Shutter Speed Tests

jasolo

18

Re: Shutter Speed Tests

« Reply #20 on: 03 / January / 2010, 17:28:10 »

Advertisements

I'm testing the max values of aperture and shutter speed of the S3. Using a script with set_tv96_direct and set_av96_direct functions I obtain several images of the same scene (this time overcast sky, not sunny

) with different Av and Tv values (in steps of one step). Then I get the histogram from the raw files:

dcraw -D -4 IMG_XXXX.CRW (makes a PGM image without processing)
I've written a small C program to calculate the histogram from the PGM for each channel.

From the histogram, I study the evolution of the coordinate x of some peak. For example, when x=1000 with exposure EV=14, then EV=15 (one step, half light) will show the peak at x = (1000-31)/2 + 31 = 516. Remember, x=31 is "black" in these cameras.

Well, my first test with ISO100, 36mm, f2.8 to 16 and t =1/2048 to 1/65536s shows that f8 is the max value for aperture (as many users have said before). For the shutter speed, nothing wrong up to 1/8192s, but the next three values don't match the expected values:

- 1/16384s (1 step) behaves as a 1/2 step in exposure, that is, 1/11585s.
- 1/32768s is really 1/16384s.
- 1/65536s is the same: 1/16384s.

So the max shutter speed at 36mm (wide angle) seems to be 1/16384s but you must indicate 1/32768s to obtain it

... or maybe it's a bug of the set_tv96_direct function I've used in the script? At least, ISO value (using AutoCHDKtoEXIF) says that the selected shutter time was 1/65536s when using "set_tv96_direct 1536". I'am using s3is-100a-0.9.9-863 firmware.

The next test will be with a focal length of 432mm.

Logged

e2b

37

Re: Shutter Speed Tests

« Reply #21 on: 03 / January / 2010, 20:59:51 »

I did some similar (but qualitative) tests with an S3 and had similar results - see page 1 of this thread, http://chdk.setepontos.com/index.php/topic,232.msg1672.html#msg1672. The explanation is that the maximum shutter speed (minimum exposure) is limited by the inertia of the mechanism, which the manual over-ride settings do not make any allowance for.

Because it takes some time for the shutter to accelerate to maximum speed, the transition from 'normal' behaviour to the point where the shutter is moving at its maximum possible speed takes place over a range of speed settings.

The reason that you can get higher shutter speeds with smaller apertures (larger F numbers) is that the shutter has accelerated nearly to its maximum speed before it reaches the small area through which light passes through the lens.

With a large aperture (smaller F number) the full effect of the inertia is more evident, because the acceleration happens inside the area through which light is passing, and therefore it affects the exposure more. The result that at low F numbers the minimum effective exposure is longer (effective maximum shutter speed is slower), and in addition there is a wider transition range over which the effective speed deviates from the speed that you specify with the manual over-ride.

« Last Edit: 03 / January / 2010, 21:17:14 by e2b »

Logged

jasolo

18

Re: Shutter Speed Tests

« Reply #22 on: 04 / January / 2010, 17:49:29 »

I have made the test with max focal length (426mm), but again it was overcast so the data from the histograms were very small for a good statistic. I'll repeat it on a sunny day. Anyway, at low f values I don't get faster shutter speeds than with 36mm. Other data are yet known: at 426mm f range is 3.5 to 11.

I'm not sure which shutter system is used on the S3. I supose that the diaphragm is used, instead of curtains (DSLRs). S3 has also an electronic shutter, but this type maybe is used for video.

« Last Edit: 04 / January / 2010, 17:51:39 by jasolo »

Logged

dzsemx

302
[SX40HS][A650][A630]

Re: Shutter Speed Tests

« Reply #23 on: 07 / January / 2010, 16:12:41 »

all digital cameras use for main shutter the electronic one.
they have speed limitations, because they was not designed for extra high speed shooting

the mechanical shutter is only used to protect the sensor from light, while reading out the ccd, to avoid vertical lines, as seen on videos when pointed to a ligth source.
without this, the photos are pretty bad.

video mode uses only electronic shutter 30 times/sec for every frame. actually it takes 30 photos/sec. the readout method is different, it's done with reduced line numbers, like 480 lines

cmos sensors do not need generally mechanical shutter (like cameraphones) and generally they shoot better video, without vertical lines, but many of them use rolling electronic shutter, which creates the jelly video effect...

Logged

jasolo

18

Re: Shutter Speed Tests

« Reply #24 on: 10 / January / 2010, 16:32:34 »

New test, this time during a sunny day. As e2b did, I have set ISO to 400 to obtain bigger numbers, but the noise in the histograms has also increased. So I have to smooth each histogram before checking the position of its peak (specially in the most exposed images). Maybe averaging several raws could be a solution, but the subject is a portion of sky near the sun, so the exposition changes slowly along the time.

For ISO400 and max focal length (426mm) the results are (rounded values):
- Setting a value of 1/16384s, I obtained from 1/9000s (f3.5) to 1/15000s (f11).
- 1/32768s and faster ones gives 1/10500s (f3.5) to 1/22000s (f11).

The position of the peak for the fastest speed is x=46 in the green channel, so an error of x=1 in the position of the peak gives values between 1/23700s and 1/20800s.

About the mechanical and electronic shutter topic, in my S3 the diaphragm remains open during the shot and another part closes the hole. Also, the diaphragm gets its final aperture once the shutter-release button is pressed or half-pressed. In the script I wrote to get the photos, I only use the "shoot" command after setting aperture and time, so I'm wondering if I should add a [click "shoot-half"] line before the "shoot" to avoid any possible inertia of the diaphragm. Otherwise, the limits will be in the closing part.

Logged

NewbieToobie

102

Re: Shutter Speed Tests

« Reply #25 on: 11 / January / 2010, 04:58:04 »

Quote from: dzsemx on 07 / January / 2010, 16:12:41

all digital cameras use for main shutter the electronic one.
they have speed limitations, because they was not designed for extra high speed shooting

the mechanical shutter is only used to protect the sensor from light, while reading out the ccd, to avoid vertical lines, as seen on videos when pointed to a ligth source.
without this, the photos are pretty bad.

Not true. If you study the bokeh of the sparks in this photo from the CHDK Wiki you can see that a diamond-shaped aperture starts out wide and then closes to nothing during the travel of the sparks.

If it was an electronic shutter the spark tracks would remain the same width during their travel, starting and ending abruptly with llittle change to their track width over such a short distance during the short exposure. Unless they were traveling directly away from or toward the camera. All the sparks from the flint-wheel in this photo are traveling roughly on the same plane. Spin a flint-wheel on a lighter and you can see for yourself that the vast majority of sparks are all traveling in roughly the same direction as that imparted by the spin of the flint-wheel.

It is the mechanical aperture that is setting these high shutter speeds, not an electronic one. They work together but the mechanical aperture is the one doing the actual shutter work at these high speeds.

Logged

jasolo

18

Re: Shutter Speed Tests

« Reply #26 on: 14 / January / 2010, 16:17:50 »

New results: there is no difference in the histogram between using the "shoot" command and adding a previous [click "shoot-half"] command. So no inertia problem due to the aperture adjust.

About the spark tracks, I have found a similar behaviour but without moving parts. The next image shows a blue LED shot by a S3IS using the max focal length and out of focus. Each row has a different aperture: 3.5, 5.6 and 11. From left to right the time exposure was increased one step (double speed) from 1/2048s, although real speed will be inferior to the expected one (as noted in my previous post).

At low speeds you can see a rounded hexagonal shape (due to the diaphragm blades?). At high speeds that shape is distorted, maybe because of the shutter. So it seems the diaphragm is static and the shutter is another part.

Logged

flenser

9

Re: Shutter Speed Tests

« Reply #27 on: 18 / January / 2010, 03:54:19 »

I posted the results of some shutter speed tests on my A570 http://chdk.setepontos.com/index.php/topic,3576.0.html. I used sunlight on a bright cloudless day as the light source and I did my tests at one aperture setting only.

Instead of trying to judge the results by eye I wanted to "measure" the exposure and to do this I copied the method described on the Clarkvision site http://www.clarkvision.com/imagedetail/evaluation-1d2/index.html.

Convert the raw file to a TIFF using dcraw
extract just one of the Bayer Array green pixels from the TIFF file (effectively creating a monochrome version of the image using just that green pixel)
Get the mean value for a 200x200 pixel selection at the centre of that monochrome image

These indicated that on my A570 the fastest possible shutter speed was about 7/100K sec and the shutter speed can be altered down to about 10/100K sec while still operating pretty linearly (i.e the exposure at 10/100K sec is 1/2 the exposure at 20/100K sec and 1/4 the exposure at 40/100K sec). See for the details.

I also did a set of tests for the ISO Override feature on my A570 IS in March '09 for which I used an incandescent lamp as the light source http://chdk.setepontos.com/index.php/topic,3373.0.html.

The lowest ISO I achieved using was equivalent to the stock ISO80 setting and the highest was only fractionally higher than the stock ISO800 setting. While the available ISO settings were no more than the stock settings, using the manual override does allow you to choose from a bigger set of ISO values. eg 100, 125, 160, 320, 500, 640, etc instead of just the stock values 80, 100, 200, 400 and 800.

Logged