about manual exposure

Some photographers are very proud to say they use manual exposure mode for tuning their camera. When shooting in a studio with flashes, yes, I agree... ISO set to minimum to maximize the image quality, speed set to synchro flash to neutralize ambiant light and aperture set accordingly to the quantity of flash light: this is the common way. But, in a common context, why using manual exposure?

Let's remind the basics of tuning a camera exposure thanks to a light cell embedded into the camera that measures the average (weighted average, multi-zone,...) luminance L in cd/m2 (candela per square meter). As said in a previous post, tuning a camera to get a correct exposure amounts to move in a 3+1 dimensional space with one contraint. If each dimension is represented by a variable: N the f number (1:5.6 → 5.6) defining the aperture, S the arithmetic sensitivity in ISO and t the exposure time in seconds, then 1 variable is dependent on the 2 others. As a reminder, I say 3+1 and not 3 dimensions, because an extra-light quantity, brought for instance by light flashes, can be used to modify the ambiant average luminance. But for now, let's forget the extra-light.

Let's write the constraint as a relation between the 3 mentioned variables.

The higher the aperture (lower f number), the shallower is the crisp area. The lower the ISO sensitivity (less amplification of the sensor signals), the less grainy is the picture. The smaller the exposure time, the more motion can be frozen on a photograph. It means that all the possible solutions of the 3-dimensional space, with one constraint, are not equivalent.

In most modern cameras, automatisms make it easy to evolve in the space of possible solutions corresponding to a measured luminance. In case the luminance is not properly measured, there is an exposure compensation to fix it. ISO amplification mechanism can be fixed with an auto-ISO mechanism that set the ISO amplification such as the minimum time exposure is fixed to avoid blurred pictures due to camera shaking, given an aperture, in case of aperture priority. Therefore, the exposure time t and the ISO amplification are directly resulting from the aperture. The exposure time can also be set. In that case, the aperture is set according to a rule to avoid too much ISO amplification.

Whatever the context is, after correcting the measured average luminance, the solutions are lying in a contrained 2-dimensional space. Therefore, manual exposure that let the 3 variables independent is not meaningful: it's a kind of snobbery.

To make good photographs, it's much better to make it easy with the camera tuning to direct the whole attention to the image capture: subject, framing,...

As a final thought, I would add that raw shooting accept much more tolerance in irrelevant average luminance measurement... It's going in the same direction than automatisms. Personnally, I recommend to control the aperture because it has a strong impact on the photograph, and auto-ISO that adjusts ISO amplification in order to avoid blurred photographs if possible. If needed, the rule used by auto-ISO mechanism can be forced. Shooting should be as easy as possible to let the photographer focusing on making great pictures. Old time photographers with manual exposure and even without embedded cell were compensating this complexity by tuning their camera before a interesting situation occurs. It's maybe a justification for using manual exposure: when tuning the camera in advance...

what about focal reducer or speedbooster?

When using lenses designed for bigger sensors thanks to a mount adapter, the resulting illumination circle is bigger than what is needed... the light is absorbed by the mount adapter i.e. it is lost... although only the best parts of the lenses are used: the lenses borders, which are known to be softer, are not used. An interesting idea is to use a speedbooster, also named focal reducer, which is concentrating the illumination circle to a smaller one that match the actual camera sensor size. Metabones proposed it first, but much cheaper ones have been proposed by Zhongyi (LensTurbo), Pixco, Viltrox, etc...

Using a focal reducer with lenses for 24 x 36 sensors on a APS-C camera will increase the aperture by 1 stop (for instance, when it is written 1:8 on the aperture ring, you have to understand 1:5.6, and if the lens maximum aperture is 1:2, the actual maximum aperture is equal to 1:1.4!). The angle of the focal lens is also changed: it remains the same than the one obtained for a 24 x 36 sensor. No need then to multiply by a scale factor to get the 24 x 36mm equivalence. If full frame lenses are used with a micro four third sensor, aperture is increased by 2 stops.

Manufacturers use to suggest that, when using focal reducer, the resulting image quality is better than without focal reducer because of the bigger image concentrated on a smaller surface. Nevertheless, there is extra glass... that might affect image quality.

Let's do some experiments before concluding. I am going to use a Fujifilm X-T2 camera with an APS-C sensor and 2 full frame lenses: an ukrainian Helios 44-2 1:2 focal length 58mm, with a Zhongyi Lens Turbo II M42-FX, and a (excellent) Micro-Nikkor AI-S 1:2.8 focal 55mm with a Pixco NG-FX focal reducer. The Zhongyi is famous although the Metabones speedbooster is more: here, they are compared. The conclusion is that they are closed but with a little advantage in the image quality for the Metabones one. However the price had discouraged me from buying it (from 300 to 500€).

I used a tripod and took the following shots with lamp on (yellow square) and off in order to appreciate the resulting flare. Then, I focused on the upper left corner.

• Helios 44-2 1:2 58mm with a Zhongyi Lens Turbo II M42-FX

The picture excerpts have been ranked according to image quality from left to right and top to down. 2 apertures have been tested: 1:5.6 and 1:2.0, without focal reducer ("helios-xx") and with ("helios-red-xx"). Yellow square means the lamp was on.

The ranking is:

1. 1:5.6 no lamp

2. 1:5.6 lamp

3. 1:5.6 no lamp with focal reducer

4. 1:5.6 lamp with focal reducer

5. 1:2.0 no lamp

6. 1:2.0 lamp

7. 1:2.0 no lamp with focal reducer

8. 1:2.0 lamp with focal reducer

The Zhongyi Lens Turbo II M42-FX decreases the image quality and increases the sensitivity to flare. The loss of quality is more important at large aperture. However, it may be argue that the comparison is not fair because one stop should be removed for a fair comparison. Indeed, 1:8 → 1:5.6 with focale reducer and 1:2.8 → 1:2.0. Let's see the results (lamp off).

The ranking is now:

1. 1:5.6

2. 1:8.0 with focal reducer

3. 1:5.6 with focal reducer

4. 1:2.0

5. 1:2:8 with focal reducer

6. 1:2:0 with focal reducer

Increasing by one stop improves slightly the image quality must not suffiently to reach the image quality without focal reducer.

• Micro-Nikkor AI-S 1:2.8 55mm with a Pixco NG-FX focal reducer

Same experiment that before but testing at aperture 1:5.6 and 1:8 (right hand side is with lamp on, and off on the other side, top is without reducer and down with).

Here also the focal reducer decreases the image quality but it is not sensitive to flare.

• upper left corner, lamp on: 1:5.6, higher image quality

• upper right corner, lamp off: 1:5.6, higher image quality (no change)

• lower left corner, focal reducer and lamp on: 1:8.0 (eq. 1:5.6), lower image quality

• lower right corner, focal reducer and lamp off: 1:8.0 (eq. 1:5.6), lower image quality (no change)

Maybe the extremely expensive Metabones focal reducers are better but does it worth to pay such a price? Although the decrease in image quality, focal reducers behave properly when shutting down the aperture and, together with a simple mount adapter, you can get 2 focal length with an old lens.

depth of field and sensor size

I am gonna compare APS-C and 24x36mm sensors.

For a lens projecting a rectilinear image (focused at infinity), the angle of view (α) can be calculated from the chosen dimension (d), and effective focal length (f) as follows:

d represents the size of the sensor (or film) in the measured direction.

For 35mm film which is 36 mm wide and 24mm high, d=36 mm must be used to obtain the horizontal angle of view and d=24 mm for the vertical angle.

For APS-C sensors, the size is 23.7 x 15.6mm. So, we can compute the angle with respect to the focal length and notice how it is nonlinear: a difference of 7mm at wide angle lens area between 9mm and 16mm yields 30° whereas a 7mm difference at tele lens area around 150mm doesn't lead to any perceptible variation in the angle.

Here is another more convenient representation, I found on the web, for 24x36mm sensors:

Now, 2 more convenient representations I did for APS-C sensors:

Let us see now, for APS-C sensors, what is the minimum shutter speed that can be used without tripod, flashes or image stabilization:

It's also nonlinear and we can see why it's not easy to use a tele-lens: shutter speed must be very fast for crisp images and the maximum aperture is usually smaller (f/2.8 max) than standard and wide angle lenses.

Let's now check the crisp area for different focusing distances and wide angle lenses, for APS-C sensor still. X axis stands for aperture (with vertical red line for common ones) and Y axis stands for the distance in meters: the lower curve models the closest distance for the crisp area and the upper one of the same color models the furthest crisp distance for a given focale length.

And now for standard lenses (tele-lenses have of course very narrow crisp area... very limited kinds of photographs):

Another way to appreciate the effect of focal length on depth of field is to analyze the hyperfocal focusing distance i.e. the focusing distance that will lead to crisp images from infinity to the closest distance. The dotted line stands for the hyperfocal distance and the plain line for the closest crisp distance:

Once again, large crisp area can be obtained with wide angle length.

Now, it's interesting to compare 24x36mm with APS-C sensors because there are not equivalent. Actually, if a 23mm focal length for an APS-C sensor is equivalent in terms of angle to a 35mm focal length lens for a 24x36mm sensor, the depth of field of the 23mm is the one you get with a 23mm lens, whatever the sensor is: it's an optical property. To make it easy, let's analyze separatly different focal lengths. There are 4 curves for each focal length. The upper left corner contains the hyperfocal distance and the closest crisp distance at hyperfocal. The upper right corner and lower curves contains the closest and furthest crisp distances for different aperture.

1) focal is 9mm. At aperture f/2.8 and focusing distance at 1m, the crisp area is 50cm → ∞

2) focal is 15mm. At aperture f/5.6 and focusing distance at 2m, the crisp area is 1m → ∞

3) focal is 21mm. At aperture f/8 and focusing distance at 1.5m, the crisp area is 1m → ∞

4) focal is 25mm. At aperture f/8 and focusing distance at 2m, the crisp area is 1.5m → 8m

5) focal is 28mm. At aperture f/8 and focusing distance at 3m, the crisp area is 2m → 8m

6) focal is 35mm.

7) focal is 50mm

8) focal is 90mm

I conclude that increasing the size of a camera sensor has limitation because of the optics. As we saw, it's easy to get narrow depth of field with a 50mm focal length and more. It's all the more easy that very large maximum aperture, close to f/1, use to be available for small sensors because lenses are smaller... What is difficult to obtain is large crisp area... A standard angle is 35mm for APS-C, 50mm for 24x36mm and 70mm for GFX medium format! Just remember the previous curves. The difficult question is not how to obtain bokeh but how to obtain crisp images. From that point of view, smaller sensors are much better... and smart-phone cameras perfectly adapted to capture crisp images...