TTB Dads Rock On from Steve Green on Vimeo.
I have just completed a music video for The Tall Boy: Dads Rock On from his latest EP, Take it to the Kids. I thought I would share with you the details of the way I put it together.
Filming
The basic idea for the video is a pretty simple one. It was Matt’s idea, I can’t take any credit for it. The idea was that he should do some embarrassing Dad dancing to the song, and that I should film it.All filming was done using my trusty 550D with the 18-55mm kit lens. I use the Cinestyle Color Profile with the settings as per the recommendations to get as flat a picture as possible. I have also recently installed the Magic Lantern software. I turned on the zebras, to get the maximum dynamic range out of the sensors. Magic Lantern also gives you good control over the ISO range, adding a whole load of intermediate steps that you don’t get from the out of the box camera.
I had my Rode Videomic Pro on the camera - not strictly necessary but I’ve only just bought it and I wanted to check it worked. Incidentally, I have seen a few reports on the Videomic Pro that suggest that the battery is incredibly difficult to change quickly - I don’t know whether they have changed the design but I tried it several times and it couldn’t be simpler. Perhaps they have changed it
The camera was set on a tripod for all the shots except the last two - the feet close up and the top half of the body where I wanted to make sure I got the arm movements. With the IS on I find I can keep movement down to a level where it’s not distracting. I’ve got a huge problem with the modern shaky cam style of filming - why when lighting, hair, make-up, sound etc still has to be perfect, is it acceptable to make it look like the camera is being held by a monkey?
Lighting was a lot of natural light (a large bay window behind the camera position) supplemented by my Neweer LED light with the orange filter over it. With the camera white balance set to Shade I find that this combination gives skin tones that are pretty close to the skin line without the need for a massive amount of tweaking during the post production phase.
Importing
I imported the footage onto my Macbook Air by popping the SD card straight into the card slot. I’ve experimented with various ways of doing this but for this project I made sure that Final Cut Pro was running before inserting the card, which makes the FCP import window open automatically. I think that, now I’ve figured out the best way of backing up my work from FCP libraries, this is the way I’ll keep doing it from now on.Editing
The first step was synchronizing the takes on the timeline. After fiddling about with the multi cam option for a bit I finally figured out that FCP tries to use the record time to put the clips onto the timeline, even when you tell it to use the audio to match them up. Luckily, the Synchronize clips option works a treat - I simply selected all the clips, together with the sound file of the track, pressed the button and voila! a perfectly synchronized compound clip.It was at this time that I noticed my first mistake - you can see the lighting stand in one of the key clips (where Matt is playing the guitar). A quick round trip to Motion was required to fix this. Integration between FCP and Motion is non-existent so the sequence was:
- Right click (or 2 finger click on the magic track pad) on the clip in the FCP browser and select Reveal in Finder.
- Open Motion, create a new project and Drag the clip onto the time line
- Create a copy of the clip
- Mask the lighting stand on the top clip
- Slip the bottom clip over so that the background wall replaces the lighting stand
- Blur the edges of the mask so I couldn’t see the join.
- Use share to write the new file out to disk, accepting the default settings
- Switch back to FCP, import the file and drag it into the multi clip.
- Using the audio waveform as a guide, synchronise with the audio track (this was really easy in this case because the track starts with two clicks)
- Delete the old clip with the lighting stand visible and we’re done!
The next step was to separate the video from the audio for all the tracks and delete the unwanted audio. After doing a few I discovered that separate audio works if you select multiple tracks so I did that. Note that after you done this you have to be really careful not to time slip any of the tracks as you no longer have a reference point to resync. I didn’t have a problem in this case but I suppose an alternative is to mute the audio on all the tracks you don’t want to hear. There’s probably a clever way of doing this that I’ll have to figure out in the future.
So now I had a big stack of tracks on top of the audio track and I had to figure out how to cut them. This was where I made my second mistake - I should have done the basic colour correction before starting to cut - it would have saved a bit of time, but I didn’t think of that til later.
Dads Rock On as a song is in two parts - very cleverly linked by the chorus. In the first half our hero stumbles across a gig being played by one of the heroes from his youth while in the second half he decides to get his own band back together and get out on the road. I decided that the first half should mainly feature the Dad dancing elements with short stabs of close up guitar action while the second half would reveal our hero as the singer and guitarist, intercut with more stabs of dancing. I decided that the close up of his face should be reserved for a couple of key phrases to emphasise the journey in the song. The next phase was all about selecting the right clips in the right sequence.
The way I did this was to make all the tracks except the one I wanted to use invisible, then cut that track roughly on the beat. I would then select where I wanted to the clip to end, cut again and then make the before and after sessions invisible. I decided to do this, rather than delete the unwanted sections as FCP tends to drop tracks that are above the main timeline down to the lowest possible level. This would have resulted in a jumble of tracks. By just making the tracks I didn’t want invisible I kept the clips lined up, like a conventional editor. This turned out to be a good decision and I would definitely do this again if I was working with a multi cam or synchronised track.
Once I had got all of the right clips I wanted, I concentrated on getting the cuts exactly right. Dads Rock on is a very beaty track so it made sense to cut very rhythmically on the beat. This took several passes through until I was sure that it was ok.
Colour correction
Once I was happy with clips and the timing, I did the colour correction and the shot-to-shot matching. As I said before, this would have been easier if I had done the colour correction before starting to cut, but heigh-ho, lesson learned. In doing this I use the workflow suggested by Denver Riddle on his excellent Color Grading Central website. That is, for each clip:- Optimize the dynamic range using the exposure controls. Generally I set the low and high to give maximum dynamic range, then use the mid to bring the skin tone back in to line (in this case Matt’s skin was around the 50% mark)
- Correct the colour balance for skin tones. In this case it was pretty much bang on, thanks to the orange filter on the LED light. There was just a very slight green cast to the shots, probably due to the green curtains.
- Next shot-to-shot matching. I probably should have followed Denver’s advice and used a secondary for this but in most cases I just messed with the primaries until all the shots matched the look I was going for. The only shot I had to use a secondary for was the head shot, where for some reason there was a much lower contrast between Matt’s face and the background. For this I used a secondary with an oval mask and just raised the exposure on the outside until it matched the other shots.
Applying the look
Finally, it was time to apply the look. This was when I realised the power of compound shots. When you drag them to the project timeline, they appear as a single clip and any effects or colour correction apply to the whole clip rather than the individual shots. This is incredibly powerful and, in fact, better than other NLE’s where you generally have to apply looks on a track-by-track basis.Anyway I thought the FCP Indie Red effect might suit the video so I whacked it on. Although I like the effect, I thought it boosted the highlights too much so I went back into the compound clip, lowered all the highlights to 75%, then tried again. For some reason, FCP doesn’t seem to recognise changes to compound clips once they are on the timeline, so I had to delete and re-insert the clip to get the changes. I don’t know if there’s a better way of doing this but since there was only one clip it didn’t take too much time. I reapplied the effect and this time I was really happy with it.
I uploaded it to Vimeo and sent a link to the client. However, when I looked at the on-line version myself I was more than a bit disappointed - what had been a vibrant red leather colour on my Mac had been turned into a dull brown wash! Back to the drawing board…
I decided to go for the good old standard bleach bypass look. Rather than use the built in effect I thought I would have a go at one of the methods Denver Riddle recommends. This involves overlaying a copy of the clip, turning the saturation all the way down to zero, and changing the composite mode to screen. This produced a lovely result, increasing the dynamic range very nicely without crushing the blacks and highlights (which the built in effect does).
Finishing off
The second time I uploaded to Vimeo I selected 720p upload rather than 1080p. There is no discernible difference in quality (in fact I think the 720p version is slightly better) presumably because it is not as heavily compressed.I uploaded to Vimeo and this time was rewarded with a rendering that looks just about identical to the one I get on the Mac.
That’s all there was to it! It was a lot of fun, I learned a lot about FCP and I hope you learned something too.