An explanation on digital sensors size, and what it means for your pictures.

When buying a digital camera,  2 things require your attention: the lens, and the sensor. I'll write the next post about the lens, but for today let's focus on the sensor.

A bit of context first: image quality, appart from your talent, comes from the optical characteristics of the lens, and the characteristics of the sensor. Now with film, the sensor is the film, so the chemical characteristics of the film at the contact of light are responsible for the tones, noise, dynamic range of your image. 

When shooting digital, instead of photons hitting film, you have photons hitting a sensor that will send an electric signal, according to how many photons of a given "color" he receives (physics PhD please forget my vulgarisation).

Enough theory, my natural geekiness would quickly bore many of you. My point here is : the lens is like the chassis, the sensor is like the engine. Everything else (LCD screen, buttons here and there) are just options and they should not be the main things to look for.

I talk about sharpness, noise and dynamic range as variables of your image quality. Let's first see what they mean, then we'll see how sensor size impacts those variable, and for those who just want the final word, I invite you to skip to the end for conclusions :-).

• Sharpness is about image elements being distinct from one another. In other words, can you clearly see where something ends. It has nothing to do with resolution! Resolution is how many pixels you can count on the picture, the size of the image if you prefer. You can have a huuuuge resolution and a totally blurry image at the same time. Now sharpness is nearly 99% under lens influence, so don't get confused by bad salesmen bull-shitting you with mega pixels. If you care about sharpness so much, you can get a fix lens, but mostly you should start reconsidering your approach to photography :)

• Noise corresponds to sensors (or film) made so sensitive to light that they even see light when there is none. Those inaccurate dots on your image are what we call noise. We push a sensor to be ultra light sensitive when we want to shoot in low light. That sensitivity is what you change by raising ISO. The higher the ISO number, the more sensitive, the more noise the sensor is likely to generate. If you see that a camera goes up to 12000 ISO...well that means nothing. The noise level might be so high that image quality would be awful anyway. Also one thing to keep in mind: film noise can look real nice, when digital noise can be properly awful.

Ilford 3200 ISO film on a 6X7 camera (huge sensor equivalent): gives it a nice looking grain.

• Dynamic range is a little trickier to explain. Let's use an example: I'm sure you already took a picture of some one in front of the sky, and the sky came out totally white. Or the person is coming out black and the sky is blue, and you can't see no details at all on the subject. Dynamic range is the maximum difference between dark and bright that your camera can handle. It's terribly important for landscape in particular, because a valley, a snowy mountain and the sky can have very different brightness level. With poor dynamic range, you only have one part correctly lit, the rest comes out all white or all black. Ever wondered what HDR is? It's when you take several pictures of the same thing, at different light exposure, and pile up picture to make sure everything is lit properly. It's nothing more than faking very large dynamic range.

Courtesy of http://kimletkeman.blogspot.com/ on this image you understand how DR can be important: you want the shades and the far right part to still be detailled, not just one of the 2.

Ok now back to today's topic: how do sensors impact those image characteristics?

To cut corners a bit, the bigger the sensor the better. 

With film, portrait is done with medium format, up to 60x90mm size film. Landscape is done with large format cameras with even bigger film surfaces. 

What we call a 35mm film camera (24x36mm) is the pocket format, the film equivalent of a compact, the low quality stuff...And that's what we call full frame in the digital world! The top of the range Canon 5D's and Nikon D3 and D700 are "only" 24X36 sensors cameras, also called FX. 

- Most DSLRs are DX camera's, not even a full 24X36 sensor. They are 17X25mm sensors also called APS-C. All the Nikon D3100, D5100, D90, D300, Canon 7D, 550D and other are DX cameras.

- Compact cameras have even smaller sensors. Rather than going thru each of them, look at the characteristics of the camera you're into, and see in the table below what is the sensor size:

Sensor size is crucial to noise and dynamic range for a simple reason: the bigger the sensor, the larger the pixels for a given picture size. 

For example, compare a 10MP medium format (MF) camera with a 10MP compact : same amount of pixels, but the MF camera's sensor has a surface about an hundred time larger. The pixels are then a hundred times bigger. Each pixel being bigger, it can rely of a much larger amount of light hitting it to gather information: it is more sensitive in low light, and generates less noise as it doesn't take decision out of very little amounts of light.

Same with dynamic range: the pixels need less exposure to collect the proper amount of light in the dark areas of you image. So while they get exposed for brighter parts of the image, they still receive enough information to obtain detailed dark objects.

Last but not least: depth of field.

DOF is the distance between the nearest and farthest sharp object on your image.

Long story short, the smaller the size sensor, the more DOF you get. It means that small sensors, i.e. most compact cameras can hardly blur the background behind your subject. Not that it is a necessary thing to do, but it limits your creativity by prevent you from exploring a key variable in photography.

One way of allowing less DOF (=more blur behind) is to have lenses that open very wide (those F1.8, F1.4 lenses), but few compact camera offer that, if any.

Very little DOF: you will NEVER be able to do that with a compact due to sensor size.

If you are into portrait, don't spend 500$ in a premium compact! Get a used Nikon D90 or Canon 550D with a 50mm 1.8 lens, for about the same price.

CONCLUSION

- The bigger the sensor the better

- Bigger sensor = capacity to shoot in low light

- Bigger sensor = image with better exposure, details in dark zone, which is crucial for landscape

- Bigger sensor = the possibility to get a very sharp subject with a very blurred background, ideal for portrait.

Let's force manufacturers to stop making those mini sensor cameras, when you buy GO FOR BIGGER!!! 

My choice? The only compact with a bright lens and a APS-C size sensor (same as most DSLR) is the FUJI X100. Otherwise go DSLR.

The upcoming Fuji X10 will have a smaller sensor, so careful with the appealing feature, the X100 might still be the best camera for you if you shoot more portraits !

This being said, a rubbish camera can still pull the best picture in the world. So go out and shoot whatever you have :-)

Portrait "how-to" part 2: advanced composition, use of depth of field

Welcome back ! Last time we saw how to set up your camera in order to expose correctly for portrait, and introduced some basic rules for composition.

Today we'll simply see some example pictures, and explain why those are interesting. This is in no way exhaustive, or properly academic; my point is to trigger your creativity, by having you noticing different techniques and ideas that have been used by other photographers. Then you're on your own, just go out and shoot to find your way.

We'll conclude by a little "how to" on depth of field", really basic.

• Different types of composition & framing.

Photo by Maree

If your model is looking straight ahead, you can center the subject, but make sure the eyes are high up in the first third of the picture. Don't hesitate to cut the top of the head, it gets the viewer closer to the face, the eyes, and creates proximity. Having the eyes in the middle and empty space on top of it, unless the subject looks up, can look cheap.

Photo by Tanel Varek.

If you shoot a full body portrait, and even if it feel like your shooting it centered, the rule of thirds apply. Look at how the horizon on the top third line, and how, overal, space around the subject is split in thirds. For a very academic portrait like this one, it guarantees the adequate balance. Also notice, in line with part 1, that exposure was done on the model, not the entire image. If the sky was correctly exposed, i.e we could see the details, clouds etc, the model would be too dark.

Photo by Dave Tata.

Here the photographer cares mostly about the scene, rather than the face, hence taking a little distance but yet cropping the legs. Once again the rule of thirds, and free space where needed, face is top right, looking down left.

Finally an important part of that picture: the light. It was necessary here to capture it in order to restore the mood of the actual scene. By using techniques explained in part 1, i.e. spot metering, the photographer was able to maintain a correct exposure on the face, and get that halo of light, even though shooting against the sun. So yes you can shoot against the sun, just meter on the subject.

It's important, when you see a scene you want to capture, to think about what part of the scene makes it interesting. If it's someone's reaction to a situation, you might want that situation to somehow appear too, it it's an intense facial trait, you'll want to be has close has possible to remove any disturbing element. Try to remove everything that is not necessary to tell your story, all useless subject will be distracting. If your framing does not seem to work, move, turn around the subject, don't be lazy :)

Photo by me.

Look at this on as an illustration. On that photo the baby's eyes are top right corner (rule of...ok ok I said it enough), leaving room from his mum on the foreground. We can see her back and she's blurry, which allows the viewer to understand enough yet have room to imagine. The rest is blur since it is not relevant. Keep reading for the blur part.

Position yourself in line with the 2 subjects, don't think "I have to step back to get the entire scene". Again, baby's head is cropped, because we don't need it all on the image.

• Using depth of field.

I explained in my lens guide what DOF is about. It is of the highest interest for portraits as it allows you to extract your subject from it's environment. See for yourselves (also consider the baby above):

Photo by me, 85mmF1.8.

Photo by me 110mmf2.8 (medium format)

How to minimize DOF (i.e. subject sharp, everything else very blurry)? It's fairly simple:

- maximizer the aperture (low "f" numbers, below f2.8 and down to f1.4 to get such results)

- Increase the distance from your subject

- Use a long enough focal length (from 85mm and up it gets really good)

Long story short, set your camera to aperture mode, fix the aperture under f2.8 if your lens allows it (once again, the importance of buying the right lenses can't be highlighted enough).

Speed will adjust automatically, usually it'll be very high so you can shoot fast moving subjects.

Pay attention to one thing though: since the focus is going to be on very narrow depth, if you are focusing on the wrong area, you might totally miss the shot, by having for example the nose sharp and the eyes blurry !

All of it being said, keep a last thing in mind : fuck the rules. Look at other photographers' work for inspiration (check those links), painters, and try for yourself. Use the rules in a vicious way, do the exact opposite, there are no rules, it is art. Just keep those in mind since, appart from rare cases of pure talent, you often need to start with rules to deconstruct them with creativity.

Now stop reading this blog and go shoot !