TriX wrote:
No, you are NOT accomplishing the same thing. When you change the ISO, you change the gain of the sensor/amplifier chain prior to the A/D which ALSO changes the D/R of the A/D and the S/N of the signal. When you change the exposure slider in PP, yo are simply multiplying the raw value by a constant.
Let me give you an example. Suppose you set the SS, Aperture and ISO such that you underexpose and by so doing, you limit the DR and amplitude resolution of the A/D (since you are not using all the bits of the digitizer). Suppose you have a 12 bit A/D with a max digital output of 4096 and you set the amplifier chain gain such that you just set the MSB (most significant bit). The smallest amplitude value that the A/D can resolve is 1/4096, but this accuracy and resolution is limited by several other factors including the sensor noise, the amplifier noise AND the uncertainty of the digitizer (at least 1/2 LSB), all of which are not necessarily linear in nature. Now instead, instead of setting the gain (ISO) correctly, you underexpose by 3 stops. Now you are only utilizing 9 bits of the digitizer resolution (512 discrete levels of amplitude resolution) or 1/512 max resolution. Not only that, but that 1/2 LSB is now a larger proportion of the total signal than if the full DR of the A/D was used. You have not only limited the resolution but decreased the S/N ratio. Now when that digital value is multiplied by 8 (3 bits) in PP by moving the exposure slider, the max value is again 4096, BUT you cannot recreate the resolution. You can’t magically recreate those missing 7 points between every real point - they are simply interpolated, but the actual resolution is lost and the S/N is still a larger proportion of the signal.
This is the old ISO invariant argument that you can shoot at base ISO, using only a portion of the A/D and “bring it up in post” with no changes in the final product or loss of resolution or S/N. You cannot. You pay big $ for a high resolution sensor and A/D and then limit it to a much lower resolution by underexposing and multiplying it in post. If you or anyone else really believes this, then why not use a much cheaper 8-bit A/D and work only with 8 bit data? Why not record in 8 bit audio instead of 14 bit and simply multiply it in the player? The designers of both cameras and audio systems know the reason and that is why we use high resolution A/Ds, and a signal chain to utilize the entire DR of the A/D (the ISO control on your camera). This is digitizing 101 - not rocket science.
No, you are NOT accomplishing the same thing. When... (
show quote)
What is LV = Light Value, please explain DR of the A/D,...ETC.
Totally T.M.I. for my 70 year old Brain to understand.