Pages: [1]   Go Down

Author Topic: If sensor DR exceeds subject DR by 5EV, does +4EV in ISO worsen detail, noise?  (Read 2049 times)

markwilliam

  • Newbie
  • *
  • Offline Offline
  • Posts: 18

If a sensor's dynamic range exceeds a subject's DR by 5 EV, then does a 4 EV increase in ISO cost much detail, add much noise? For example: a painting whose dynamic range is 7 stops (as lit in studio), imaged on a sensor whose dynamic range is 12+ stops and whose baseline ISO is 100. If one exposes to the right without clipping highlights, then will the fact that all the shadow detail is at least 5 stops to the right of the left edge of the histogram mean that one can shoot at ISOs higher than baseline ISO without compromising detail and increasing noise? If so, how much higher?

I have read (see quote below) that this is the case for a scanning back; if it is, then how much less true is it for a DSLR?

To explain:

A few years ago, a superbly helpful representative at BetterLight wrote this:

"[This scanning back] has an 11+ stop dynamic range, and seldom does fine art repro exceed 6 or 7 stops, which means we theoretically have 4 or 5 stops of overhead to use for increasing the ISO without intruding on the detail, so let's see... ISO 200 to 400 to 800 to 1600 is 3 stops while still leaving a stop or two of "underhead" at the highlight end of the scale. We are not film and we are not instant capture, so 2 or 3 stops of gain dialed in by increasing the ISO should have no effect on shadow detail because there is no "black" or "white" in any painting that is the equivalent of Zone I, II, or III or Zone IX or X. Black paint is not the same as black velvet or the deep shadows found in grass under the shade of a bush under the shade of a tree under the shade of a cliff wall. Practical experience in the studio shows that our Super 6K HS with a Base Line ISO of 200 can be easily turned up to ISO 800, 1200, even 1600 without concern of noise in the deeper tones. ISO 2400 and 3200 (and a little beyond) are available for problem shots under special circumstances but noise becomes visible at these settings."

But in cases where one must use a single-capture DSLR, how much of this would fly out the window?
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

Increasing ISO increases noise, four times the ISO - two times the noise. Detail is not really affected by underexposure, unless noise reduction is applied. On the other hand noise can cower over fine contrast.

The low end of the grayscale is normally dominated by readout noise. Cameras like Canon amplify the signal when increasing ISO (within limits) and can thus reduce noise. Cameras using sensors of more modern designs (most Nikons, all Sonys) gain little from increasing ISO.

If you work with JPEG, the camera's ASICs will apply noise reduction at higher ISO, so there will be a loss of detail.

Best regards
Erik

If a sensor's dynamic range exceeds a subject's DR by 5 EV, then does a 4 EV increase in ISO cost much detail, add much noise? For example: a painting whose dynamic range is 7 stops (as lit in studio), imaged on a sensor whose dynamic range is 12+ stops and whose baseline ISO is 100. If one exposes to the right without clipping highlights, then will the fact that all the shadow detail is at least 5 stops to the right of the left edge of the histogram mean that one can shoot at ISOs higher than baseline ISO without compromising detail and increasing noise? If so, how much higher?

I have read (see quote below) that this is the case for a scanning back; if it is, then how much less true is it for a DSLR?

To explain:

A few years ago, a superbly helpful representative at BetterLight wrote this:

"[This scanning back] has an 11+ stop dynamic range, and seldom does fine art repro exceed 6 or 7 stops, which means we theoretically have 4 or 5 stops of overhead to use for increasing the ISO without intruding on the detail, so let's see... ISO 200 to 400 to 800 to 1600 is 3 stops while still leaving a stop or two of "underhead" at the highlight end of the scale. We are not film and we are not instant capture, so 2 or 3 stops of gain dialed in by increasing the ISO should have no effect on shadow detail because there is no "black" or "white" in any painting that is the equivalent of Zone I, II, or III or Zone IX or X. Black paint is not the same as black velvet or the deep shadows found in grass under the shade of a bush under the shade of a tree under the shade of a cliff wall. Practical experience in the studio shows that our Super 6K HS with a Base Line ISO of 200 can be easily turned up to ISO 800, 1200, even 1600 without concern of noise in the deeper tones. ISO 2400 and 3200 (and a little beyond) are available for problem shots under special circumstances but noise becomes visible at these settings."

But in cases where one must use a single-capture DSLR, how much of this would fly out the window?

Logged
Erik Kaffehr
 

markwilliam

  • Newbie
  • *
  • Offline Offline
  • Posts: 18

Thank you for your reply, Erik. So does this mean that you believe the BetterLight representative is incorrect to claim what he did about "4 or 5 stops of overhead to use for increasing the ISO" when doing copy work of paintings? Or that you believe that DSLRs would have no such "overhead"? Or both?
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

A scanning back, like the Better Light is more like a scanner than like a camera, I guess. If you increase ISO it essentially means underexposure. So if you set a hig ISO and overexpose the net effect would be the same, that is the same number photons captured. On the other hand, signal processing may be affected by ISO setting but also how data are saved.

As the dynamic range of scanner/digital back is wider than the dynamic range of the scanner it seems reasonable to increase exposure, as increasing exposure will collect more photons, thus reducing noise. That is what we call Expose To The Right, ETTR. What increasing the ISO gives is less than clear. It depends on Better Light readout circuitry and how they save and tag data.

Best regards
Erik
Logged
Erik Kaffehr
 

markwilliam

  • Newbie
  • *
  • Offline Offline
  • Posts: 18

Thank you for elaborating, Erik. I very much appreciate it.
Logged

Fine_Art

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1172

For scanner type equipment ISO is probably the rate of scan. The BetterLight rep is probably saying you can get the job done faster.

For a normal sensor on a solid mount you are better off leaving the left of the histogram empty. On some things like sunsets the color can seem to wash out, you have to figure out how to correct it in raw conversion.
Logged
Pages: [1]   Go Up