A lot of my decision-making is based on a basic fact: a camera is simply a device for capturing light. The more light a lens can take in, and the more accurately, the better the lens. The larger the area a camera records light onto, and the more accurately, the better the image. Yes, there are many other factors, but all else being equal, it is indeed this simple.
In a digital camera, the light is captured by the sensor. In an analogue camera, the light is captured by photographic film, which is simply a light-sensitive emulsion bound to a plastic backing layer. The size of the sensor or the size of the film frame dictates image quality. A larger sensor/frame produces an image with greater dynamic range (the difference between darkest black and lightest white) and less noise than a smaller sensor/frame.
If you have a point and shoot digital camera it likely has a 1/2.5" sensor that is 5.76×4.29mm in size. You might have a superior model such as the Canon G9 that uses a 1/1.7" sensor that is 7.6×5.7mm in size. The Four Thirds System, designed to be an interchangeable lens system that is also portable, is 17.3×13mm in size. I currently shoot with Pentax digital SLRs, which have ASP-C sensors of 23.6×15.7mm. This is the same size as DX class Nikon cameras; though Canon is slightly smaller the difference is not significant.
We commonly measure these sizes relative to a frame of 35mm film (36×24mm), rather arbitrarily designated "full-frame". Anything less than that is considered a "cropped" sensor (though this term is a poor one since it is only analogous with the more common usage of cropping a photo to reduce its size).
To get a useful measure, we first look at the diagonal size of each sensor. For the formats I discussed above, these are 7.18, 9.5, 21.64, 28.35 and 43.27mm respectively. Now, if we take the ratios of these relative to 35mm, the result is this list of crop sizes:
35mm.... 1
APS-C... 1.5
4/3..... 2
1/7".... 4.5
1/2.5".. 6
Now we can see easily why a point and shoot camera, no matter how well made, no matter how good the lens, no matter how many megapixels, simply cannot compete with an SLR for image quality. The sensor is many times smaller than full frame. Similarly, the Four Thirds system has made a large compromise to reduce the size of their cameras; the crop factor of two is significant. (But whether this matters to you depends on what other photographic priorities you may hold.)
Nikon, Canon, Sony and now Leica all offer a full-frame digital camera (though the Leica is a rangefinder, not an SLR). There is much gnashing of teeth on the Pentax forums I frequent as to the fate of a company, like Pentax, who "only" offers a 1.5 crop factor camera.
My point of view is that while there is certainly an advantage in stepping up to full frame, the net gain is not as significant as some proponents would have you think. The gain from point and shoot to Pentax digital is 4x. The gain from Pentax to full-frame is only 1.5x. This might not be noticeable depending on your application.
So how is full-frame particularly advantageous? First, in situations where you need to boost ISO to photograph in low light, full-frame will produce less noise. Second, when you need to enlarge, full-frame will permit a print 1.5x larger diagonally than APS-C. And thirdly, because of the relationship of the crop factor to angle of view, full-frame allows for a wider field of view for the same focal length. This is important to photographers interested in extreme wide angle shots.
Let's shift our perspective a moment. The ancients among us can remember that far from the pinnacle of professionalism it is often viewed as today, 35mm film was introduced as a cheap and easy format for amateurs. It was an alternative to large format cameras and their smaller siblings, medium format. Indeed, 35mm was the runt of the litter!
So why not look at digital sensors larger than 35mm film? The main reason is price. For example, the Hasselblad H3DII-50 has a 36×48mm sensor with a 60mm diagonal. That works out to a "crop factor" of 0.72 (though the term hardly makes much sense once the numbers get below one). For this 40% increase in quality you pay $28,000. And yes, that's the current selling price with one lens, not the original list price, which was much higher. This puts it well beyond the reach of any but professional users.
Photographers who want better image quality but are not made of gold are not without recourse. Medium format film cameras are still available. And because camera bodies were once made to last, they likely need little more than a cleaning before they can be pressed into service once again.
While some medium format cameras can be quite bulky, the smallest size in this category, 645, is available in bodies not much larger than full frame digital. These are also SLR cameras and so have many of the advantages of contemporary cameras. The name refers to the frame size of 6×4.5cm (though the actual image is 56×41.5mm). The image has a diagonal of 69.7mm and a crop factor of 0.62. That's even better than the Hasselblad just mentioned!
Why not take the 645 format as our new baseline? If we do so, we get the following ratios:
645..... 1
H3D..... 1.2
35mm.... 1.6
APS-C... 2.5
4/3..... 3.2
1/7".... 7.3
1/2.5".. 7.3
I find this quite revealing. On this chart APS-C and 35mm don't look so different after all. I guess it all depends on your perspective, a lesson particularly pertinent to photography.
RELATED POSTS
4 comments:
Hi there - I've read a few of your posts with great interest - thanks.
The difference between the sensor sizes is greater than you state. In terms of area the full frame is 2.5 times greater than the aps. This really changes how a lens functions - it has to do with mtf curves - don't ask.
The upshot is that a FF of the same megapixel count, using the same lens as an aps will have around 50% more actual resolution. Additionally it will record about twice as many colors.
For a lot of stuff it's not so important - but for critical work it really can be.
I haven't mentioned the point and shoots because even the best is fighting an uphill battle. And the medium format is too expensive for many to consider.
I would like to see a convincing argument for whether a linear or quadratic measure is better for measuring sensor "quality", so if you have one, send it along.
In terms of lenses I believe that it is easier to make one for a larger sensor, since it does not need to be so precise. The denser the pixels the harder a lens has to work to resolve them.
If you use a FF lens on an APS-C body only the center of the glass is used, so light fall-off and certain other undesirable qualities are minimised. However, only the very best such lenses will perform optimally.
There is indeed a distinct difference between APS-C and 35mm, as I point out in the article. I wanted only to indicate that much larger differences to be found by moving up to medium format. I have rewritten two key sentences a little to clarify this context.
Argument why 'quadratic' measure is better than 'linear':
The quality (resolution, dynamic range, signal to noise, colour recording) of any recorded image is dependant on the number of photons falling on the sensor (film or digital). No matter how hard you try with clever processing to reduce noise, expand dynamic range, up-scale pixels (or what ever else you try to do) the more information from the original will give you better results. The amount of light falling on the sensor is proportional to its area, not length of any side or diagonal so the quality difference between two (otherwise similar) sensors is related to the ratio of their areas - hence a 'quadratic' measure is the better one to use.
This article is stale. Much better to read Equivalence of Camera Systems.
Post a Comment