The official vid – we got commissioned to make this by the label, which was terrifically exciting, espesh as we’d seen JFB DJ years since. Anyway, this was made over just-over-a-couple-of-weeks.
The puppet was made from a thick sock, carefully selected and bought in a two-pack from Marks & Sparks, and a couple of ping pong balls, bought from Argos in a pack of far-more-than-we-needed. Ping pong balls stuck in place using some PVA glue we already had knocking about, and the eyes drawn on, et voila. Our star.
Filmed on our new posh Canon 550D, in front of a makeshift greenscreen made from a stretched green t-shirt. And this really didn’t work as easily as we’d have liked – not enough light from internal lights, we needed to shoot in the day, and it was a right pain in the arse to greenscreen – we thought it’d be easy, with a posh camera now and all that, but no. Still a tricky fucker to key the right bits out and not cut into the other bits. And a bit of a pain lying on the floor fannying about holding a puppet in the air. To get the best results, we shot in the highest res available, and in daylight, and other stuff. We led there through the entire track playing twice-over, a few times, dancing about.
Our original idea was far more grandiose, but time and competence pressures meant it got pared down. We had some idea of what would happen – in the treatment we pitched it outlined the basic steps of the puppet walking in, then moving around the edges, then multiplying, then kaleidoscope, before standing in front of a crowd, then the crowd vanishing. But we didn’t have it planned in detail, so the filming was just some basic walk in, walk out, and a lot of standing there dancing.
It became clear as we started that we’d be best filming it static-ish, centre frame bobbing up and down, and not move it around the sides or back and forth much at all. Also, all at the same point, not moving it back and forth much. All that, we’d do in the animation.
So… First things first, but not; the backdrop. The AE puppet tool is a bloody amazing wee thing we only found out about recently (forget if it was specifically when looking for something to do what we wanted for this vid, or just unrelatedly beforehand.) So we used that for all the warping business, along with a bulge warp. Various mucking around with keyframing it all.
Ooh, and firster things first. We took the flat hi-res JPEG of the record cover we were supplied with, and added the two bars either side in Photoshop so we could show it all and fit a 4:3 trad TV aspect ratio. And we cut bits out. We were originally wanting to do a lot more of this, platforms and elements made of and in the style of the cover, for characters to run around and be backdropped, but time meant we dropped all that. We cut out the ring, cloned bits to create a full ring, animated it rotating in AE (not so simple, as it’s not a clean perfect circle, so some resizing was also needed in the animation), with layers to keep the recylcing sign on top. The warping is only applied to that composition.
The splitting apart, fairly simply done, but we made effort to keep the ring rotating. Something cocked up, so in the final comp we added a flash on a white frame to cover a join where the ring anim jumped as we moved to the split comp.
We used 3D frames and depth of field, trying to approximate the camera settings despite not quite knowing what we’re doing, so the main puppet sits in the focussed bit, with the backdrop slightly out behind it, and puppets moving in and out of focus as they move back and forwards. One big thing we wanted, and helped partially achieve by using this, we think (if anyone notices,) is that it doesn’t just look very ‘AfterEffects’, with cutouts resizing – we wanted it to look more solid and authentic.
So, a load of different bits of puppet composited together and moved around at the start. A couple of simple bits in the middle.
The big rows shots were a pain because our computer started to creak a bit (it can’t handle HD video, really, and lots of instances of nested comps caused trouble.) But they’re composed of a row composition (itself composed of multiple instances of a greenscreen keyed composition) stacked over itself in different bits. And they are a big bit where time and experience constraints cut our ambition – simple block colour backdrops. And a lot of time saved using a simple AE effect of the mirror, but animated whipping around, and then a second copy of the comp overlapping.
The big close-up bits we again cheat on the authentic aim by just using a simple resize on the semi-opaque one. We did always aim to do a mixing of looks-authentic-for-1950s-TV with AfterEffects-trickery – that was the heart of the idea – so it’s not a big betrayal, but it was very simple and looks as simple as it was.
We tried to avoid reusing footage – all the main puppet stuff never repeats, but the shots of the rows will have been used more than once as that comp was only a few seconds long.
(This is horrifically rambley, isn’t it? More than usual.)
So, quickly. We wanted to be authentic, so looked up the resolution of 50′s TV! Ludicrously, we cocked it up and got out horizontals and verticals mixed up, so this was actually done to a lower res than we should – tsk.
We applied a bunch of effects on the final thing – a bit of grain, a load of bloom, a subltle bulge, something that darkened the edges (may have been the bloom…) and a ‘wiggled’ movement on both the position (full comp sized so it was slightly oversized, so it could move around a bit) and the opacity, to give that old footage flicker.
We exported it interlaced, then imported that footage and exported in HD res. We didn’t do artifical scanlines, as they’re a pain and any tutorials we saw looked shit, and sadly AE smooths out the interlacing a bit when you blow up the size. This meant the effect looked a bit accidental, like a bad encode – as confirmed when the chap from the label questioned it. So we re-did this final step, using AE to blur the interlacing (via a filter, not the other method) to have an authentic-ish approach to how you’d deinterlace actual old footage, maybe.
This was quite the palava, working much of our spare time, but although we did have to scale back our ambitions, we reckon we did a good job of making the most of it, and are happy with and proud of the final vid.
http://www.superluminescence.com/games/blackmirror
This was done quickly, over a few days, in the week after the first episode of ‘Black Mirror’ was on, and ready before the second episode was on. Still, for a cheap, confused gag, it took too long. The idea was to just do a quick pseudo-game, using the structure of Obstacle Climbing Bastard. It turned out to be quite a bit trickier, the main problem being to make it so you couldn’t win by just holding the spacebar down (not a problem with OCB, as there you got punished for pressing the button on any ‘off’ frame.) Forget what we did, but we sorted it – we think we just added extra variables to record when the button had been pressed, rather than straight just listening for the button press.
The graphics took a bit of fannying about. Unnecessarily composed to be vaguely authentic – 8×8 blocks a la NES/MasterSystem, composing with Photoshop’s preview of pixels at a 1.2ish aspect ratio, outputting to twice the size, adding a layer of scanlines (forget what we went for, but some sort of darken blend method to the layer,) exporting then as square pixels because you seemingly have to, but then stretching the final SWF in the HTML embedding. Aaand we did all the bits and bobs in one Photoshop file, turned the appropriate layers on and off and outputted each whole frame as a PNG – so the SWF is inefficiently working through entire fullsized PNGs, rather than compositing different screen elements. It’s all a big cheat.
Music, since upgrading to Snow Leopard, we can’t run our old Garageband with the music instruments and settings we used for OCB et al, so the sound here was done in NanoStudio, exported, cut up and downgraded the bitrate in Audacity, and again downgraded in the final export. That clipping sound where the beep loops is accidental, but we’re lazy and quite liked the horribleness of the effect.
So, the ‘game’. It’s supposed to be grim, morally ambiguous, and ultimately a bit empty and pointless, with a pompous finger-point at the audience at the end. A-ha ha. How post-modern.
There is an ending. Just hammer the spacebar like you’re playing Track & Field, and you’ll get there, after an (intentionally) gruelling amount of time.
WHAT AN INTERESTING MORAL STATEMENT.
(Didn’t dislike the programme as much as this may suggest – fairly ambivalent about it, really, and thought making a game out of it might amuse C Brooker and get attention. It didn’t.)
Playing catch-up; we did these ages since. Fairly cheap Photoshop jobs, but hey ho. They’re like the Total Recall and Commando ones. The ones we didn’t put up here, apparently. Hrm.
A-ha! Iron Man, in Irn-Bru colours. Hilarious. Done quite lazily and imperfectly – forget how, probably wand selection by colour, and don’t know if it was fannying about with the hue or using colour layers. Whatever. The logo was a piece of piss too, just removing an O and adding the hyphen, but hey, nice hyphen, right?
A-ha! Instead of the pic of Ceasar’s head, it’s the Monkey thing from off of those adverts and that. Very cheap, but quite nicely done, just cropping out the Monkey head and fannying about with some image adjustments, darkening or ‘burning’ some of it, and that’s it. Changed the tagline from ‘Evolution Becomes Revolution’, too, for some reason.
Ah, now this one, though, inordinately proud. Good joke! It’s the Tinker Tailor Soldier Spy poster, but with Tinker the character off of Lovejoy! Golden. And done pretty fucking well, if we do say so. Go and have a look at the proper poster with Gary Oldman’s face, then look at this. It’s nicely done, innit?
Found best picture I could of Tinker, did a bit of cheap contrast and darkening stuff to make it stand out a bit better. Created a page full of numbers, with some Lovejoy words hidden in there (the actual posters do something like that – ATTENTION TO DETAIL.) Some sort of filter (can you tell we’ve forgotten the specifics) to pull out only the Tinker colours. Um, and a background gradient, with the same numbers, cut out and in opposite. And, cut out some of the bits in blocks so it sort of breaks up at the bottom.
Should have replaced Gary Oldman’s name, on reflection. Hey ho.
So, this’ns a competition entry, for a Vimeo/BUG/Moby/Saatchi&Saatchi(erk) competition called ‘Hello, future.’ Wonder how it’ll do.
Cinema 4D, again. MoGraph dynamics, again. Some text, again. Global Illumination again.
Was done in a few discrete bits, because my computer couldn’t handle it all together (starts struggling at the end of the first bit.)
Bit one – camera moves up, text dynamics switch on, a fracture object falls and lands on the one below it, light moving around is only acting on one section of text, but GI reflected light gives a nice gradient on the other stuff.
Bit two – static sections of text appearing, a light source each, then one for the lot. (All light sources infinite.)
Bit three – dynamics being switched on again. Global setting changed to make the gravity negative. Else, same as bit one.
Bit four – a light source acting on everything, with a visible floor and, for the first time, sky. Sky and light raised together at the beginning. Text (as always, extend NURBS, these not in fracture objects) rising up. At the end, light and sky falling together, and the final bit of text with its own light.
Camera just doing keyframed movements, and as with the lights, some points set linear some spline. Bit of a cock-up between bits two and three meant they didn’t line up, so I rejigged the points there so they did.
Assembled the different exports (it took days to render in total) in FCP. The only FCP noodling I did was a keyframed blur filter on the beginning and end of bit four (and ended up cutting short the very end of bit four.)
And hey, finally found out that to get FCP to allow you to play footage on the timeline without needing a render, you just have to set it to the same res and have SQUARE PIXELS. Only taken years and years and years to stumble across that, but hey ho, here we are, and it’s good to know.
The final export I stuck slavishly to Vimeo’s recommended compression, H.264 and all, and HD, and it looks like compressed shit. And took an hour to export, madly. But hey, it got up there in time for the deadline, which is something.
Ooh, look, an embedded Vimeo vid! Yep, we put this on there, because for once we don’t have to worry about any tedious copyright infringement stuff.
So, music made in NanoStudio. Pretty simple – using three tracks, different loop on each, all using just the basic undefined synth for 8-bit sounding funtimes.
Forget what kicked off this idea (other than procrastination, obv.) After Smashcube, did some more playing about with glass cubes and physics. Lead to the idea of loads of them falling, from an emitter. And then natch, that lead to the idea of pulling back to reveal the cubes being spat out of a massive Japanesey cute… thing.
And then we’re here, all in white. Cube’s just a cube. The monster thing is a very simple edited sphere, given a bit of depth and then hyperNURBed, and arms and feet and eyes stuck on. Texture applied to the body just to paint the interior black.
Two light sources, one blue, one pink. Don’t show as specular, and a non-light-casting white for the specular in the eyes. Artificial, but looks nicer.
We were thinking of using the white infinite light, but wasn’t necessary. We have a white floor and sky, Global Illumination, and a white environment with fog.
Emitter inside the mouth for the particles, and MoGraph tags on the particle source, which seems to do just what we wanted. Only problem was it seems pretty much impossible (on a little bit of trying) to bake the particles, because you’ve got a particle emitter emitting MoGraph instances… Couldn’t find a way to do it that didn’t switch off the particles or fuck the physics, but then we’re no experts.
Not sure we didn’t fuck the physics slightly, as inititally don’t think there were any rebounds, whereas now you can see a few going back into the mouth. We put a MoGraph tag on the body to make sure no cubes were flying through the body, and rebounds are better than that, but we thought we’d tested it and sorted the emitter’s spread so it avoided the edges enough. But maybe the cubes are colliding as soon as they’re generated, and rebounding, and so being a bit unpredictable and uncontrollable. Anyway, it’s fine.
Did fancy it showing a continuous build up of cubes, but this proved too intensive. To get a decent number falling and to look good coming out of the mouth, it’s at something like 100 per second. Those soon add up and grind render times to a halt if it’s preparing collisions of hundreds of objects. So instead we’re just in each shot showing the same period of time (give or take, as the shots are prepared all the same length from Cinema 4D, but then cut down and edited in Final Cut Pro.)
And the motion of the colliding cubes was bloody fast on normal settings – they would whizz past at ludicrous speed. It just looked a bit crap. The bounceyness was taken (we think) just the same settings as we’d found worked nicely in the glass smashing vids, perhaps with less friction on the floor. But noodling around with various settings didn’t help slow things down, it always looked really fast and therefore small and inconsequential – we wanted it to be like these cubes are big and smashing down on the ground, falling from a huge height. Solution we used was to arrange the stage at 25FPS, but render it all at 100FPS.
Not sure we fully achieved the sense of size we were going for – in some shots better than others. As well as the fog environment, we also used for the first time depth of field – same setting on every camera, to just give focused foreground going into a blurry background. Does some nice things when you have objects like the cubes falling from the background into focus. And in a bunch of shots kept the camera low, sort of half way up a cube’s height, looking up.
Edited in FCP, early on thinking of doing long shots to coincide with the first music loop, then shorter shots for the more complex instrumentation (i.e. two loops) and slowing again when the third loop comes in… But we found ourselves not wanting to cut down, or intercut, these nice long tracking shots we’d prepared. So stuck to the full 4 bar length until the end, where we go a bit clip-show and recap a few shots. There’s not much of a shift in action, but starting with just the cubes, then revealing the source, then static shots for the change in the music, then recapping shots when the music recaps… It’s only that recapping that is the especially lazy bit.
So, we first did a bunch of angles and camera movements. Exported quick hardware renders. Did a quick edit. Then exported 640×360 full renders, which took 5+ hours for most because of the GI and shadows and everything. Then subbed those in to FCP, and then worked out what other few sorts of angles and movements we needed, then prepared and rendered those.
All in all – with the mad render times, leaving the computer rendering when we went out and stuff to get them done – it took a week. Or just over. Started previous weekend, made everything and set it up. The week was rendering. Then last weekend finishing up.
But all in all, pretty good, right? Feels like a proper thing, doesn’t it? We think so.
