Three short timelapses that I’m going to include in a longer video from Nairobi, Aberdare, Lake Nakuru, Masai Mara, Lake Viktoria, Serengeti and Ngorongoro.
For the HDR time lapses I used the following flow
- +-2 stop bracketing on a 5Dmk2 fixed on a tripod in full manual (WB, aperature, ISO and shutter). If you use full manual you normally shouldn’t need to worry about sw deflickering, etc. Shot in RAW to enable setting WB later.
- In Lightroom, loaded the images to correct color, sharpness, etc and output jpegs
- In Photomatix, loaded the bracketed processed photos, used the tone compressor to batch create the HDRs and then tone mapped them (tried the tone enhancer first, but it imho creates that hideous “HDR look”)
- In Premiere, added all the photos to a sequence as numbered frames and used the app to zoom and pan
The two only real obstacles so far with this flow, software and hardware:
- Only two stop bracketing, which if you start with an over/under exposed image gives you a 4 stop margin for changing lighting conditions. This is an arbitrary and silly restriction imposed by Canon. I want minimum 4 stops up and down, preferably 6, AND I want it in 1/2-3 stop increments.
- Neither camera nor intervalometer can be set to take all the bracketed frames in immediate succession, halt for a longer period of time, repeat. This is always preferable, but the practical advantage is obviously greater if you’re shooting a scene that changes fast. In a dynamic scene, HDRs from bracketed images shot at large intervals suffer from severe ghosting.
I did a bunch of research to figure out how to build a video capable one man DSLR rig and thought I’d share my findings here to save others some time.
Why not get a video camera? Ken Rockwell said I should.
He’s got a bunch of really good points, BUT I still want to use my DSLR to shoot video for 3 reasons
- It already takes good photos, which means I don’t need to buy and carry two camera bodies, two sets of lenses, etc. Device convergence FTW. To me, this is the #1 advantage.
- In the grand scheme of optics, 35mm DSLR lenses are very affordable. A kick ass 50mm f/1.4 lens costs less than $500. It’s small, sharp, light, insanely fast (important in low light situations), and offers shallow depth of field (important when you’re taking a photo of something and want to blur out something distracting in the background)
- 35mm fullframe (5D) and crop sensors (7D/x0D/x00D) have amazing low light capabilities so you can shoot useable video without a lighting rig in most situation
Canon 5dmk2 was the first video capable DSLR to hit the market and at the time of writing it pretty much represents state of the art in the sub $3k price range. Still, it’s plagued with a few pretty severe weaknesses that you’ll need to address to turn it into a viable camcorder.
- Stabilization: if you pan or move the camera, video will be jumpy because there’s neither digital nor mechanical stabilization of the sensor in the camera, and Canon’s lens-based Image Stabilization gets jerky when you pan
- Audio: If you record audio with the built in mic it’s going to sound like crap. And if you plug an external Mic straight into the mini stereo jack on the camera it’s still going to sound like crap. No matter how good the Mic is. The reason is that the preamps in the 5Dmk2 suck.
- Focus: If you move towards or away from your subject it’ll be out of focus unless you adjust manually because the 5Dmk2 video digital autofocus is completely worthless.
I learned all of the above the hard way, but fortunately I did so well in advance of needing the rig so I had enough to time to figure out passable workarounds. I’ll share how I addressed weakness #1 in this post, and move on to the other two later.
Stabilization of moving shots
When you’re moving around, the last thing you want is for your video to look like this:
To prevent this from happening, you can either a) shoot a shaky handheld video and try to fix this problem in post production using a plugin like After Effect’s Warp Stabilizer or Deshaker for Virtualdub or b) shoot a smooth video to begin with using a Steadicam, which fixes the problem upstream by eliminating shaking altogether.
A Steadicam is a mechanical balancing tool. From a newtonian perspective it consists of 3 parts:
- At the top of the rig you have a weight: your camera
- In the middle you have a gimbal which is located slightly above the system’s overall center of mass
- At the bottom you have a counterweight preventing the top weight (your camera) from tipping over
There’s a little bit more to it than this, but you get the mechanics. Like most really good solutions, it’s very clever yet very simple and elegant.
When balanced correctly the system is a tiny bit bottom heavy. This means that the camera will not tilt around any horizontal axis regardless of how much your hand is moving around. All you can do is move the camera up/down and forwards/backwards, and rotate it around the vertical Z axis. THERE’S NO TILT.
The Steadicam was invited by an American cinematographer named Garrett Brown back in 1976 and they’ve been used so much in pro filmmaking that he was awarded an Oscar for the invention.
What are the pros and cons with the different stabilizers on the market and which one should I buy?
Fortunately you don’t have to pay $60k for a Hollywood Steadicam rig. If you have time, skills and access to a machine shop you can build one for for less than $100, and if you want to purchase an off the shelf system the cost is ~$200-$800. The compelling DSLR stabilizers I found were Steadicam Merlin, Glidecam HD2000 and CMR Blackbird. Here’s how they compare:
Merlin beats HD2000 & Blackbird on:
- Weight. It’s the lightest, which is important unless you’re built like the incredible hulk. A 5Dmk2 with a 16-35mm lens already weighs 5 pounds, and that’s without the stabilizer.
- Size. It’s small and easy to transport.
- Brand. It probably has the best 2nd hand value, partly due to brand recognition and partly due to the points above.
Merlin & Blackbird beat Glidecam on:
- Operations. The grip handle rests right under the gimbal so you only need to apply upward force to keep the rig steady. There’s no torque on your wrist whatsoever. This is a REALLY big deal.
Blackbird & Glidecam beat Merlin on:
- Price. They are both 40% cheaper than the Merlin
- Versatility. You can flip them upside down and fly the camera very close to the ground (this might be important if you’re going to shoot a real life version of Disney’s Bug’s Life)
- Capacity. You can use them to balance slightly heavier cameras. Merlin supports maximum 5 pounds with the counterweights included in the box and 7 pounds with additional weights. The other two systems go a tad bit higher, although in practice you’re going to need some sort of exoskeleton (vest + arm) to be able to fly a camera weighing more than 5 pounds for more than some minutes.
The Merlin was the right compromise for me. I bought it and loved it. If you understand the simple mechanics of the tool, it’s contrary to what I read before on some forums super easy to set up and balance. If you’re willing to give it 10 minutes you’ll figure it out. While I imagine it’d take a life time to truly master this tool, it’s easy to get quite good results, see for example the below:
My only complaint would be that the finish of the product isn’t quite what I would have expected given the price. I don’t mind the fact that many of the parts are plastic since this keeps the weight down, but there’s a difference in finish between this product and similarly priced mechanical equipment from say Arca-Swiss and Really Right Stuff. However, since this doesn’t affect operation of the Merlin at all, I don’t think it’s a big deal. I wholeheartedly endorse the product. It’s amazing.
Stabilization of fixed shots
I’ve always used my DSLR primarily to take still photos and will continue to do so. Hence, I needed to interface the Merlin with the stabilization system that I use for stills. My photo setup consists of
- A 3 leg carbon fiber tripod. This is imho the only way to go. It’s more stable than a mono-pod and lighter and easier to setup than quad-pod. You want a tripod with a center post that has a hook at the bottom so you can hang a weight there stopping it from swaying in the wind - especially if you’re shooting with long lenses or time lapses.
- A ballhead with a quick release clamp. This allows you to pan and tilt the camera, and mount and dismount it quickly and without tools.
- An L bracket for the camera allowing me to clamp it in both horizontal (landscape) and vertical (portrait) position while keeping the center of mass right on top of the ball head. This makes it much easier to make fine adjustments than if you flip the entire ball head 90 degrees over on its side, putting torque on the tripod and making it want to flip over to the side.
In my opinion, Gitzo makes the best carbon fiber tripods and Really Right Stuff make the best L brackets, the best quick release system and the best ball heads. They’re both pretty much indestructable and have travelled with me all over the world.
Integrating the best Steadicam stabilizer and the best tripod setup
I wanted to interface the Gitzo/RRS tripod with the Merlin to be able to easily switch between a) tripod, b) Steadicam, and c) hand held shots. Unless you’re willing to machine custom parts, the only way to do this is to stack the Really Right Stuff quick release system on top of the Steadicam Merlin’s dovetail plate.
To make sure this was possible before doling out my hard earned Amex Points (w00t w00t) on the Merlin, I started looking for schematics of the dovetail plate. When I couldn’t find any I reached out to Garrett Brown who invented the Steadicam and asked for one. Much to my delight he put me in touch with the Robert Orf in Steadicam’s engineering department, and Robert got me the drawing below. Kudos to both! It’s a top view of the dovetail plate with the front-facing edge at the top, and Robert gave me the permission to republish it to save you some time.
As you probably guessed, the small holes are just pilot holes to keep the camera from rotating once mounted. The large holes take 1/4-20 threaded screws holding the camera. The center of mass for a Canon 5Dmk2 with a Canon 16-35mm lens and a RRS L bracket attached is situated pretty much spot on over the H hole, so the best way to interface the two systems with this camera + lens combo is to
- Purchase a RRS B2 LR II clamp
- Run one locating screw through locating hole above “K” and one through locating hole above “O” and then screw in the included locating pins. The diameter of these pins is about 1mm or so too narrow to fit snugly into the cavities of the QR plate, so I strapped a cable tie around each.
- Run the Merlin 1/4-20 flat head screw through the hole labelled “M” in the dove tail plate and into the center hole of the RRS clamp. Since the RRS hole is threaded 3/8-16 you need to purchase a 1/4-20 thread reducer bushing and I’d recommend fastening it using LocTite 242.
This is what it looks like when you’re done and the whole setup weighs ballpark 7 pounds. It’s stable, light, sturdy and sparkling with awesomeness.
Next up, a blog post about how to solve the audio problem. But first I’ve got some stabilized video to cut. :P
A few friends of mine recently started a new company. They are brilliant PhDs and leaders in their field, and even though they have no previous business experience I don’t question whether they’ll be successful or not. My question is rather: how successful will they be, and how fast?
Sun Tzu said “Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat.”
Some ventures have iterated rapidly towards success with little or no strategy. I think most people agree they are the exception, not the rule though. Most successful companies do not operate without strategy, regardless of what they’re telling you. By doing a minimal amount of upfront thinking and testing before you start executing, you eliminate a lot of risk. My friends recognized this and asked if I could help them outline out their strategy and write a business plan. Given the company’s nature (b2b SaaS) and stage (early), I advised them to use a pitch deck as the medium instead. A thick business plan document often drives people to overthink things that are bound to change anyway, and if 30 pages are needed to explain the strategy it’s most likely way too complicated to be executed effectively.
A pitch deck forces you to articulate your strategy in a concise way. Basically you answer the high-level questions that an investor would ask you during a 10 minute meeting (or if you do well enough, maybe 60). There’s lots of value in building a pitch deck even if you don’t plan on raising funding though:
- Building a compelling deck forces the founders to answer a lot of difficult questions and reconcile their differences
- Your pitch describes your strategy. Your strategy drives your execution. It helps you evaluate whether you’re on track, provides a context for making sound day to day decisions, and enables you to estimate what resources you’ll need.
- The pitch helps the founders convince themselves the plan is solid (which reduces founder stress to just slightly less than the “unbearable” level), and it helps the founders “sell” themselves to potential customers, employees, consultants, distributors, board members, buyers or investors, wives, and others.
- Since you’ve built a pitch you have the mental model of your company required to tell a concise and compelling story when you meet people WITHOUT your pitch deck. This is when having built pitch deck really pays off.
The guiding template (see below) I sent my friends contains the following slides in the following order:
- Solution (+Demo)
- Sales & Marketing
- Challenges & Risks
- Milestones and Projections
Each slide contains two sections: “Slide goals” and “Comments”. Prior to these slides I also provided high level answers to the following questions:
- Why have a company pitch deck? (also answered in this blog post above)
- How do I build a solid pitch?
- How do I validate my assumptions?
My friends found the guiding template to be a useful tool when creating their own deck, so I’m posting an anonymized stripped down version here. Next up is the fun part, filing the slides in the template with content! Bob Goodson once suggested to use a Ted talk as a mental model when developing your company’s story - I’m not sure if it was his original idea or not, but I think it’s a very good one.
European countries proclaim “The king is dead. Long live the King” following the accession of a new monarch. I think a startup venture is quite similar to a King. It definitely takes on a life of its own, and even though you’re in control it is also true that it becomes a deity you serve. And like a King, who might be good or bad for you, a startup eventually comes to an end, and if you have it in your blood another one follows.
In January we decided to shut down Hug. This is simply a short obituary and a cheer for Hug, our supporters and entrepreneurship in general.
- Automatically turn things like thermostats, lights or power sockets off or down when they’re not needed
- Make the right tradeoffs between purchase price and operating costs when buying new appliances
- Identify behavioral changes that are low effort and yield big returns such as making minor changes to how you drive your car
- Energy management
- Home energy efficiency upgrades
- Social smartphone-based hypermiling
I’ve melted rocks for several days now without much sense of progress. What I’d like is some sort of counter showing me how MANY rocks I’ve melted to date. I’d also like a counter showing me my current rate of rock-melting, i.e. something along the lines of Rocks Per Month.