the writer if you’ve made a big change to their script.

Some might look at all this as cheating, and I confess I have mixed feelings about it myself. I believe in getting things right in-camera, but the reality is that all films are prototypes, and you’re often well into post-production – at least – before you really figure out what the best way is to make this particular movie.

The Power of Post

Window Smash VFX Breakdown

Here’s one from the archives. This is one of the featurettes from The Beacon‘s long-forgotten DVD in which I break down a crude but effective VFX shot.

Compositing elements shot against black in my living room was an MO I heavily expanded on when I made my next feature, Soul Searcher, and you can see the extensive break-downs for that film by renting or buying the deluxe package below.

Window Smash VFX Breakdown

Ghost-trainspotting VFX Breakdown

As promised, here is the VFX breakdown for the main shot of the ghost train in my recent Virgin Media Shorts entry, showing how a crude model shot can work a treat, given the right compositing:

Due to the looming nature of the competition deadline, this was a pretty quick and dirty shot. Having said that, my VFX skills are pretty limited and I doubt I could have significantly improved it even if I’d had more time. It took a couple of hours to set up and shoot the miniature, and no more than an hour to do the compositing you see above. (I re-used smoke footage shot for Soul Searcher.)

As always, my approach was low-tech, avoiding any CG elements, and I did all the compositing work in Final Cut Pro. Soul Searcher had loads of shots that utilised this low-tech method, creating effects with everything from indoor sparklers to milk being poured into a fish tank. You can see a breakdown of all those effects as part of the Deluxe Package rental of Going to Hell: The Making of Soul Searcher. And remember, if you embed this video on your own site, you get a cut of any sales made through it.

Ghost-trainspotting VFX Breakdown

Acetate, String and Gaffer Tape

Shooting the train against black
Shooting the train against black

Last weekend I recorded the model shots for Ghost-trainspotting. The script was written with the train miniature from my 2005 feature film Soul Searcher in mind.

This model was built by Jonathan Hayes, with the wheels and linkages by my uncle and my late grandfather, and even some detailing by Yours Truly. If you’ve seen Going to Hell: The Making of Soul Searcher you’ll know what an absolute pain in the arse this train was to shoot back in the autumn of ’04. You can also check out the key blog entries here, here and here to experience the pain.

I’m pleased to report that filming the train this time around was considerably easier. The model has been sitting in my parents’ loft conversion for the last eight years, and that’s where I filmed it. All I had to do was rig some black drapes and line up the camera angles to the live action plates.

In the past, whenever I’ve needed to shoot a VFX element that had to match another, pre-existing element, I’ve used acetate. I would play back the existing element, typically a live action plate from principal photography, on a TV or monitor. I’d tape a sheet of acetate – the stuff teachers used to use on overhead projectors – over the screen and draw around the important landmarks with a felt-tip. Then I’d plug the camera into the same TV and adjust the camera angle until it matched the lines on the acetate. Low-tech, but effective.

Express Train to Hell
Express Train to Hell

Thanks to Magic Lantern – a firmware hack for the Canon HD-DSLRs – there’s now a more elegant solution. Magic Lantern provides a large number of handy features that Canon themselves were too tight or too lazy to include, but there are only two I regularly use: manual white balance, which lets you dial in the colour temperature in degrees kelvin, rather than relying on presets or holding up a white sheet of paper, and crop marks.

On Stop/Eject I used the crop marks to show us the  2:35:1 widescreen aspect ratio we were framing to, but you can actually overlay any monochrome image on the viewfinder by simply creating a .BMP image of the correct specifications on your computer and saving it to your SDHC card. So for Ghost-trainspotting, I loaded still frames from the two live action plates into Photoshop, and outlined them with the paint tool. After saving these outlines to the memory card as bitmaps, I could superimpose them on the Live View image on my Canon 600D’s flip-out screen. This made it a doddle to line up the shots.

Left: live action plate. Right: overlay for Magic Lantern
Left: live action plate. Right: overlay for Magic Lantern

So, the model shots are now in the can and yesterday I incorporated them into the film. I also tightened up the edit – finally getting it down to 2’20 exactly, including credits – mixed the audio and graded the images. The film is very nearly finished now; I’m just trying to decide whether to change the ending….

If you’re lucky, there may be a little VFX breakdown of the train shots coming in the next week or so. Meanwhile, here’s a nickel’s worth of free advice about shooting miniatures:

Acetate, String and Gaffer Tape

Wasteland FX walk-through: Empty A50

Here’s a look at how I created one of the FX shots in the Wasteland trailer. Tom, the director, wanted to see the central character, Scott (Shameer Seepersand), walking along a deserted dual carriageway to show his isolation in the post-apocalyptic world.

We started by shooting the A50 from a bridge (image 1). Traffic was light, but we were never able to get a clean, car-less shot. (While filming we were interrupted by a couple of blokes from the Highways Agency who wanted to cover their arses in case of accident, so gave us a safety briefing: “Be careful when you cross the road, lads.”)

The next day we drove around looking for a footpath or cycle path with similar tarmac to the A50 which could be shot from a bridge to get the same elevated viewpoint. Having located one, we filmed Sham walking (2).

Shrinking Sham down and feathering the edges of his element into the shot was the work of moments (3), but the illusion wasn’t complete until this element had been colour-corrected to match up the tarmacs.

Next the cars had to be erased, which I achieved partly by overlaying cropped sections from later in the footage (when the cars had moved on, leaving a space behind) and partly by exporting a frame to Photoshop and using the clone tool. The former technique is preferable because using a motion element retains the movement of picture noise; the lack of this movement can be an FX giveaway. Therefore I kept the Photoshopped sections small – just way in the distance where the cars were tiny.

At this point (4) the shot is essentially complete, but I added some extra touches in the form of smoke elements from an FX library, a faint one in the distance (top left) and one on the van which I had left frozen in the picture as if it had been abandoned. I also duplicated the van’s smoke and distorted it to create a shadow for that smoke (5).

All images copyright 2011 Light Films.

Wasteland FX walk-through: Empty A50