Thursday, January 13, 2011

Which lens blurs the background best

This subject has bothered me for a long time. Most people say that a long ('telephoto') lens is best for blurring the background in an image and providing selective focus. I think this is only correct under certain conditions. It all depends on the absolute image magnification.

I'm mostly concerned with portraiture here. When making a portrait, the choice of focal length is determined largely by the perspective effect of different lenses. Too short a focal length and people look like pinnochio with noses exagerated. Too long a focal length and they look like a cardboard cutout. However, for environmental (as opposed to studio) work, you also want to be able to blur out the background, and so the effect of the lens on blur is also important.

So here is my bold statement: The focal length has almost no effect on background blur as long as the subject is kept the same size in the frame at all focal lengths (i.e. the magnification is constant). The final caveat is that the magnification is small i.e. that you are not in the 'macro' region.

How come? Surely people dont spend $000s on long fast lenses for nothing? No they dont but they want the maximum magnification possible for wildlife shots etc, and shallow depth of field is a side effect. In portraits the subject is always the same size whether at 50mm or 500mm - that means on a longer lens you step away from the subject until the person is the right size. For shooting aminals you get as close as you dare and if you use a longer lens then great, your subject is bigger in the frame - probably they are always too small already!

Are you sure? have you tried it out?

I have tried two things:

Take a 50mm lens and a 200mm lens. Take something with writing on it like a news paper and stand it up on something. The place another object that is easy to focus on about 100mm (4 inches) in front of it. Now put on the 200mm lens wide open , stand far enough away that the newspaper fills the frame from top to bottom when you focus on it. You should be about 2m away or thereabouts. Now focus on the other object and observe how out of focus the newpaper is. You maybe cant read the text, but some of the smaller headlines are just legible.

Now put on the 50mm lens and stop it down to the same aperature as the 200mm lens (if its a zoom then this is easy!) and make sure you look through the viewfinder with the lens stopped down in DOF preview mode. Make the newspaper the same size again so that it fills the frame. Refocus on the object in front. You will be much closer this time. Observe the blur - it will be very similar. Probably ever so slightly less blurry than the 200mm

I have tried this in an optical raytrace program with 210mm and 105mm lenses (only 2:1 this time) Set the object distance and field angle so a 100mm object 2m and 210mm gives a 11mm image height. Now shift the focus by 200mm (bigger shift this time) and ask the program to draw the spot diagram. An out of focus spot is drawn with a radius of 0.315mm. Now switch to 105mm and 1m subject distance which gives the same image height of 11mm. Shift the object distance by 200mm and now the spot size has radius 0.423mm.

You will notice that the blur spot is only marginally bigger and definately not 2x. The reason it is bigger is because 200mm is 10% of 2m but 20% of 1m and therefore the conjugate point of the new object distance is disproportionately further from the original image plane. But it isnt much. So to a good approximation the lens focal length is irrelevant in determining the magnitude of the blur of out of focus areas.

So what can you do to control selective focus?

There are two things:

(1) Increase the F-ratio of the lens: this increases the angle of the marginal rays and so the blur is bigger in due proportion to the opening of the lens. Faster lenses have a shallower depth of field and a shallower depth of focus (i.e. thickness of the acceptable image in the camera).

(2) Make the sensor/film bigger. This decreases the other magnifiaction that everyone else forgets about. This is the magnification between the sensor/film and the print/monitor image. The same rules apply as the lens magnification. If the picture is very small then it will all look sharp. Take the same picture and make it fill the screen and then you can see the blur in the image and the perceived DOF looks narrower. Go on try it with a picture on your computer!

If you use an APS-C camera, your DOF is 1.6 times wider than those with a full frame camera at the same conditions of distance, fstop and subject size because the focal length is 1.6x shorter. This means that the magnification at the sensor is 1.6x less and also the blur distance ratio (as above) is also 1.6x less because of the reduced focal length. You then magnify the image by 1.6x more to make the print. So overall the depth of field is 1.6x deeper (you know that point and shoot cameras with microscopic sensors have everything is in focus! the lens is 2mm focal length, so 1m is infinty to it!). So you can see that DOF and OOF blur is proportionall to sensor size. You can think of this as affecting the 'effective' Fstop of the lens in the same way as you have an 'effective focal length' as people do when comparing lenses to 35mm. So compared to 35mm, some formats have the following effect:

Consider an f2.8 lens, used for portraits such that the subject always fills the frame:

Format Size Ratio Effective Aperature Aperature for f2.8 blur Aperature for f1.4 blur
4/3rds 0.54 5.19 1.51 0.76
APS-C 0.63 4.48 1.75 0.88
35mm(full frame) 1.00 2.80 2.80 1.40
6x4.5 1.55 1.81 4.34 2.17
6x7 1.94 1.44 5.43 2.72
4x5inch LF 4.00 0.70 11.20 5.60
10x8inch LF 8.00 0.35 22.40 11.20

The first column is the size ratio of the sensor or film. This assumes that the print has 10x8 aspect ratio and that you crop to get best use of the format.

The second column is how the blur compares to a 35mm format f2.8 lens. This is really revealing. How much would you pay for an F0.7 lens on a 35mm format camera?! And yet there are several 5x4 lenses out there at affordable prices with a maximum aperature under f2.9. Food for thought!

The other columns show what fstop is needed to mimic an f2.8 lens on each format. So on APS-C, my 50mm f1.8 plastic fantastic is like an 85mm f2.8 lens (just like my old tamron manual focus f2.8!) . Also even that hyper costly f1.0 50mm lens on your APS-C samera wont get you even near to F1.4 on full frame.

On 4/3rd's, you will obviously struggle to get good selective focus at portrait distances. On 5x4 most standard lenses wide open are about 4.5 - 5.6 giving f2.8 performance easily and cheaply.Becuase the demands on magnification are so little, even 1950's 4 element lenses are good enough wide open.

For medium format, the advantage of 6x4.5 film seems to me not so great. Alot of extra hassle for a small improvement in image quality and selective focus. 6x9 or 6x7 are a better choice, although pricer gear than 5x4.

So for portraits my favourite rigg is my 5x4 speed graphic with an Aero Ektar 178mm F2.5 or better still my Pentac 210mm f2.9 (which is smaller, lighter and longer despite being not so pin sharp, but thats fine for portraits - also the Bokeh is better than the aero wide open at the edge of the field)

Thats enough rambling for one night - sorry it got so techie. Exec summary - in film/sensor formats, bigger is always better, at least for portraiture.

The Speed Graphic:

the mystery solved thanks to those nice guys at . My thanks to Dann Fromm, 45PSS and the others who helped explain my camera's oddities.

The back on the camera appears to have an after sales modified spring kit to admit a roll film holder. Obviously this makes it a bit tricky to use other lenses, without a ground glass to focus on. However its perfectly serviceable with its current lens and rangefinder setup.

The lens is 1938 and although the manual is newer, the camera itself is likely to have been produced in a batch in 1940. So now I know.


I need to check the RF and try a roll or two now and see if the holder is OK !

Wednesday, December 22, 2010

Here are some pictures of my latest acquisition - a miniature speed graphic. It seems not to quite match the usual pattern..

its serial number is 260434 (stamped into the top of the shell inside the front do


Interestingly it has a graphic 23 roll film holder despite the lack of graflok back...

Wednesday, November 24, 2010

So it been a long term interest of mine that the primary supply of energy in the world today is a limited resource. Its part of human nature to spend ones resources until a habit is formed before suddenly realising the barrel is nearly empty. So for many decades now we have used almost free energy with not a lot of interest in the cost and the sustainability of supply.

Go look at :

There you can see the work of people who have analysed the oil supply who dont have a vested economic or social interest. If this prediction ( the mean of a number of learned peoples estimates) is correct then the following might well be the case in as little as 5 yrs.

Cars will be a serious luxury affordable by the very rich only

Food will become the most expensive commodity in our lives

The world economy will never get out of its current spin and back to full 'growth'

Why? Because the cost of oil is in nearly everything we do. The food we eat, our transport costs, the cost of nearly all our goods, and the costs of heating and water. 

Politicians don't like to talk about this issue because there is no way they can get votes for it. Its an uncomfortable truth to be ignored in case they get the blame for not having prepared.  Instead they discuss global warming - which is more likely to be a natural phenomenon and out of our control in reality (thats another debate).

The current western society is completely unsustainable in a world where oil is in limited supply. The in-elasticity of the supply of oil means that prices will be highly volatile. Soon the $150 peak we saw in 2008 will be a normal and frequent occurrence - its no mystery that every time the world economy tries to get going again, the oil price rises sharply. That is because oil supply can only just keep up with demand.

And despite an exponential increase in exploration, the number large fields being discovered is diminishing steadily. Sure there is half as much left as we've all burnt so far - but that half is mostly hard to get to, and will cost more to extract - Energy return on energy invested will fall from 8:1 to 2:1 before we are well through the existing 'reserves', making that oil do less useful work for us. And yet al the time global population and demand for energy is every increasing.

So what to do? Well I'd make sure you have some land to grow veg on, keep a horse and grow some wood for the winter. And try to cut your energy consumption by a factor of 10. Don't live in the desert, build houses that have thermal mass and good insulation, drive cars that are better than 0.5% efficient in practise, install a ground source heat pump. etc. etc.

As long as the pace of change isn't too fast, it could actually be a whole load of fun learning how to survive in the post oil era, making better use of energy and learning some good basic skills. 

Thursday, November 18, 2010

This is anEnsign Autorange 20 by Houghton-Butcher. A folding medium format roll film camera from the 1930's. It is equipped with a 100mm F1:4.5 Ensar Anastigmat in a 'Mulchro' shutter. Serial no. H23908, probably from the late 1930's

When I bought it, the shutter was sticking and the lens was cloudy. Also the rangefinder was filthy and the door wouldnt close properly.

The lens cleaned up easily - someone had squirted light oil into the shutter and this had got onto the lenses. The glass is now clear and showing only a few light cleaning marks.

The self cocking "mulchro" shutter was another story. Its a crude design and relies on the rather soft brass parts to keep their shape with some accuracy. There are several pawls which run up ramps to catch and they are quite worn. Also the B/T mechanism requires undue precision from some of the levers and wear and tear has made this unreliable. Worst of all the main pinion that engages the segment arm of the timing mechanism is well worn and didn't always stay engaged with the segment arm.

Basically the shutter is worn out. In the past some of the arms have been bent with pliers to make up for worn out pawls. In doing so the slow times got shortened and this had to be sorted out by very careful bending of some of the internal levers to return them to their intended travel. There is no gear train mechanism like a compur just a rather odd star wheel escapement and a lever with a gear segment.

I managed to sort out most of its trobles, but the B/T mechanism is beyond help. B is OK but T is unreliable and sometimes the shutter closes immediately. After using B or T you must turn the shutter to 1/100th and back to release the timing mechanism. Still it works well enough for me to put a film in this weekend.

One odd thing with this camera is that it has rise and cross slide movements (about 12mm each - see right..). This is puzzling as the lens - a triplet, like a cooke lens - ought to only barely cover the 6x9 film format. Also because this is a roll film camera, there is no way to see what the effect of the rise/slide is. Both the 'brilliant' finder and the main finder on the body are unaffected by the movements.

The focussing is by the chrome lever visible at the front left of the bed in the photos. This swings out to slide the lens bed along - no front cell focussing! There is a lever arm that couples the bed to the rangefinder. Inside the rangefinder there are two adjustments to control the near/far points. Unfortunately the rangefinder is a seperate window to the viewfinder so you have to focus; then move over to frame. The viewfinder has a lever for a 6x4.5 mask - but of course the film plate mask is missing so I am stuck with 6x9.

Now that its ready for some film, maybe I will find out if the somewhat excessive efforts on the shutter were worth the trouble....

Thursday, July 23, 2009

So today I installed Sage on my PC - looks interesting - I have used Mathematica in the past for doing symbolic maths and I use Matlab at work alot for data processing. Work dont have Mathematica and its nice to think there is a free tool that will do at least some of what Mathematica can do (preferably without the sort of error messages designed to make you realise you know only 0.01% of the sum of human maths knowledge...)

First look - it uses vmware to run in embedded linux and then you start a server to use your windows browser as a front end. This is a bit of a clunk but seems to work. I guess I just need to have a play with it (and learn yet another syntax).

Also today- been looking at the Quad 22 preamplifier schematic. This is vintage valve hi-fi - I have half an intention to build a hi-fi preamplifier with a switch that say 'Hot/Cold' which incorporates a valve gain stage to give the warm valve sound.

Many years ago, I did A-B comparisons wiht a Quad valve preamp/amp and a transistor preamp/amp. We tested all four combinations - Valve-valve was sweet and warm as you'd expect. Tranny -tranny dry and cold (probably more faithful as long as volume's not too high). But most interesting was that the combination of valve preamp and tranny PA was indestinguishable from the valve combo. An inspection of the schematics showed that the Quad PA had overall feedback from the transformer to the input stage and was therefore linearised, but the preamp used a single triode flat out for gain as the input stage. Interesting... hence my desire (some 20years on ) to build something with hot wire and vacuum...

Enough for now.

Tuesday, July 21, 2009

Well, being a technocrat doesn't make me an early adopter - infact I'm quite the technology luddite. This is my first foray into the 'blogosphere' only some 9 years behind the form. Why now? I have no idea. Its probably just straight egotism - a desire to share my thoughts with people and get some form of feedback (preferrably positive) .

You'll be likely to find thoughts on a rather diverse range of subjects here, given that my interests include Photography,Music,Astronomy,Woodwork,Audio etc. etc.

OK so much for the introduction. Now all I need to do is find the time to write something worth reading...