I was slow to respond. In fact, I did a very careful test, but the results were so ugly that I couldn’t believe they were correct. Such degraded performance could not be the fault of Kodak and its Tmax 400, whose poerformance in terms of resolution compared to the TriX was well established…
So I redid all my settings, then repeated my tests. And to make sure I didn’t make a mistake, I scanned a TriX and a Tmax with the same settings.
The strip on the left is the Trix with a 100% crop of a 36MPX. The middle strip is the Tmax treated in the same way. How to understand such horror? As far as I can make out, the tabular grains are flat and larger, but less distributed in the thickness of the gelatin. Above all, they are more homogeneous in size, with far fewer very small grains in the interstices. Is it the behavior of point light in this type of structure that fails so miserably?
To try and understand, I added a frosted glass 2.5cm above the top condenser. The result in the strip on the right. I find the difference quite spectacular. And the result is better (for my taste) than a scan with an Epson V600 at its maximum resolution of 6400 dpi, which is even blurrier.
It seems that tabular grains only like diffuse light. So it’s unlikely that you’ll be able to get good results with spot light on color films, many of which are made up of a thick sandwich of gelatin layers seasoned with a variety of tabular grains.
I also tried to process these results in post-production with my Diffusion or Sharpness module preset in Ansel/Darktable/. The result is somewhat similar to the addition of frosted glass, but not as good. This is logical, since the module’s algorithm is not quite equivalent to the optical phenomenon of diffusion.
So I hope this can help others avoid a dead end. And if by chance someone could have an explanation for such surprising results, I’d be happy.