The crop factor of the Q will make it interesting!

Started Jun 23, 2011 | Discussions thread
bobn2
bobn2 Forum Pro • Posts: 62,445
Re: Again, improperly stated.

PicOne wrote:

bobn2 wrote:

Of course, I don't think anyone is talking about 'exposure purposes' we're talking about photographic purposes. 'Effective' f-numbers go with 'Effective' focal lengths like a hand fits a glove. Look at your 35/2.8, it has an aperture of 12.5mm and a focal length of 35mm. Fit it on a four thirds camera, or simply crop your FF print to the same size, the effective FL is now 70mm, but the effective f-number, for any purpose other than exposure, must have changed, since the aperture is still the same, so the f-number is now effectively 70/12.5 = 5.6. Now crop again (or mount it on a Q) and the effective FL is 200mm, and the effective f-number is 200/12.5 = 16. This f-number is 'effective' to judge DOF and the amount of noise in the image for a given scene brightness and shutter speed, and hence the 'speed' of the lens. The only thing it isn't 'effective' for is setting exposure, but then the consequences of 'exposure' are not the same as they were, I don't think anyone goes into photography with the aim of setting a particular exposure - or at least, I've never seen a DPR challenge based on that objective, (let's see what you can do with 1800 lux, f/7.1 and 1/300 second)

I'm confused how aperture has an impact on noise.

Many people are, it comes from the effects of all those who push film-era (or, more accurately colour reversal and lab processed print era) notions of what exposure is all about.

Are you saying that if I took a shot with a FF lens on my FF camera, but underexposed the shot image by 5.6 stops, and then pushed it 5.6 stops in post processing, I would have the same noise exhibited as taking the same shot accurately exposed but the lens mounted on the Q camera?

First, let's get rid of the 'underexposed' and 'accurately exposed' notions. For the moment all there is is exposure, no 'accurate', 'over' or 'under', those are all relative to a notion of 'correct' which will be imposed in processing, not in capture. So, with a lens, any lens, on the Q, set at f/2.8, the sensor will receive the same amount of light for the same brightness of subject as will a lens, any lens, on a FF set at f/16 (which is 5 stops - sqrt(2)^5 = 5.7, the crop factor of the Q with respect to full frame). For the sake of convenience (though it's not essential to the discussion) let's assume that the FF camera has 12MP, something like the D700 or D3S (my guess is that the D3S is a closer match to the efficiency of the Q sensor). Since both sensors receive the same light, and have the same number of pixels the number of photons received in corresponding pixels is the same , and since the per-pixel photon noise SNR is given by the square root of the number of photo-electrons collected in each pixel, the per-pixel photon shot noise is the same . If both sensors contribute the same amount of read noise (and it's very likely that they are similar) then the number of photons per pixel is the same, the shot noise is the same and the read noise is the same - i.e. from simply reading the signal from the sensors, you could not tell which was which. If you 'processed' (by which I mean all the electronic and digital work downstream from the sensor) them exactly the same you could not tell from the photograph which camera had taken which picture (apart from the fact that the FF one would probably retain more detail, since it doesn't work the lens so hard with respect to resolution). How do we set the cameras up to process both the same? The answer is simple, set the ISO's 5 stops different.

In terms of your question, can this processing be done with a raw file, in the computer? That depends on the camera. Many DSLR's skimp on a full DR analog to digital converter, and so at low ISO's small signals (which is what we're talking about here) get swamped by the noise of the converter system. Those cameras you have to set the ISO differently. Other cameras can take it in their stride, such as this example from Pierre Sottas, of a D7000 image pushed six stops from 100 ISO

and one with the same exposure at 6400 ISO

(from http://forums.dpreview.com/forums/read.asp?forum=1034&message=36903045 )

The problem in comprehension comes because people think that it is raising the ISO which causes the noise, it isn't - it is reducing the number of photons received in the image.
--
Bob

Post (hide subjects) Posted by
Izu
Izu
Izu
Izu
Izu
Izu
igb
Izu
Izu
Izu
Izu
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow