There is now an usual post-processing routine I follow, it’s really just something I came up with and since I don’t “go by the feel” and like to believe there is a set of formula on touching up your photos before publish them. Continue reading
It has been sometime since my last attempt on black card photography, and it also has been a long while since my last fireworks photography. So to sum up, waiting at prime location attempt to shoot decent New Years Eve fireworks was a gamble of luck. Continue reading
Doing Signaling a Wave – A Timelapse Impression of Vivid Sydney 2011 was a great learning experience and I decide to document down the making and thoughts behind this music video. This will help me a lot when I look back as journal, and hopefully this will be some way helpful to others who also interested in timelapse video.
Note: I realise there are fair number of Chinese audience to the video. Please let me know in the comment if a Chinese translation of this article is desirable. I may decide to translate the whole thing (which isn’t my greatest skill…) if enough people asking for it.
It all started by an inspiration of Sydney Harbour timelapse video done by Philip Bloom where he managed to capture the beauty of Circular Quay from only his hotel room in 1 night. This was eye-opening to me as I always thought you will need access to spectacular views and the instinct to pick the right time frame to start the shootings.
I made up my mind to try out on timelapse video and bought a cheap intervalometer. Originally I thought I’ll have fun: shoot a sequence, make a small GIF animation or short video, show of to friends and repeat. However the colour and light in Vivid Sydney amazed and motivates me to start something bigger.
So the ‘fun’ self evolved to a half decent project, where I only have 1 year of experience shooting and no experience dealing with video clips. I even don’t know how to apply text overlay to a video (which will be further elaborate later in this post).
Test the Water
Before everything, I have to make sure I do have the capability to capture timelapse. So I do a test run on 1 scene of the Sydney Opera House (which is the scene from 0:21 to 0:28). I did my first timelapse sequence with the following camera equipments:
- Nikon D90
- Nikon AF-S DX Nikkor 18-70mm f/3.5-4.5G
- Wired Intervalometer
- Standard Weight Tripod
It worked great and I’m glad I only have to invest in the intervalometer like initially intended. After took the photographs, next step was to convert into a video format. After some search over the net, I quickly learnt that it can be done by Adobe Lightroom: A existing software without further investment. With ‘Slideshow’ function in Lightroom, I was satisfied and now able to proceed with the project (of course, only afterward I realise the technical difficulties of recompiling music, splicing and video effect).
In summary, the software used in this making were:
- Lightroom 3.3
- LRTimelapse 1.3 Dev
- Audacity 1.3 Preview 2
- Format Factory 2.60
Plan it out
There are many light sculptures in this year’s Vivid Sydney, many of them are interactive artwork and I want to cover more than just the signature Lighting the Sails (scene of Sydney Opera House). I planned to shoot a maximum of 3 minutes long video exclude intro and outro, and I decided 15fps frame rate is acceptable instead of going for standard 24fps. And not knowing how I will use each sequence of timelapse, I think a 15 seconds timelapse video for each scene should give plenty of padding for post editing.
I dropped down scenery that I wish to capture such as:
- People on treadmill (#19: Re-CYCLE!)
- Bright light beams shooting from the Overseas Passenger Terminial while standing in front of the light.
- Kids playing with MCA projectile (#20 Interactive Paint Projection)
Roughly 25 scenes were picked include backups, I planed out a routine of 2 nights and the order I should visit each scene with Sydney Opera House and The Rocks on separate nights.
15 seconds per scene means I will need to have 225 photos when at 15 frame per second. Most timelapse sequences were captured in 1 second exposure and 1 second interval: this is the fastest setting my intervalometer able to perform. So to shoot 225 photos I will have to standby for 450 seconds (= 7 minutes 30 seconds) of shooting per scene. I tend to stand for a full 10 minutes, you know.. just in case. Weather on both night are dry and not windy, otherwise the entire process may be much more dramatic. 2 video clips were also captured designated to be used as outro. I later realised video editting was way over my head and decide to dump them and stick with frames of photos.
In the end, 21 of 25 scenes were captured. Abandoned scenes were either because lack of interaction, little creative light or simply ran out of time.
This is the most relaxing part of the project where I can sit back and sample through several create common songs from sites like Jamendo until I find one that able to delivery the mood I want to project onto the video. To make sure I don’t browse through endlessly, I keep the link of ones I like until I hit 5 and I stopped. I waited till next day (as a second opinion from myself) and run through the 5 songs I picked and make decision on 1. There is the possibly where I might ended up disliking all 5 the next day and have to through the whole audition process again. But gladly, that did not happen.
I use Audacity to analyse the song, adding notes as I repeatly listen to it. The first thing I look for is mood: identify low key, excitement and high. Secondly I search for “stops” that are suitable for scene changing. Although the stops doesn’t nessecary lands on the exact seconds, but I can safely said it is in either 3, 4, 7, 8 or 15 second sequence. This is the deciding time spot to change timelapse scene within video. I also use this software to truncate the song from more than 5 minute to under 3 minutes and apply fade in and fade out to both end of audio clip.
Disclaimer: I know nothing about audio engineering, so some of the chosen strategy might seems hideous and not recommended if you know what you are doing. But that was what I did anyway.
most people will tend to think timelapse video is meanly about taking photographs then photoshopping them together. I think building a storyline as if i’m directoring a film is also a critical part of an exciting timelapse.
Knowing the stops in music means I can draw boxes and fill in names of timelapse scene when listening to the song again and again while picturing the scene I picked.
I throw timelapse scene into storyboard in try and error approach: complete a storyboard, look at photos of that scene for designated number of seconds and switch to the next photo sequence emulating follow up scene. I will quickly find mismatchings such as long boring scene, or similar scenes too close together causing repetitiveness. When the storyboard is built, I have to sleep over it to get a second opinion from myself the next day, in which I did made minor changes to the storyboard and is then happy with what I setup.
18 scenes out of 21 were used. 1 abondoned scene was a reshot of MCA interfactive light paint. The other 2 were before and shot scene of Fire Dance which was too dark and not as exciting. Originally there was plan of reusing same photo sequence and alter with zoom or order, but this idea was forgotten somewhere along the process…
Disclaimer: similar toaudio editting, I have no knowledge of plotting, so it may seems extremely inefficient on the way I do this..
Now I have a huge stack of photographs and the precise instruction on how to puzzle them up, it is now time to create timelapses.
I start by using photoshop batch processing to shrink photo resolution from 3216x2136px (average 1.8Mb) to 1608x1068px (average 220Kb). Originally was to speedup sampling, but evantually I used the shrinked size in final production as size is good for 1280x720px video.
In Lightroom, I created a catalog for the entire shootings and separate scenes with different collections. Knowing the length of video for each scene, I know the exact number of photos I will need to make the timelapse (for example, opening scene is 11 seconds which is 165 photos). I pick a section that I believe is the best and trim of the rest.
I touch up the first photo of each scene (straighten, brightness, color balance) and use Sync Settings to apply same settings to all photos in the collection: One of the most favorable feature in Lightroom). Then I use LRTimelapse to create transition effect: panning, zooming and fadings, which was extremely straight forward and easy to try and error to find the best configuration.
An unexpected scenario was I was unable to find an easy solution to add text overlay for intro and outro. After hours of frustration, I decided to fade out my outro to black and insert blank frames with text to the end of timelapse sequence as my solution of outro. Extremely fiddling, but it worked!
Finally, after all has aligned, it is time to make timelapse video. By using the “Slideshow” function in Lightroom, you can create slideshow sequence of 15fps and output in video format (mp4). Originally I want to create different timelapse scene as individual video file and merge all together with Format Factory, however I found that Format Factory will create an undesirable ~0.25 seconds blank between sequence, which really bothered me. In the end, I have the slideshow video as the entire thing.
Sampling a 320x240px takes less than 30minutes to generate and creating the full quality (1280x720px) video takes 1 and half an hour to complete. When video is generated (correctly. Yes, it does fails), I use Format Factory to join the video with the audio I cut up earlier. Since my video is 2 seconds longer than my audio, I will have a 2 second silence in the end, which is the exact output I wanted.
The project span over a week and was shot in 3 seperate nights. There was late nights but was enjoyable and seeing the result was really rewarding.
Things I should have done better are:
- I skipped few testings timelapse samples in later half, and some transition effect was unexpected. For example, Customs House was suppose to zoom in slowly and Crystallized was meant to rotate.
- I off-sync the beat by exactly 1 second, because 0:23 to 0:24 is consider as 2 seconds not 1. The end result is actually off beat, but I still consider them as acceptable (enough that I don’t want to wait for another 1.5 hours generating the video file)
- Scene selection was not great. Several moments I have doubt on my plot, however I locked in and decided not to make changes to the plot. The advantage of that is the project completed on time, otherwise I will be be perfecting the video right now…
- I caught a flu afterward. Should have wear more warm clothing when out shooting. Body temperature drops surprisingly fast when you are standing not doing much.
I had fun, I have you enjoyed the video as well as this article. Feedbacks are always welcome as they will contribute greatly toward my next timelapse video.