Jan 092013
 
 January 9, 2013

IMG_4047

This was a tough one, make a music video starring 50-100 people, who were only available in groups of 10-20 at a time, had no specific talents, and had no experience with acting.

That was the brief with this piece. A bolt on video to a live event we were planning, that involved all of the staff of the end client.

So the concept had a few constraints from the go, but a miniature style piece using tilt-shift photography seemed a perfect fit.

IMG_4041

To achieve a successful miniature look, you’re shooting from afar, with a high angle, often focussing on many people/objects at one time, with a shallow, selective depth of field.

This long distant style of shooting, along with shallow focus gave scope to use many different ‘actors’ and extras, without relying on facial expressions, or even relying on the same person to play the same role.

IMG_4119

Tilt-shift photography has fascinated me, much like anyone else who has scoured Vimeo over the last 18 months. There is certainly an abundance of videos online that uses the effect, and this has been much of the reason as to why I’ve been hesitant to use it in the past.

But with criteria that lent itself so well to the effect, I felt no other style suited it better. I have also struggled to find tilt-shift photography pieces with a narrative, the majority seem to be past-time pieces or at best, experimental videos. So a music video with a narrative seemed a refreshing format to bring to the effect.

IMG_4122IMG_4113

I got sign off for the script, but before we jump forward to the production of the video (where my involvement in the project continued) it’s worth touching on the intense pre-production that took place in able to make the video possible.

It was nothing short of a logistical nightmare, sourcing good locations with high vantage points in and around Portsmouth. I didn’t envy the Producing role of my colleague Laurie Gibbs. Who spent weeks on the phone clearing locations with the council to film at, to get on the rooftops of, or squeeze a 27M cherry picker into. It’s a story in itself, and he did a smashing job! But I’m going to focus on the production of the shoot.

IMG_4082

IMG_4088

So the first thing to consider, was what this was going to be shot with. There were a few considerations.
– It’s going to be shot in a time-lapse fashion (a successful miniature look requires sped up motion)
– It’s going to be premiered on a huge 20-foot screen at an awards show.
– It’s going to require a small footprint, as a lot of the shooting will be from a 27M cherry picker, which sports a small metal framed carriage.
– Tilt-shift lenses were required, so the format must be compatible with EF mount
– There will be 11 locations over 3 days, so setup must be light, and rely little on power as each setup is quick and therefore power access is limited.

IMG_4056

I did initially considered the Red Epic. 5k resolution, EF mount, good size. But the cost, weight and DIT requirements were a drawback.

There was also the Sony FS100/FS700, but these lacked a little in the resolution department, and EF compatibility is only achieved via a dumb adaptor or hard to find/hire metabones adaptor.

I then started experimenting with DSLRs. The immediate problem was that the video function lacked far too littler resolution for the final output, but luckily we didn’t need to be shooting real time, so the photo side of a DSLR became much more attractive.

In high burst mode, the 5D mark iii shoots 5fps. This means that a photo sequence shot in high burst mode, which is then converted to 25p video, would give you video at x5 speed with a resolution potentially exceeding 4k.

A few tests concluded that x5 speed would work great in regards to the motion of subjects when considering the miniature style. (In hien sight it was a little fast for a few applications, more initial testing may have proved this).

IMG_4090

So the 5D mark iii on paper was the perfect candidate; huge resolution, very lightweight, cheap, perfect EF compatibility, little to no power requirements).
The only problem was operation. In high burst mode, the first thing that happens is the mirror flicks up, blocking the viewfinder, and the output via HDMI is disabled. Shooting blind was far from ideal!

So I came up with a solution. Two cameras next to each other, with matching settings and framing, one for monitoring, one for capturing. May sound a little strange, but was perfect for achieving cheap lightweight high resolution, without losing the ability to monitor what was happening in frame.

IMG_4107IMG_4076IMG_4101

I made a rail system with a few manfrotto and Jag35 bits I had lying around, so that two cameras could mount onto the same tripod (small footprint). We then hired in another 5D mark iii, a tilt shift prime set; Canon 24mm TS-E f/3.5 L, TS-E 45mm f/2.8, 90mm TS-E f/2.8, and a canon x1.4 extender.

The first camera utilised the tilt shift primes, and with the x1.4 extender, I was able to achieve 6 different focal lengths between the 3 lenses. The second camera then used a 24-105mm to match the framing, and careful note of the plane of focus was made.

At the time, it felt like a wiser choice to go with the 24-105 rather than another tilt shit prime set. This was for two reasons. First cost, second (and more importantly) amount of gear. With potentially 6 lenses (2 tilt shift prime kits) it would’ve meant that all camera gear would not have fitted into one Pelicase. This was a big deal when considering the small carriage of the cherry picker, so the 24-105mm saved us a lot of space.

IMG_4068IMG_4033IMG_4028

If anyone has used a tilt shift lens you know of the endless possibilities they can produce. And looking back there were occasions where the framing of the 24-105 simply didn’t have the ability to match what I was able to produce with the tilt-shift lenses. But the conscious note to keep the gear small was a good one; we would have really struggled with additional gear in the cherry picker!

The setup was topped off with x2 SmallHD DP6s. Best field monitors I’ve used to date, and worth they’re weight in gold on a production like this (especially when you have a director on the ground that needs to see what’s going on).

IMG_4073IMG_4040

IMG_4035IMG_4036

The monitors also used the same LP-E6 batteries as the cameras, which was great. 21 batteries kept both cameras and monitors alive all day without needing to re-juice any (although we did have a procedure put in place for re-charging at mid points during the day).

IMG_4026IMG_4029

The post-production was a breeze. It was refreshing to edit with photo sequences that hold so much colour and latitude. You could manipulate them so much and they’d still hold up fine.
For shots that required moderate colour editing I processed them through Lightroom as sequences before converting to video. For the rest I processed as video in Quiktime 7, then added contrast and saturation in FCP 7.

When shooting high up, particularly from a cherry picker, you’ve got to consider the weather. We knew we’d face camera wobble, especially at high heights around the coast.
Due to the format we we’re shooting in (JPEG Sequences) we didn’t have to worry about attributes which normally collide with image stabilisation, such as CMOS wobble. As a result, Warp Stabiliser in the Adobe CS5.5 package worked an absolute treat. There’s no doubt in my mind that some of the footage we came back with would’ve been un-usable if shot differently (i.e. using the video mode on a DSLR).

IMG_4109

I think the project came together really well as a whole. There are definitely points for improvement. Directing and timing scenes that are x5 faster than real time were a challenge, and I feel this is reflected in a few scenes where certain actions needed sustaining for longer, or cutting short.
Hope you enjoy

IMG_4038           IMG_4094

 

  One Response to “Old Flame Alight Music Video”

  1. [...] One member of the production team, Tim Fok, wrote up some production notes to give some insight into how the music video was shot. You can read it on his site here. [...]

Sorry, the comment form is closed at this time.