xtol dilutions

[Kirk Gittings]

Bruce interesting, my vintage TRI-X negatives all dev. in HC-110 have visible grain at 16x20, with a prosumer flatbed kind of mushy vague grain, with a drum or pro flatbed at sharp-grain almost identical to a traditional silver print at that size from the same negatives!?

[Bruce Watson]

Bruce interesting, my vintage TRI-X negatives all dev. in HC-110 have visible grain at 16x20, with a prosumer flatbed kind of mushy vague grain, with a drum or pro flatbed at sharp-grain almost identical to a traditional silver print at that size from the same negatives!?

<disclaimer>
I'm just a engineer / photographer who has done a lot of reading and too much subjective testing. I don't have access to the type of lab that would let me really investigate this stuff, so I haven't. Also, for the purposes of this discussion I'm just going to use "grain clump" when I really should say "grain clump and/or dye cloud." Anyone reading can extrapolate from B&W to color if they want to.
</disclaimer>

Well, this is why I said it's sort of nebulous. For one thing, when you scan film you can't actually image the grain. Grain clumps look to me to be almost fractal in nature. IOW, it would take a good many pixels to allow you to see the actual shape of the grain. For the sake of argument lets call it 10 pixels per grain clump. Now I read somewhere (don't remember where or even when) that film grain clumps range fairly widely in size, more than an order of magnitude. So for film with an RMS graininess of say 10 microns, grain clump sizes could be as small as say 2 microns and as big as say 18 microns. To get 10 pixels from a 2 micron grain clump you are going to need a spot size of less than 0.2 microns. Without doing the math I'd guess around 0.1 micron. That's about 30x smaller than the smallest aperture on the best commercially available drum scanner. IOW, that's a microscope.

All I'm saying is, even the best drum scanner isn't imaging actual film grain clumps. The laws of physics are too hard to violate.

So what they do instead is they create a perfectly square pixel (decidedly not-fractal in shape) that has a single color space value (say hue, saturation, and luminosity). You loose film grain and replace it with "digital grain" for lack of a better term.

On top of this, the distribution of film grain clumps in the emulsion is stochastic. All film scanners, CCD or PMT, are deterministic. That is, they create a virtual grid and create pixels by sampling the film through the holes of this grid. So if a film grain clump is centered in a grid hole you get a good idea of what it's HSL value is. If instead, which is the more normal case, the grain clump straddles two or more holes in the grid, the pixels still show what the scanner sees, but the resulting pixels contain a mixture of the HSL value for the grain clump and for the non-grain clump part of the film that the scanner sees through this particular hole. This is in part what causes scanned images to be somewhat soft and is the reason some people like to apply some "capture sharpening" to a fresh scan file.

I'm not even going to talk about lighting differences between flat bed scanners and drum scanners, and the effects thereof. Too painful.

But wait, there's more. You get the inverse effect happening when you print with an inkjet printer (or a lightjet, chromira, lambda, etc.). Here you are trying to create your pixel on paper using a number of ink droplets (laser or LED spots, etc.). The printer / driver does this using a dither algorithm which takes into account the HSL values for surrounding pixels, this done to improve gamut, smoothness, sharpness, whatever the particular dither pattern is optimized for. This is what I think of as "printer softness" and is one of the main reasons for people to apply "output sharpening" before printing.

The reason any of this works at all is that it takes a large number of film grain clumps to record actual image detail information. One can think of this detail information has being at a considerably lower resolution than the film grain. For example, one can think of capturing the majority of the image detail information with a scanner resolution of, say, around 2000 spi (actual number will vary widely from film to film, process to process, scene to scene, exposure to exposure, etc.) while one must go considerably higher to capture the actual film grain (say, 50,000 spi). Most people scan in the range of 4000 spi which lets them capture practically all of the image detail and coincidentally tends to be the maximum of many scanners.

All of this stuff probably adds together in some interesting ways. Visually I suspect that it's partly responsible for why inkjet prints can look smoother and less grainy than darkroom prints of the same enlargement.

Then there's that phrase "vintage Tri-X." Tri-X 30 years ago was a grainy film. Now, not so much. I've got access to a 20 x 16 Adams print, made with a 10x8 camera and Super-XX film. Visibly grainy and only a 2x enlargement.
So a lot depends on the film used and how it was exposed and processed.

And finally there's personal taste. What's too grainy for one person is another person's perfectly smooth.

In the end all we are left with is prints. And all we can do is work to improve our workflows so that we can make the best prints we can.