Counter-intuitive

I had settled on scanning my negatives at 2400 dpi because, starting at lower resolutions and working upwards, that was where the grain became noticeable. Any higher resolution, I reasoned, would only make the grains larger and not reveal any more detail. I was wrong.

There seems to be an 'interaction' between the scan resolution and the grain size when the two are close which actually produces a more pronounced appearance of graininess. At higher resolutions, though the grain is still visible, it is less distracting.  Here are some examples:

2400 dpi

3200 dpi

4800 dpi

6400 dpi

9600 dpi

Surprisingly, the least grainy in appearance is the last scan at 9600 dpi (which is a shame since it takes between 7 and 8 minutes per negative at this resolution).  So it looks like I'll just have to learn to be patient.        

Finding Gamma

Really, this is a note to remind me that, based on my empirical testing, for low values of gamma, the gamma value at the scanner and the gamma value in GIMP are additive, and sum to about 3.2. (This puts Zone 5 at a digital (linear) value of 127 out of 255.

There's probably a mathematical proof of this, but I'm not going to try to derive the result. Suffice it to say that I will now start using scans with a gamma of root 2 and then add 1.8 in GIMP. This is just a mnemonic device; root 2 is common and 1.8 is the typical display gamma.

Neither value has any substantive significance here. However, given the odd things the Epson software does with the tone curve and the top and the tow, I decided not ti use a gamma over 1.5 at the scanner. I also don't want to have GIMP to all the rescaling since I think this is best done at nearest the source in the hope that it's achieved in the AD converter, and not afterwards.

Acros 100 dynamic range test

Procedure

I metered a fairly flat (in terms of contrast) wood panel (the garage wall). I then exposed 8 frames each one stop greater than the last, starting from the metered exposure. This represents zones 5 through 12 (5,6,7,8,9,10,11, and 12). With a second roll I did the same, but stopping down, giving me zones 5 though -2 (5,4,3,2,1,0, -1 and -2). Development was normal (no push / pull), courtesy of Fuji Labs and Walmart.

Results

The results were interesting. Ansel Adams talks about Zones 0 and 10 as the points at which detail can no longer be resolved: but film evidently has improved. Acros 100 has a very broad dynamic range, with detail discernable in zones -2 and 12!

This composite shows the negatives from zones 10, 11 and 12 and 0, -1 and -2 and some others (1,2,8 and 9) thrown in for good measure. This was scanned such that the dynamic range on the 'input side' was from 40 to 248, values measured individual negatives as being the 'metered' values for -2 and 12 respectively.

      
Compare this to Ansel Adam's Fig 4-3 on page 50 of The Negative; in Adam's figure detail is resolved in 8 zones (from 1 to 9 inclusive). Here, detail can be discerned in zones -1 and 12 (there is some in -2 in the top right but it's a little hard to make out. The dynamic range of Fuji's Acros 100 is comfortably 13 stops. Looking at Zone 12 suggests that there is considerable scope for further over-exposure, perhaps another 2 stops. 

This compares rather favorably with two recent digital cameras, the Olympus OM-D and the Nikon D800. Their dynamic ranges are shown in the chart below taken from dpreview's measurements. The D800's range is only 11 stops.

Conclusion

This gives both the flexibility to get the exposure wrong and still recover the image and, if the negative is properly exposed, the ability to resolve enormous range of light levels. It also provides huge flexibility in creating the final image since there is so much information on the negative with which to work. 

The resolution at the over exposed end (11, 12) seems better than at the other end (-1 , -2), so my inclination is to over-expose by a stop (in other words rate the film at 50 rather than 100). 

P.S. Exactly 30 years ago, I listened to arguments about the merits of the CD over the vinyl LP.  Analog proponents (and I'm still one having never gotten rid of my Linn LP12)  asserted that a good turntable / tone arm could resolve greater dynamic range than could be digitally encoded in the CD; detail in the quiet passages and in the loud ones was superior on vinly with a good setup. This was sufficiently important to make up for the pops and scratches. Film seems analogous; the greater dynamic range over the dust and debris (no discernible grain if the format is large enough). And with some digital spotting, even the pops and scratches, dust and debris can be removed. Looks like analog has some life left in it yet...