On a more topical note, once I’ve finished my current job for TreasureBags, I fully intend to update on the challenges of shooting a large gig at the Sydney Entertainment Centre.   Time’s been extremely scarce, but it’s all been extremely rewarding.

Truly impressive stuff …

A few months ago I got the opportunity to do a large-scale shoot at a local playcentre.  Photographing just under 300 kids over five days presents some interesting challenges, not all of them obvious; on the bright side, getting to play with 300 kids is pretty darn fun!  It’s been a few weeks since I finished the shoot and I thought some of my readers might find it interesting what goes on during such a complex process.

It was pretty obvious before I’d even shot the first photo that good planning was going to be key; taking candids in a large playcentre isn’t easy.  Candid photography and portraiture can quickly go south if you don’t know what you’re doing – prior to shooting, I’d set the following goals:

• An average of 20 high quality, compositionally varied shots per child.
• No use of staged shots – only natural compositions.  Staged shots have their place, but they never show the person, only the staging.
• Minimal ‘play-time interference’ – I was there as an observer, not a director, and the kids should still be having fun.  Like a good wedding shoot, my aim was to be as invisible as possible.  ’As possible’ being the ter, as a madman running around with a massive f/2.8 lens is … shall we say, rather obtrusive.
• Processing turnaround and proofs delivery within seven working days.

I knew that it wasn’t going to be easy; ignoring the stuff I couldn’t predict, I already knew that:

• Light quality was going to be bad.  Very bad – industrial lighting, variable levels of natural light throughout the day, and a two-storey high ceiling don’t make it easy to set a consistent white balance, let alone have enough light to freeze the motion of playing kids in flight.  As the day progressed, I knew from experience that I’d have to deal with rapidly changing light conditions from class to class.
• Keeping the kids happy and relaxed was essential.  Kids freeze when they’re unsure or scared, and a stranger looming over them with a camera’s the best way to reduce a two-year-old to tears, not the greatest ‘natural’ shot to say the least.
• Time was short.  Each class was only an hour long, and with the number of kids in each class, getting a 20 varied good shots of each kid was going to be a challenge.
• The sheer volume of photos was going to create workflow and processing issues.  With roughly six classes a day on average and approximately eight kids in each class, I was projecting nearly 1,000 finalised photos a day.  Given the lighting conditions and the rapid subject motion, I estimated that I’d need at least double that to ensure an average of 20 sharp final shots, giving me a total of around 2,000 photos a day, or 250 an hour.  At around 12 meg a file, that’s around 3 gig an hour of data, or 24 gig a day.  Simply copying and crunching all that data was going to be a challenge, especially given I’d have minimal available time between days.

The Approach

To handle the above challenges, I ended up using two cameras; a D700 as my primary camera for its low-light capabilities and a backup D80, used for wide-angle photography during the short ‘playground’ play session without requiring a lens change.  I mentally broke each class up into multiple sessions where each session used a different focal length and composition to suit the action and class involvement.  

I couldn’t afford the time to change between more than a handful of lenses, so the main focal lengths I used were 50mm at f/1.8 to f/2.8 (to allow for a lower ISO due to the increased light availability as well as combat hand shake) and 100mm – 200mm at f/2.8.  I used the longer focal lengths to be ‘outside’ of the action and let the kids have fun, the main tradeoff being reduced light and increased hand shake, leading to a higher ISO / faster shutter speed with slightly increased noise levels (but still well withing acceptable limits thanks to the D700).

I also shot a close-up session within each class at approximately 28mm to provide compositional variation.  These were shot using the D80 backup camera; the tradeoff was higher noise, but the advantage was absolutely no setup time, something that was really key given the short periods of free play time the kids had.

I used multiple CF cards along with a laptop / USB drive backup system to stream photos off my inactive card onto a backed up system while I was shooting.  I intended on limiting my use of flash photography as much as possible, but ended up compromising on this somewhat from the second day on.  Bounce flash was next to useless given the high ceiling, so I mainly used flash to provide some weak fill where absolutely necessary.

Lessons Learned

• As bad as you think the light’s going to be, it’ll be worse.  White balance was impossible to set with any consistency (even with the use of a grey card) and needed to be adjusted every time the cloud layer changed.  During the worst periods, this was as bad as four different white balance points within a single hour, all of which being significantly different to each other.  Even worse, the first two classes of each day consistently struggled to have enough light to maintain acceptable shutter speeds; only skill, luck, experience, and thinking on your feet can get you through this one.
• Selecting and processing over 10,000 photos takes longer than you might expect.  A lot longer; while I met my 7 day turnaround for most people, the final two days stretched out to around 10 to 12.  While not the end of the world, the delay was frustrating (probably for me moreso than the parents).
• Processing and coordinating shipped orders becomes increasingly complex as the number of prints increases.  This was something I hadn’t planned for and could have definitely been done better; there’s a big difference between dealing with individual portraiture / wedding shoots and dealing with over 200 individual orders.  I got some key learnings out of this one for next time, as well as a lot of ideas around how to streamline my ordering workflow; overall, while it was painful, it was invaluable.
• Not everyone’s always easily identifiable.  Lots of parents prefer to have their kids photographed without nametags on (as their clothes are lovely).  While it does make very nice photos, it also make matching names to faces a lot more complex, leading to additional work and delays.
• Communication is key.  I already knew this one, but it was good to see that it helped.  Communicating early and communicating often with everyone led to little additional time required doing one-off communications. 

What worked well?

• Planning the shoots beforehand.  While in post-processing, I was regularly surprised at how much compositional variation I was able to get given the large number of kids, the limited class times, and the highly limited amount of time I had to spend with each child.  The planning worked!
• Optimising the shooting workflow.  I spent quite a bit of time before the project planning out how to optimise my shooting / processing workflow.  This worked extremely well and in doing so, I’ve learned a lot of tricks that’ll help me in the future.  Overall, the best measure is that out of close to 300 kids, there were only two who I didn’t end up getting enough photos of and, while I can’t be one hundred percent sure, I’m pretty sure that least one of them disappeared for the majority of the class.
• Pre-planning the communication and delivery process.  By putting a fair amount of thought into the entire process beforehand, it proved easily manageable even when complications set in.
• The D700.  It’s rare that you can specifically point to equipment making the difference on an entire project but, in this case, it’s true.  The low light capabilities of the D700 made what would have otherwise been utterly impossible totally achievable.  And, that’s saying something. 

 What would I do differently next time?

• Spend more time optimising my shipping workflow.  Packing, sorting, printing, and posting all takes time, and the scale of this project highlighted a number of inadequacies in my current system.  
• Optimise my ordering system.  While it worked and was usable, it didn’t achieve one hundred percent success.  Usability was an issue for a small number of people, and some small language issues within the automated mailing system led to some ocassional confusion.  I don’t think there’s ever and end-state for this; I think it’s something that needs to constantly evolve and improve.  Having said that, I have ended up with specific changes that I can make that’ll improve things.
• Drop the second camera as an active tool.  While it was useful, the benefits of using a second camera weren’t outweighed by the increase in noise and the difficulty of getting good exposures using an APS-C sensor-based camera.  The extra minute it’d take to swap lenses and reset white balance are worth the increase in image quality.  While nothing that came out of the D80 was necessarily bad, it was just simply too slow to catch the more energetic kids on the trampoline and in the swings. 
• Pay closer attention to my suppliers.  One messed up order can waste hours.  And, it did.  I’m now weighing up the benefits of doing printing in-house because of the time I ended up having to spend sorting out a set of orders that were incorrectly printed.

Overall, it’s been a fun, interesting, and educational project, one that I’d happily do again.  Pending sorting a few final things out and time permitting, I might post some example shots at some stage.

I wonder what the sheep thought about the whole situation?

Cloud computing’s often confused with grid computing; they have similarities, but they are different.  Simply put, it’s a movement away from the desktop to an Internet-based system of applications.  Rather than keeping all your data on your computer and running applications installed on your PC / Mac to manage that data, the application and your data live on the Internet.  Knowing how it works isn’t anywhere near as important as knowing that, ‘it just works’ – the whole point is that the application is reusable, virtualised, and always available.

Big words, but what do they mean?  Basically, it’s the realisation of what Sun always wanted Java to be; a replacement for the operating system and all the application you’d normally run on your PC.  In Microsoft’s world, the PC is the centre of the universe – sure, there are servers, but servers are basically bigger PCs that do more.  The distinction is one of Access and Microsoft SQL server; both are database applications, but one is built for the desktop, one is built for the server.  Either way, you still need a fully-featured operating system and PC to run them.

Cloud computing takes a different approach – if the data’s important, it shouldn’t matter how or where you want to get access to it.  Instead, you should simply be able to get access to it regardless of channel, interface, or application.  Google’s one of the major proponents of cloud computing; almost all of their applications are built to leverage ‘the cloud’.  What’s the advantage of taking this approach?  Let me paint you a picture …

I use multiple devices on a daily basis; at last count, I’m using an iPod Touch, a BlackBerry, a laptop, a desktop PC, and an assortment of other Internet-enabled devices (including a Wii).  Needless to say, these all run different operating systems, have vastly different interfaces and screen sizes, and very different forms of control and portability.  I also maintain multiple email addresses, at least five of which get regular use, not to mention two separate calendars (one personal, one work-related).  To keep track of what’s happening around the place I also subscribe to a large number of RSS feeds – these are a simple way of subscribing to sites in a way that’ll notify me when new content is published.

In the absence of cloud computing, there’s a few major problems with this approach:

• If I read an email on one device, I can’t guarantee that it’ll be marked as read on all the other devices
• If I add or change an appointment, there’s no guarantee that it’ll update across all my other devices

By migrating my data to the cloud and simply pointing all my devices to it, I can make sure everything’s in sync everywhere, irrespective of what device I’m using at any given moment.  But wait, Evan, I hear you say; how’s that different to a server?  Well, ‘not very’ is the honest answer.  The real difference is one of philosophy; cloud computing is founded on a general philosophy of a reusable service, one that doesn’t care how you connect to it.  It doesn’t matter whether I read my email using Outlook, a browser, an iPod Touch, or my BlackBerry – everything works, it works well, and it was trivial enough to set up that I could get everything working in less than 20 minutes.  Try doing the same with MS Exchange or Lotus Notes sometime if you’re not an experienced administrator.

In practice, spending a bit of time thinking this through and consolidating my online activites has freed up a lot of time and simplified how I work and play.  It’s also meant that I have instant access to most of the things I care about, irrespective of where I am around the world, something that’s important when I’m out of the country and realise that I’m missing a document I really need.

The Internet gets rid of a lot of this pressure, unfortunately; software that gets patched after release is the norm, as is stuff that just flat-out frequently doesn’t work.  So, it’s quite possible that this’ll be two firsts; one of the first lenses to get a firmware update, and one of the first examples of a slackening of pressure on the manufacturers and a move towards fixing things in software rather than getting it right in the first place.

So you’ve bought your home theatre system, you’ve positioned your speakers, and you’ve calibrated your levels.  That’s it, right?  Everything’s hunky-dory now?

Not quite – the unfortunate reality of acoustics is that it’s real.  Well duh, you say, but my point is  that it isn’t a abstract model; the environment we’re in is affected by and affects the exact thing we’re interested in, namely how it sounds.

The core problem is that audio is a wave.  Without making things too complex, think of your basic waveform – pretty much all of us have at some stage flicked a hose up and down with a friend.  When we both flick the hose up at the same time, we end up with a really big peak, and when we flick them in opposite time, we cancel each other’s waves out.

Simplistically speaking, acoustics is exactly the same; our speakers produce various waveforms (acoustical energy), and they shoot out and bounce around the room.  In the smallest room one can imagine, these waveforms bounce off the walls almost immediately after they’re produced by the speakers.  As they reflect back and forth, gradually losing energy and sounding quieter, we hear these echoes and they cloud what we wanted to hear in the first place.

In an infinitely large room with no other sound sources, these waves never reflect back to us – they keep shooting out forever, so all we ever hear is exactly what the speakers produce.  That sounds like paradise until one realises that the whole attraction of live music (and, to a lesser extent, studio music) is that they were recorded in situations where things did reflect off various things.  The speakers at the front of the stage, the glass between the recording room and the microphone, and so on.  So, the ideal situation is somewhere in-between that infinitely large room and a nutshell – something that allows some reflections, but not so many that they become overpowering.

The two major issues with any and every room are reflections and reverberation. Reflections are just what they sound like – echoes.  While high frequency noises are technically waveforms, they tend to behave similarly to particles; think of them as billiard balls. A sound bounces off one wall and the echo hits your ears after the original sound.  In practice, this can make it very difficult to clearly hear what you’re listening to – you lose clarity, and you try to turn up your system so you can hear it clearer.  Typically annoying the hell out of your neighbours, straining your amp, and making you feel fatigued in the process due to the higher volume, of course, but hey, it’s all good, right?

Reverberation (more technically known as standing waves) is a little different – low frequencies tend to behave exactly like the waves they are.  As the frequency gets deeper, the wave gets longer, and as the wave gets longer, the closer it gets to the length of your room.  Think of your friend with the hose – if you’re unlucky enough to be sitting at the point where the reflection of the wave of a specific frequency cancels itself out, guess what?  You can’t hear it.  No matter how much you turn up the volume, you’ll never be able to hear it.  Your neighbours, on the other hand, will hear it just fine.

What are the cutoffs between the two, you ask?  For the technical, standing waves tend to be an issue below around 50hz, while reflections tend to kick in above around 200hz.

So, with all these potential problems, how do you know if you’ve got an issue?  Reflections are easy – clap your hands.  If you can hear a fluttery echo, you’ve got some issues.  Reverberations are a little more difficult to detect; the simple test is to measure your room dimensions.  If the width or length of your room is less than around 4m either way, you’ve got a problem.  If both are less than 4m, I’ll guarantee you that you have pretty significant problems.  You might be deaf by now as well, but at least you can still read this.

How significant, I hear you ask?  In practice and in a bad room, certain frequencies can easily be up to twice as loud or twice as quiet as everything else.  How does this make things sound?  Like crap, to be blunt.  It’s enough to ensure you can’t hear anything at all (having your explosions sound like a wet pop’s never much fun) or you get a headache and your windows break because of sudden spikes in specific frequencies.  I’m not kidding, either – in my old room, sounds at a specific frequency were so much louder than everything else that even when I was listening at “normal” levels, my windows almost fell out of their frames because of the resonance.

Interested in a more technical approach to working out exactly where these problems exist (in terms of specific frequencies)?  Room EQ Wizard is a brilliant program for automatically plotting your room’s frequency response.  Writing a tutorial is a little beyond what I’ve got time for at the moment, but trust me – it’s worth playing with if you’ve got the time and inclination.

So, assuming you have a problem, how do you fix it?

The naive, but understandable approach, is to assume that one can use equalisation to fix any issues.  And, to a certain extent, that will work.  The catch is that it’ll only work for one spot in the entire room - if the characteristics of the room are affecting the acoustics produced by a set of speakers, boosting or decreasing specific frequencies will only add additional complexity to the problem.  It may flatten the frequency response in one specific area of the room, but it’ll throw everything out even further everywhere else in the room.

The better way is to change the acoustical properties of the room.  The good news is that it doesn’t have to be difficult to do so, but the bad news is that it may mean moving things around a fair bit and / or changing what’s in the room.  The thing to bear in mind is that you’ll never achieve a “perfect” room – it’s impossible.  So, the best you can hope for is to make things better.  With that in mind …

Reflections are probably the easiest to fix.  As they’re caused by waves bouncing around the room and hitting your ears, the logical thing is to either stop them bouncing or make them bounce in lots of different directions, primarily away from your ears.  The reason for the second option is because every bounce makes the wave lose energy, making it quieter.  The first thing to bear in mind is that hard things make the waves bounce while soft things dissipate the energy.  So, the following are bad:

• wooden floorboards or concrete floors
• glass
• large open spaces of wall
• leather
• wood in general

What’s the easiest way to fix it?  Try:

• adding bookshelves filled with books (the random shapes break the reflections)
• hanging an attractive rug on the wall
• laying some carpet or a rug in the room
• adding a fabric couch or footrest

If that’s not enough, the more advanced approach is to build some acoustical baffles.  I’ll write some more on that soon, but the basic idea is to build something light and large that you can hang on the wall / ceiling to help dissipate acoustical energy by preventing it from reflecting.  Egg cartons also work well, but look ugly – an interesting test is to simply hang eggshell foam behind the listening position.  Typically, it’ll noticeably change the sound simply by preventing the waves from your speakers from reflecting off the wall behind you directly onto your ears.

The other approach is to build a large wooden board with varying sized pegs in random configuration to help reflect the sound away from your ears.  They look good, but they require quite a lot of work.

Standing waves are a little harder to fix.  Without getting into the specifics, lower frequency waves tend to behave a little differently, making typical solutions to reflections fairly ineffective.  Plus, the room itself frequently acts a large speaker in its own right, reverberating and making certain frequencies every louder.  So, with that in mind, how does one go about fixing it?

The easiest way is to build dampening baffles for the corners of the room.  Corners tend to enhance low-frequency reverberation, so by packing an appropriate baffle into the corner, one can prevent the reverberation from occurring in the first place.  Photos and instructions will follow soon …

The other, more technical option, is to build a resonator that traps specific frequencies.  Think of blowing into a bottle – it makes a tone, tuned to the shape of the bottle.  A Helmholtz resonator does exactly the same thing, except it traps specific frequencies by converting them into physical energy.  The main problem with them is that they’re like a sniper rifle – they hit specific frequencies enormously effectively, but the trade-off is that they only hit specific frequencies.  And, tuning them, to be blunt, is an absolute pain; it’s not for the faint of heart.

So, there’s your easy options as far as fixing a room goes.

 Dog Sledding, courtest of The Big Picture

The reality is that while more megapixels gives you more to play with, that doesn’t do much good if those pixels are largely noise.  As I’ve discussed previously, the sensor is the major limiting factor in extracting quality out of a given capture.  Garbage in, garbage out; it still irritates me to no end to see cell phones with 6 or 12 megapixels.  Sure, it sounds great, and it makes a spectacular bullet point on the back of the box, but when the sensor and lens can only capture two megapixels of usable quality information, those extra megapixels just translate into additional storage and management costs.

Sure, storage’s cheap – I personally have over four terabytes of primary and secondary storage at home, and relatively to many, I’m not that extreme.  However, those larger images take longer to process, they mean I have to spend money on backing them up, and if the quality’s poor to being with, why am I bothering in the first place?

Enter Olympus, stage left.  In a remarkably insightful discussion, the head of Olympus’s planning department recently said that 12 megapixels is enough for anyone other than a professional.  Surprisingly, there’s been a remarkable dearth of cries of ‘luddite’ from the bowels of the Internet – it seems like most of the people ‘in the know’ get this and agree with it.  12 megapixels may not seem like enough, but it’s more than enough to get an extremely high quality standard print.  It’s even enough to blow up to around a 20cm x 30cm, if you know what you’re doing, use quality lenses, and are careful in your enlargements.  For the vast majority of non-studio / non-professional photographers, that’s good enough.  More than good enough – the current trend seems to be towards canvases, and they’re far more forgiving of enlargements than standard print technology.

Instead, Olympus thinks the future is in autofocus and other ‘non-megapixel-related’ technologies.  And personally, I agree with them – it’s one thing to be shooting for billboards and posters and yet another to be shooting for everything else.

Now, why can’t every other consumer photography vendor clue in?