Wednesday, June 2, 2010

Most home viewing environments tends to fall into one of two camps. Man-cave or blazing incandescent wonderland. There is however, a third choice...

In a pitch black room with nothing but the tv as a source of light, your pupils dilate and constrict with the constant changing between dark and light scenes. This eventually gets really uncomfortable for most people and can lead to headache and fatigue.
On the other hand, people who view their TV with the room lights on are introducing reflections and impacting the contrast and colour perception of what they see on screen in negative ways.

The alternative? Light the area immediately behind the TV. This is commonly known as 'Bias lighting'.

You may have seen before with Philips Ambilight. However the colour-changing, mood lighting and overly aggressive implementation of that solution isn't what I'd recommend to anyone.
I’ve worked in a few professional video editing suites and they all use a degree of bias lighting. Its important to point out, the people that make the content we all view, make that content under strictly controlled lighting conditions. If it is good enough for them....

But giving your eyeballs a less stressful workout isn’t the only benefit that well implemented bias lighting can achieve. You will have....

• Subtle ambient light in the room which helps you to find the remote/mobile phone, long-lost coins and avoid sitting on the cat.

• No lightsource shining/reflecting on the display.
Its bad enough our modern flatscreen TVs have super glossy black or silver mirror frames and many have highly reflective glass panels, we don’t need to see ourselves reflecting back at us as we sit on the couch (although it is a good reminder of how stupid you look playing on your Wii).

• No need to crank the contrast up.
With bias lighting, the perception of contrast actually improves, so your blacks will look, well...blacker. It gives the picture more ‘pop’ without looking unnatural. You can tone-down the contrast on the display which means you don’t have to drive your display as hard.

• Better colour definition compared to household lighting (*conditional).
If all you wanted to achieve is to alleviate eye-strain and get some indirect light in the room, then you could stop reading now and go out and buy any number of suitable products from incandescent lamps, fluorescent strip lights, those new power saving fluoros, LED rope lights etc. You probably already have something laying around the house.

There is however a reason why this isn’t ideal and I’m going to go into it in more depth, explain why it matters and what steps I took to settle on the setup you see demonstrated here.

Why colour matters

*The common incandescent lighting that you have in your home is generally marketed as being 'Warm' and is tuned towards the ‘yellow/orange’ side of the colour spectrum. Generally around 3000-4000 Kelvin. However, everything you watch from games to movies and broadcast TV was colour-graded in an environment tuned for 6500 Kelvin which is a much cooler colour temperature.

When you watch the TV with ambient conditions that deviate away from 6500k, your whole perception of colour gets altered. If you also have picture settings that are too warm or cold (most people), then the effect can be cumulative when the ambient lighting gets added to the mix.

Most of us have walls that are coloured to some degree, or curtains and other furnishing that also shift our perspective of the colours we see onscreen. Sometimes subtly, sometimes dramatically. For the most part, that can’t be helped, we don’t all live in a grey submarine where everything is perfectly neutral and drab. Unless you do your viewing in a particularly cold coloured environment (blues and greens), typical warm coloured household lighting will push everything you see on screen towards yellow/orange.

A light with a more appropriate colour temperature will at least mean you are within a closer tolerance of the ideal target.

However, having a 6500k light isn't a guarantee that you will have perfect results. The CRI (Colour Rendering Index) is a measure of how accurately a light can reproduce the source it is emulating when it illuminates an object. The higher the CRI, the better that reproduction will generally be.
This is not a recipe for success either. A 6500K light with a high CRI can also have a bias towards Red, Green or Blue which also colours the end result. 
Ideally, we want the RGB balance of the light to converge perfectly, but this is seldom the case. There will generally always be one colour that will dominate slightly. Ideally, that colour should be blue. We have a much harder time discerning a bias towards blue than we do with Red and Green.

Picking the right light
I wanted to get the right balance of spread (diffusion) and light intensity behind the TV, to make sure that it was easy to setup and adjust, that it was low-power and low heat (don’t want anything hot or magnetic messing with my TV thanks!) and that it was affordable.
Unfortunately, ‘affordable’ and ‘reference standard’ do not go together, so I had to weigh up a few different options and test them. This also made me realise just how much these lights vary and how they can influence the end result.

My ideal choice was fluorescent. It gives a very even and diffuse spread of light without putting big hotspots on the wall behind. There is also a huge range of different tubes to try.
There are many people that try bias lighting with LEDs or simple incandescent lamps. Many are happy with the results, but they tend to be nowhere near reference standard and they are often too bright, too dim, or leave hotspots all over the wall. The light shouldn’t distract you, its there to improve what you see on the screen, not to show off your decor (but if it does, its a bonus).

I have a 50” TV that is 1.5ft from a chalky-beige wall, so I would be able to get away with a single 2ft fluorescent T8 (which is a size now in decline due to the thinner and brighter T5’s)

I picked up a cheap and lightweight fixture from a major hardware store.

With fluoros, its important to get a fixture with an electronic ballast as they don’t have any starters or magnets and run at a higher frequency so they don’t flicker (at 50Hz) like old style magnetic ballasts. They are also more energy efficient, turn on instantly and are slimmer and lightweight. If you've ever worked in an office under oldstyle magnetic ballast fluoros flickering away at 50Hz while you stare at your 60Hz LCD, you'll know how headache inducing it can be. Electronic ballasts eliminate this problem as they tend to cycle at around 20,000Hz.

I have in the past kept tropical fish and reptiles so I already had a small stash of 2ft T8 tubes . Some are 6500k and some deviate a bit (either colder or warmer in colour temperature).

I decided to test a different one each night to see how they felt and how fussy I’d ultimately have to be in order to be happy. This basically told me nothing.
I discovered its rather easy to adjust to a particular tube when the tubes you are using are all within a certain tolerance of each other, so I devised a more scientific (ridiculous) test.

The testing begins...

On the left is a Kodak gray card. This is used by photo/videographers as a reference to calibrate colour balance. On the right is a clean bright white piece of card in a non-glossy stock. Both are placed on a black sheet.
Each tube was photographed from the same spot with the same settings and warmed up for the same period of time.

Putting them all together showed how much variance there was between them.

#1 and #6 are both 6500K tubes, both around 82 CRI, both made by Philips and both sold at the same general hardware store for the same cheap price.

#2 is 20watt (all the rest are 18w) and is a 6700K colour temperature (bluer than 6500K). You can see how the black sheet in the middle is a bit brighter than the other samples.

#3 is the generic tube that comes with the fitting.

#4 is my control (every test must have one). It is a dirty old 3300K tube which is over 20 years old. About as far away from ideal as you can get.

#5 is a 6000K tube with a 90 CRI. I was told (by the person who sold it to me) that it was in widespread use at a major New Zealand TV studio. I found it a bit hard to believe they were using something that was ‘non-reference’ but just because there is such a thing as a standard doesn’t mean everyone follows it.

So this little test illustrated the vast differences out there and gave me an understanding of how much colouring a lightsource can introduce to the environment.
But rather than just trust my eyes, I took it a step further and used the handy little eyedropper (colour picker) in Photoshop.

This allowed me to test the same areas of each test card and see on the colour scale if the results were too blue, too green or too red.
Almost all of them (with the exception of #4) were in the ‘too green’ camp.

#5 tested really well on the white card (the right side of the test pic) with its very smooth and cool grey tones, but when I used the colour picker on the left, it was showing up as extremely green. I couldn’t believe the results I was getting and had to retest it using the multiple camera shots I had taken.
Finding them all the same, I decided to hook it up behind the TV for one evening and sure enough, the right hand side was great, but the left was utterly green.

Unfortunately, as I was about to return it to the store, it slid out of the cardboard sleeve and shattered on the driveway. I’ll never know if it was just a bad sample or which side was the defective one.

So I decided to use the #1 tube (best alternative) while I hunted for something more suitable....

Pro results without the pro price

The choices of affordable T8 tubes that are 6500K with a CRI in the 90’s are extremely limited (and shrinking by the day as T5 becomes more popular). Most of those that are available are for aquarium/reptile applications. Some are tuned to show-off the greens of the plants or to give a vibrancy to marine reef fish and some of the reptile tubes are downright dangerous as they are designed to emit UVA and UVB frequencies. I want accurate colour, not skin cancer.

There is one US seller who has tubes specifically for bias lighting for home use, but the shipping costs made this prohibitive.

After much hunting, I felt that the Philips AquaSky was likely closest to reference and most affordable option I was likely to find.

Its a 6500K aquarium tube and has a rated CRI of 92. Compared to your household general purpose flouro its more than four times the price. Its still far cheaper than Philips other CR90 pro tubes that are aimed squarely at the pro video/design industry.
What made me most convinced it was worth taking a punt on was this –

It shows that the blue is the dominant bias and thats precisely what I required. The manufacturers' own graphs of course need to be taken with a large pinch of salt.

On the test card however it performed superbly, hands down beating every other tube I had tried.

The swatch testing returned some positive results with colour not wandering into the green like all the other tubes.
In use, its impact was immediately felt. I had been running the #1 tube (a cheap tube also made by Philips funnily enough) for more than a week. I was so amazed at the difference I thought I must have been suffering from placebo effect. I deferred judgement to the following night, but it was confirmed to me that without any adjustment to the TV whatsoever, colour felt so much more natural.

How do I mount this thing?
*updated setup
Placement of the light really comes down to personal taste. The further away from the wall it is, the more diffuse (spread) it will be and less bright. I controlled the spread with black PVC tape running along the diffuser. It was important to stop the light from shining onto the back of the TV, onto the ceiling above and the TV table.
If you wall mount your TV, you’d probably want two or more (depending on the size of the TV) lights placed vertically towards the left and right outer edges. These run cool enough that you don't need to worry about them heating up the back of your TV. Many people simply velcro them to the back.

*The Upgrade

I ran the above setup for a year and was very happy with the result, (and created this blog in that time) but of course there is always room for improvement. I wasn't happy with my center speaker sitting so far back from the front of the TV, or the column of shadow that the shelving cast with the biaslight, or having to lean back behind all my gear just to turn the light on and off. 
I also invested in a better colorimeter and was able to get some data as to how well my bias light was actually performing (more on that later).

The Hood

The black PVC tape on the diffuser was an effective, but not very flexible solution. If I wanted a wider or narrower field of light spread, the tape would have to come off and be repositioned, tested, rinse/repeat. 
The solution was to use some plastic plumbing pipe with a portion cut out for use as a visor. I also made some end-caps which sit over the visor so that when the caps are twisted, the opening (apeture) of the hood closes and opens. This gives me very good control of the light spread with very little fuss. 

This design also allowed for the use of coloured gel filters which hadn't been practical on the old diffuser. The blue gel can be seen attached to the visor above and in use here -

If the Walls could talk

My initial (not overly scientific) testing of the tubes colour performance kept me satisfied that it was giving me the results I expected, but there was always a factor that I knew was skewing the end result. The wall.
Having the perfect light source only carries you so far. Bouncing that perfect light off an imperfect coloured surface means you are moving further away from the ideal target. Of course that is better than using both an imperfect light source as well as imperfect wall, but hey, I'm aiming for the best result I can get without breaking out the battleship-grey paint...

My walls are chalk-beige coloured, which is to say, they are a little towards the red/yellow side of white. It was fairly subtle, but I was always aware of it with the bias light engaged. 

I purchased a more accurate Colorimeter for doing calibration and realised that I then had a tool that could help me measure the fluorescent tube and also measure the light reflecting directly off the wall. With these two data points, I could see how much difference there was between them and if possible, use a colour correction filter to rebalance it.

First, the tube.

The result was very pleasing. The amount of error is small and 6418 is not far off our 6500K target at all. Most off the shelf lamps are a long way off from this.

Next, take readings from the wall by measuring the tube's reflection from it.

Not so great... 
This confirmed the red push that I could see with my own eyes. In an ideal result, we'd want to see all three colours hitting 100%.

So these two tests gave me a colour temp of 6418 for the tube and 5375 for the wall. It was then possible to calculate what type of filter would be required to get closer to 6500K. 

Sadly, the ideal filter I required did not exist commercially. It would be a Cyan hue which cuts down both red and green. You can blend blue and green filters to create cyan, but layering filters together dramatically cuts down light output, so this wasn't an option.
The nearest commercially available filter that wasn't too dark was CT 1/8th Blue. I picked up a 1m x 1m sheet (more than enough for the small opening in the hood) and got to testing.

Its not perfect, but its certainly an improvement. The dominant red push has been neutered and the difference to my eye is very apparent. Without the filter I was always aware that the light was just enhancing the paint's natural hue, but with the filter, I'm not able to perceive it to be anything other than perfect white. 

If you've gotten this far, you may want to do some more indepth reading on this topic.