Portrait Evening at Rivington Park Creative Centre - Available Light & Studio Flash

27/11/08 Tried out my new lens a Canon 135mm f/2 L prime lens really for two reasons. Firstly for its sharpness, IQ (image quality) and secondly so I could hopefully bag some extra photos under the available light, while others were shooting studio flash.

So sat back and set the camera/lens up, using a Canon 400D set at 400 i.s.o. With the lens set at f/2.0, this gave various speeds from 1/200 to 1/500 depending on where the lens was looking. It was on centre spot focus and evaluative metering at first, but I changed that to centre weighted, it has no spot metering. I later changed to manual and set the shutter at 1/200 or thereabouts based on the histogram. Below is a photo of one of the two rooms used though one of the studio flashes off to the left isn’t visible

Note there are two sets of double fluorescent lights (green cast)which are switched on and two studio flashes, both have modelling lights giving a tungsten orange cast.

So whilst the other photographers were taking flash photos I tried some using the available light . It soon became very apparent that I was getting a green cast on some of the photos and it was often to the right of the photo when in portrait mode, but not always. The following three photos illustrate the effect; note straight from camera and vastly reduced in size converted from raw to jpg. The exposure isn’t correct, but this was early on and probably in Av mode, evaluative metering.

The shutter speed was 1/400 for the first two and 1/320 for the last all at f/2.0, so what is going on? Well I have three possible theories and its probably a combination of the three that is causing the problem. (not saying I’m right but read on)

The x-sync speed of the 400D is 1/200, yes I know I’m not using flash on this occasion but because of that at speeds over 1/200 only part of the cameras image sensor is being exposed instead of all of it at once (the 2nd curtain in the shutter is moving across before the 1st curtain has opened fully). That accounts for the banding of colour in the vertical plane, remember I was in portrait mode so the shutter is moving left to right in the photos assuming that the shutter opens from the top downwards in normal (landscape) mode.

The fluorescent lights give off a green cast that our eyes compensate for but the image sensor does not. So why different colours of green/yellow and even blue, well the third theory will also help to explain this, so second theory we are in the UK where mains electric alternates at 50 Hz, so the supply to the lights is continually going on an off 50 times a second and going positive to negative. This change in the power affects the lights output. The green/blue cast isn’t constant it varies with the alternating current. In the photos the shutter speed at its slowest is 1/320, but remember that only part of the sensor is exposed during this 1/320, so the sensor will most likely have caught part of the alternating cycle where the colour cast is strongest on the first opening of the shutter in photo three and on the latter part of the shutter at photo one. (depending on how you hold the camera in portrait mode, in this case the recommended hand over on top mode)

The camera was set to auto white balance, so immediately before taking the picture the metering has set the auto white balance to what it thought was appropriate at that time, which I believe is why they often get it wrong in indoor lighting situations . So it may have read a green light from the fluorescent's or a green-blue and then the tungsten modelling lights from the studio flash have also added their input! Not forgetting that there are four ceiling flourescents which may be of different age, giving off different light, two in front and two slightly behind the model. The light is also being reflected off the walls and pictures adjoining them, arghhhh!

So why is photo two nearly right colour-wise, well I think the white balance/evaluative metering has caught the fluorescents when they were giving their least cast, the tungsten is dominating, but the fluorescents have maintained a constant illumination through the shutters opening of 1/400.

Photo courtesy of Wikipedia - Creative Commons Licence original author Pieter Kuiper

So at 1/400 that is .0025 sec or 1/8 of the 50 hertz cycle above (ignore the 60 Hz that's for 110 volt countries, unless you are reading this in America). Therefore if the shutter is open at one of the top or bottom peaks where the voltage is above 230 then the light will be fairly consistent if my theory is correct, but if its around the centre line in the lights cycle then that's when the colour will be inconsistent - think no voltage, no power the light is actually off if only for a minute split of a second. The lights actually flicker 100 times a second due to the alternating frequency to complete one full cycle they are off twice. The human eye normally does not detect flicker above 50 Hz.

Ballasts in fluorescent lights help to stop the flicker and tungsten lights do not normally show any as the element is still hot when the voltage is zero so they still give off light.

A shutter speed of 1/50 or less will give a full cycle of light so it will average out and all the sensor will be exposed as well. Probably 1/100 may be okay to, as you are only seeing half a cycle of the lamp, but that half cycle could be from off to off, or on to on, or anywhere between. As the shutter speed is less than 1/200, all the sensor is exposed so any cast given off by the lights is seen by the whole sensor and thus the whole image will get that cast, which is easier in post processing to adjust. So in theory up to the synch(x) shutter speed of the camera (in my case here 1/200) should be okay. So you could use a grey card and set a white balance from that, but if you were at say 1/125 then depending on where you caught the alternating cycle the light given off could vary in its cast. Thus your custom white balance might still produce photos with a different cast, however below the x-sync speed, shooting in raw it will be easy to correct in software afterwards. Over x-sync is I believe where the worst problems start and you get the banding effect. In this case I needed 1/200 or more to avoid camera shake, I’m not a lover of tripods in these situations where there are a number of photographers and you need to be able to move position and camera quickly.

Later I took 10 shots in 4 seconds, well that's what the exif data said, but I was in continuous shooting mode, so it was probably 3 and a bit seconds. 2 of the 10 were usable (20%)

The shooting conditions were identical here, same settings shutter speed 1/320, no adjustment to the photo other than resizing. Just over a minute after the shots above. So what does this mean, basically if you’ve got fluorescent lights that aren’t electronically ballasted and they are not newish then you are going to get varying colour casts, that no amount of photoshop will fix, however you will have an approximate 20% chance if you’ve got your settings right and the lights are bright enough to get some usable photos. You are going to need a lens of f/2.8 preferably less and though the nifty 50 (50mm f/1.8) is handy you will be tripping over the other photographer (the one using studio flash)to get a photo. There is an f/2.0 200mm prime by Canon but you’ll need at least £3,700 for it and then you might be that far away that you are in the next room.

So what is the 3rd theory:

Well to recap the 1st was the x-sync speed, above it and you’ll get banding from fluorescents, though it may not be the same for all fluorescents, a few test photos in quick succession, should give you an idea.

2nd theory is that fluorescents give of different colours of light due to the alternating current continually switching them on/off. If its off it cant be giving out light so even forgetting the cast exposure will alter.

3rd theory is guesswork, literally, in that the metering mode, focus point and shutter speed will all alter the effect to a degree, if your shutter speed is in sync with the mains, 1/50, 1/100, then you’ll probably get an even cast from the fluorescent lights. If you are on 1/150 then your exposure might be out as the fluorescent may have had 2 off periods or 2 on ones, depending on where the shutter opened in relation to the mains frequency. Then again the super-dooper chip in your camera might be sorting all this out in even quicker time - though I think not. If you have a shutter speed outside of the mains frequency then there will be a variation in the cast of light which from the three photos at the start can be quite some difference, but at slower than x-sync all the sensor is seeing that same light so at least it will be an even cast on the whole image and not banded. Should someone switch on an electric cooker of high wattage appliance at the same time or you take a shot just as the adverts start in ‘Corrie’ (thats an English Soap tv program for those from elsewhere) then the countries alternating current may have dropped and it may not be at 50Hz.

Anyway a 20% hit rate isn’t wasn’t to bad on the night.

Right now lets deal with the 135mm and the studio flash, I’ll put that on the next page. Oh by the way my theory on Fluorescent Lights hasn’t got anything to do with Rivington Park Creative Centre, who offer excellent friendly studio facilities with tuition if needed and are highly recommended. Next time I may try the nifty fifty lens as the shutter speed could be lower to avoid camera shake and see if the same occurs, and if it doesn’t well that's my theory blown!!!!

As its quite a long article there’s some embedded piano to listen to, tracks 2 and 9 recommended.

See bottom of page re embedded music

Woods of Chaos by Rob Costlow