Grain, tonality, format size and all that
Posted 06-01-2008 at 01:29 PM by keithwms
How much is grain per detail (and tonal smoothness) a function of format size?
First, the film grain per area is a constant of format size. This is important.
We can estimate grain increase per one stop increase in required film sensitivity as ~2x. In other words, if you shoot a film at ISO 200 then there will be ~2x more grain per detail than the same shot at ISO 100.
Obviously this is a very rough approximation and a rather pessimistic one at that. There are many different kinds of film grain, many different developers and developing procedures... and there is no reason to expect a linear relationship which will certainly fail at ISO extremes. Still, this ballpark estimate seems reasonable because detail per grain is ultimately nothing more than a signal to noise ratio, and if we shoot at ISO 200, we do so because the signal is 2x weaker than it would be if we could take the shot at ISO 100. The development process is basically an amplification process, so I think it's reasonable to say that 2x weaker signal corresponds to ~2x weaker overall signal to noise ratio. Again this is just a ballpark estimate! Of course it will fail at ISO extremes an for certain films and developers etc.
Alright... if you buy that a one stop increase in ISO means 2x as much grain per detail, then you see right away that a jump in format size which roughly quadruples the film area will give the same detail per grain as a one stop reduction in film sensitivity.
Let us list the format sizes and the approximate frame areas; I will define the letters A,B,C, etc. to denote roughly quadruple area format jumps:
35mm: 24x36 mm = 863 mm^2 = "A"
645: 56×41.5 mm = 2324 mm^2 = 2.7*A
67: 56×70 mm = 3920 mm^2 = 4.5*A = B
4x5: ~12903 mm^2 = 3.29*B = C
5x7: ~22580 mm^2
8x10: ~51612 mm^2 = 4*C
So here is my simple conclusion. The following exposures will yield roughly the same detail per grain:
Equivalent detail per grain:
35mm @ ISO 400
=6x7 @ ISO 800
=4x5 @ ISO 1600
=8x10 @ ISO 3200
35mm @ ISO 50
=6x7 @ ISO 100
=4x5 @ ISO 200
=8x10 @ ISO 400
Do you believe it?
Obviously this is very rough approximation based on many assumptions.
First, the film grain per area is a constant of format size. This is important.
We can estimate grain increase per one stop increase in required film sensitivity as ~2x. In other words, if you shoot a film at ISO 200 then there will be ~2x more grain per detail than the same shot at ISO 100.
Obviously this is a very rough approximation and a rather pessimistic one at that. There are many different kinds of film grain, many different developers and developing procedures... and there is no reason to expect a linear relationship which will certainly fail at ISO extremes. Still, this ballpark estimate seems reasonable because detail per grain is ultimately nothing more than a signal to noise ratio, and if we shoot at ISO 200, we do so because the signal is 2x weaker than it would be if we could take the shot at ISO 100. The development process is basically an amplification process, so I think it's reasonable to say that 2x weaker signal corresponds to ~2x weaker overall signal to noise ratio. Again this is just a ballpark estimate! Of course it will fail at ISO extremes an for certain films and developers etc.
Alright... if you buy that a one stop increase in ISO means 2x as much grain per detail, then you see right away that a jump in format size which roughly quadruples the film area will give the same detail per grain as a one stop reduction in film sensitivity.
Let us list the format sizes and the approximate frame areas; I will define the letters A,B,C, etc. to denote roughly quadruple area format jumps:
35mm: 24x36 mm = 863 mm^2 = "A"
645: 56×41.5 mm = 2324 mm^2 = 2.7*A
67: 56×70 mm = 3920 mm^2 = 4.5*A = B
4x5: ~12903 mm^2 = 3.29*B = C
5x7: ~22580 mm^2
8x10: ~51612 mm^2 = 4*C
So here is my simple conclusion. The following exposures will yield roughly the same detail per grain:
Equivalent detail per grain:
35mm @ ISO 400
=6x7 @ ISO 800
=4x5 @ ISO 1600
=8x10 @ ISO 3200
35mm @ ISO 50
=6x7 @ ISO 100
=4x5 @ ISO 200
=8x10 @ ISO 400
Do you believe it?
Obviously this is very rough approximation based on many assumptions.
Total Comments 1
Comments
 | P.S. upon further reflection, I do believe that the factor of 2x above is a bit too pessimistic, it might be closer to sqrt(2)=1.4x due to photon counting statistics. |
  |
Recent Blog Entries by keithwms
- Grain, tonality, format size and all that (06-01-2008)
- Converting RB-545i back for Fuji Quickloads (05-03-2008)
- Christopher Beane's "Flower" (05-02-2008)
