S.I.L Broken Glass – Making the Video

I’ll run through the making of the video below, hopefully it helps you out if you’re making a music video yourself or want to know the process behind it.


The first step was asking the band what they wanted, this is a difficult step. If you make videos, music or anything creative, it can often be tough to explain exactly what you want – try writing a funding application and you’ll see what I mean!

I was asked for something a bit weird, and a bit different. Not a straight up performance video. That’s all fine, I’ve seen some amazing videos that don’t involve the band performing. Huge fan of Ellis Bahl’s Breezeblocks video, for example.

Unfortunately, the budget was limited because the band crowdfunded to pay for the album’s professional recording, mixing and mastering, so not a lot of money was left for promotion. This ruled out professional actors, complex sets, lots of lighting gear, high-end cameras, and so on.

You’ll more than likely find yourself in the position of making something out of nothing with a music video, because new bands don’t have much cash to spare. Don’t judge – neither do you/did you when you started making films and videos!

I’m a bit of a believer in a new band being seen, to introduce their fans to them, and to give the members some personality, but the straight performance video is a bit simple, it won’t make anyone stand out. Still, I was keen to include the band in some way.

We met up for a chat and I threw out some fairly bad ideas (in retrospect!) just to get a feel for what the guys were after. They still didn’t want to be in the video too much, which is fair, but my ideas weren’t going down too well, also fair. It’s their song, and their call on how to promote it. Remember that if you’re working with a band. You’d want your videos promoted the right way, expect them to want that for their music.

I asked some questions – and these are very important – about how the audience should feel while listening and watching, what the song is about and should that be communicated? The band explained the story behind the song, but told me to keep it abstract to match the lyrics. No actual broken glass was another request!

I was all set to go back to the drawing board and look for a story again – after a few days already spent listening to the song and thinking up stories and visuals, but I tossed out one last idea. The visualisation grid you see in the  finished video. They liked it, after a quick drawing in my notebook was shown to explain properly.

The idea came from a grid used as part of a credits sequence in an anime, with similar shots to the Strokes video for Reptilia, and then the visualisation you’d see in a media player to give it some motion and make it a bit more interesting to watch.

It suited the band, and was abstract, and would allow me to show the guys without showing their faces clearly. It focused on the music, which is what they, to some degree, wanted, and fit their name well. Sound Intensity Level. S.I.L. A visual representation of the name, essentially. I’d love to say I thought of each block as a window, and broken glass meant that… yeah, I’m not that clever. Or else my subconscious is a genius. Who knows?

The idea was sort of an afterthought, I was listening to the song all day and then watched the show with that grid, and thought something like that could work, but with a bit more complexity and a musical twist. Then I thought a contrast of dark and light would be very striking on screen, so in my mind I gave the shots a white background. That’s when Reptilia came to mind, so I watched that for some visual cues.

The Reptitlia Influence should be obvious in this shot!

The Reptilia Influence should be obvious in this shot!

The final step was adding the graphic visualisation, I just opened real player, played the song and checked if the bars were suitably high and low at different points in the song. They were, the track’s dynamics worked perfectly.

I still expected a story-based idea to be the winner though, and not something I jotted down at 1am. But ideas are a bit random like that. It’s a case of listening to the song as much as possible and letting what you imagine combine with what you see around you, then praying an idea comes. It usually does late at night for me.

Through drawing a few tests I decided to go widescreen 16:9 for each block on the grid (4:3 was considered), and that 64 total blocks would be sufficient to give the visualisation enough motion without being far too much work. Adding more blocks multiplied the number of shots.

Shooting the video took a day, we used the front room of the bass player’s house, moved the furniture and got ready to set up the background. No background. Not good.

Somewhere along the line I’d forgotten to sort out a white background so I was left with a magnolia wall with a dado rail across it, right in the middle of half my shots.

The dado rail of doom.

The dado rail of doom.

Simple solutions to complex problems. I just got two lights and blasted them at the wall at full power directly behind where I was shooting. The original plan was to light the setup once, and leave it the same way for the day. This turned into moving lights for every shot, because I had to get more power on a smaller area.

Window light provided key light, and a softboxed light about 2 metres away on the opposite side gave a nice fill to the person playing, just enough to capture detail in the shadows and retain contrast.

With the background blown out and the subject exposed, the wall looked white, the rail vanished and the shots looked good. Not perfect, but I knew they could be sorted in post, which is often the best you can do when something goes wrong.

The equipment used was a Canon 5D mk II, a Canon 600D, 24-105mm lens, 50mm Canon F1.8, 50mm Olympus F1.8, Sigma 135mm f3.5 and my new favourite lens, Rokinon’s 14mm T3.1 Cine. It’s a thing of beauty. I just set the on camera audio to auto, for easy syncing.

I tried to keep the shots as varied as possible, knowing there would be a lot of guitar, and 64 total setups. With an assistant (Laura Emily Downes, fact fans!) running a second camera we were able to do around 30 run throughs of the song, with some shots set up to give two shots in the final cut. With 64 squares in the video, it wouldn’t cause a resolution drop to split shots in half, or even into quarters.

Then it was just a matter of recording from lots of angles. The room was physically big, but small as makeshift studios go, so I had the person playing turn to change angles, rather than moving the camera too much, which meant the wall was always behind them. As opposed to cabinets, family photos and furniture.

I marked off a note of what each shot looked like so I wouldn’t repeat anything, but inevitably some are a bit similar when you look at the guitars. Having the band play individually meant that there were absolutely loads of different angles though, and everyone got plenty of coverage. (Although singer, piano player and guitarist Damien kind of dominates.)

The shoot took around 6 hours total, including a break for food. Not too bad for 30 takes.

I tried to get some odd angles at times – there’s a shot of the kickdrum taken with a 14mm lens on a 5D mk II sitting between the drummer’s legs, for example, and shots like that give a little bit of variety to the video. I like the idea that people might watch more than once to see everything that’s going on.

Up close and personal with the kick drum.

Up close and personal with the kick drum. Rokinon 14mm T3.1.

The final step, the edit, took the longest. There were a number of issues that emerged during this part of the process that I wasn’t expecting.

First, my computer did not like having 64 video files thrown at it at once in my editing software. Grumpy git. I tried compressing them, but that didn’t work either.

Then there was the matter of creating the grid. Choosing where everything should go was tough without being able to visually build it in the editor, I had to watch video after video and put everything into a text grid in MS word. Not exactly ideal for knowing what the end product will look like.

I did some simple colour correction to brighten the background and give some punch to the colours. I shot pretty flat, but with all the visuals moving around as they are I decided more contrast was needed for clarity, and brightness was pumped into the highlights to keep the background wall pure white.

To sync sound I tried pluraleyes, but couldn’t get it to match multiple videos to one audio track. I think you can in Pluraleyes’ latest edition, but us PC users are still waiting for it to make the leap from Mac only. I synced manually. It took a while, but with music it’s not too hard because the waveform is easy to match up. The song was playing during every take, so remember – if you’re shooting performance, the band should be playing along to the track you’ll be using – and do record audio to sync it all up.

To actually be able to edit I exported a few videos at a time, slowly building the grid by adding more on top of a video with 8 or 9 of the individual videos already exported at once. This took 9 exports before I had all 64, but at least they were playing and the quality didn’t suffer, which was a huge concern.

I had planned a 4K export to have a crazily high resolution in each block, since I was shooting 1080p and showing it at a fraction of its actual size. This wasn’t possible with the amount of videos being used, my computer couldn’t handle it. Maybe if there were only 4 blocks, but not 64.

Once I had my grid I could see how the videos looked and of course, thanks to my text grid, guitar was beside guitar, the kick drum was in the wrong spot, the piano shots were all clumped together and it looked terrible.

I duplicated the main video file a few times, cropped down to individual blocks and moved these blocks around on top of the main video with all 64 on it. That solved the problem well.

Next step was to create a grid, so I just tried a few different widths for the lines, until I found something that looked chunky, but didn’t cover too much of the videos. Then I just did the sums to figure out how big each square should be, made a title with lots of rectangular solids and laid it on top of everything.

The final final step was to create the visualisation of the sound. To do this I screengrabbed real player’s visualisation and placed it over the video tracks, with 60% opacity.

Then I created 8 semi-dark rectangular solids at 96% opacity and placed them over each of the 8 sections of the grid, from left to right. Each one was individually keyframed to reveal the videos vertically matching the sound level of the song. So 8 times the rotoscoping action!

This part of the process took an age. I tried masking, but this was simpler, allowing me to know the 9 points the solids should be depending on sound level, and move them up and down accordingly. 540 showing nothing, 400, 265, etc. etc. all the way to -540 – which revealed everything.

The keyframing process was so time consuming that it took two weeks (though I had other projects to work on in that time) on top of a few days failing to get 4K to work and exporting the video in tiny pieces. I was inevitably frustrated, but perfectionism won out, and I spent the time making sure the video matched the real player visualisation to a T.

Then I thought I should try making the sound level slowly drop back down, like a peaking meter, rather than an audio meter, which took another day to test. It didn’t work, so I lost a day to that. Then another day to downgrading the video to see if any effects such as broken video, VCR look, stuttering, and so on, worked. They were just distracting, so another day gone.

When I finally had it finished I noticed a mistake from my cropping and moving of the blocks, and had to re-export, there’s always, always, always something!

And that was it, from conception to creation, Broken Glass by S.I.L

If you enjoyed the track, please check out their website: silmusic.com

And follow them on facebook and twitter.

You can follow me too if this helped you, I hope to do more blogs like it: http://twitter.com/landsleaving

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

  • Enter your email address to follow this blog and receive notifications of new posts by email.

    Join 359 other followers

  • Archive

%d bloggers like this: