Digital BW, The Print

> Rather than comparing grain size to resolution, I think an interesting
> comparison would be between the film's resolving power and the digital
> resolution.  The data sheet for Fuji Velvia, for instance, says it can
> resolve 80 lines/mm at a chart contrast of 1.6:1.  The
> measurement for this
> is described in ISO 6328:2000.  Does anyone have a copy of this?  What I
> don't know is if lines/mm means lp/mm.

Hi Thomas,

lp/mm means line PAIRS per mm.  One black line, next to one white line.  If
you have two lines of same color next to each other, then you can't see
them.  Reason for solid black next to solid white is because it's %100
contrast.

Resolution of digital imaging systems is hard to measure...since they have a
regular structure for their arrays.  The worst case is diagonally...which is
typically about 1.4x worse than on axis...and most digital imaging systems
list only the BEST they can do, not the worst!

> If you assume it means exactly what it says, "lines/mm," then that
> translates to 4064 pixels per inch which would mean an image of
> 3840 x 5760
> (or 22.1 megapixels) is equivalent.  This is closer to the number
> thrown out
> by Gordon Brown.

80 line pair/mm is actually 160 lines/mmm, which is what you used to do your
arithmetic with apparently...since 160 x 24mm = 3840.

The big issue to remember when comparing digital imaging sensors to film is
that digital imaging sensors are regular patterns, and film is not.  This is
very important when trying to compare resolutions.  Film has a LOT more
information than just comparing squared segments (as you are doing when
calculating digital equivalent based on lp/mm)...remember the diagonal issue
I mentioned above.

You are also confused by pixels vs sensors.  Digital imaging sensors use
FOUR sensor cells to make ONE color pixel...so if you want to get the same
amount of color information, you need four times the amount of sensors!  So,
4 x 22.1 is reasonably close to my claim of 100M pixels.

There is also another issue...called sampling.  In the digital world, you
need to sample at slightly more than 2x the smallest thing you want to
RELIABLY detect.  So, in order to RELIABLY detect 160lines/mm (only on axis,
off axis is 1.4x), you need to sample at 320 lines/mm...in both axes (X and
Y), therefore giving you 4x your initial estimation of 22.1 M pixels, and
this doesn't take into account the Bayer pattern (four sensor cells for one
color pixels worth of information).  Again, showing that more like 100M is
in reality what is needed to equivalent top 35mm film.

So, even the 80lp/mm shown for Velvia, clearly shows that 22.1M sensors is
clearly not enough to give a reliable 80 lp/mm with digital.  Others will
argue that anything over 40 lp/mm is wasted...and that very much depends on
what your shooting techniques are.

If you want any of this explained any further, I'd be more than happy to.

Regards,

Austin