Why is grain visible in scans?

Hi,

 

Mostly I scan Reala, NPZ, and Provia.

 

Recently I did a raw scan of a roll of Kodak PPN 160 (made in China)

at 4000 dpi and 64 bit depth (RGB and infrared) using Vuescan. This

resulted in 521MB files. I then processed the images to invert and

adjust for film type. In this process, I changed the scan to 1/2,

resulting in a file size of 130MB. In theory, this should smooth out

the noise.

 

When I import the 130MB file to PS 7, i can see film noise (grain) or

scanner noise in the shadows. On PS levels, there are no pixels in

bottom third of the display indicating underexposure, so the shadow

region is nearer to mid value than to min value. It is no where near

a true black value.

 

The image is quite good in the highlights and midtones, but has

significant noise in the the shadows (which aren't really very dark).

 

So, what am I seeing? Film noise (grain) or scanner noise? My

inclination is to think that it is film noise because I am no where

near the dmax for the scanner with such pale shadows. Is this correct

thinking? If so, then it appears that from a scanning perspective

that underexposure is the "big no no" to avoid at all costs.

 

It is confusing though because the pic looks great except in the

shadows and the contrast is fairly low overall. Even at 4000 dpi the

scanner resolves at best (high contrast) 80 lp/mm and probably in

this case is nearer to 40 lp/mm because of the low scene contrast,

but is the grain of the film really visible at this resolution?

 

Anyway, noodling on this but getting confused, so any insight would

be greatly appreciated.

 

Thanks.

 

Mike

First make sure you are scanning in RGB+infrared for a good reason. I've seen some unusual results come out of scanning in that mode.

 

Secondly, what scanner are you using? It sounds like you might have accidentally slid an exposure bar and forced the scanner to try to push the shadows way too far up the histogram. I would suggest experimenting and placing your shadows back where they belong before drawing any conclusions. This would explain your lighter and muddy shadows as well as the lowered contrast, because you've probably pushed some of the image off the scale or sacrificed contrast in scanning through other errs.

 

If you didn't move any controls, check to ensure you didn't have lock exposure on and check to ensure that your metering point is not off in the weeds somewhere.

 

I'd venture to say that under the conditions you describe you're seeing scanner noise, not grain.

On any of the newer scanners you should not have a problem with electronic noise when scanning print films (which will be in highlights and not shadows). The Dmax is simply not strong enough to push the CCDs hard enough unlike some slide and print films.

 

The fact your histogram is showing only a partial density range indicates you have a very underexposed neg (very, very bad), or simply did something incorrectly in your translation from RAW to neg.

 

In either case, the film noise in the shadows is typical of print films, and a big reason drum scanner operators hate color neg films.

 

This is a big reason I preach so hard about not underexposing print films and shooting 160speed films at EI 100. With proper exposure there will always be some noise in the shadows, but your histogram is then adjusted in such a way that the noise isn't visible and crammed right at the edge. If the neg is thin all the density information then get pushed into the center of the histogram where it is annoying and very visible.

 

If you extend your histogram out to get strong blacks you'll also increase the contrast of these regions and possible make the problem even worse - aka; you can't win.

It is possible you are not seeing grain, but an artifact of grain called grain aliasing. For a thorough discussion of this issue see

 

www.photoscientia.co.uk/Grain.htm

 

Of course, it doesn't really make any practical difference whether or not you are actually seeing the grain. There are various methods for dealing with this problem, and some of them are discussed at the above web site.

Mike, since you already have Vuescan, there is an easy way to tell if it's grain or CCD noise. Use the multisampling option and sample the film 16 times. This reduces CCD noise by a factor of 4 which should be enough to see a difference at 100% size. If the noise isn't reduced at all, then it isn't CCD noise and is most likely grain.

 

If you do perform that test, let us know the results.

Thanks for all the responses so far.

 

Carl, I reset Vuescan to the default values.

 

Jim, I mulitsampled at 16X but the noise is still there. So, it is not scanner noise.

 

Leonard, I read the aliasing article but I don't think that it applies here. The aliasing effect comes from a difference in the sampling rate and the pattern of the grain. I think I should see grain throughout the image, not just in the shadows if it is an aliasing thing. I think the scanner resoltion at 4000 dpi is at least 2X the resolution of this film in this low contrast situation, so aliasing isn't likely.

 

Scott, the background was white and the clothes were bright white, so the histogram should be heavily tilted to the right. One thought I had, was that the shadow was not very dark because of light reflected from the clothes. This would make sense except that the faces were underexposed by 1.5 f-stops also. Duhh!! So, you are right, it is severely underexposed. And I tried moving the blackpoint up but it just built too much contrast, accentuating the effect, as you said it would.

 

So, the film is underexposed and so the grain just pops out at you. But shouldn't the grain be in the highlights not in the shadows then? The picture is almost all highlight, white on white, and none of it blew out, and none of it has any grain visible.

The only grain visible is in the shadows.

 

Mike

I have the same problem scanning color MF negs. Two things I'm trying

to help the situation: 1) don't scan larger negs at 4000 dpi, you're

scanning the grain by then. I've gotten much better results at 3200 2)

Don't turn off the scanners' sharpening, but instead set it to the lowest

possible sharpening setting...some scanners default to sharpening if none

is set.

 

I once sent a bunch of RZ 6X7 ISO 160 negs for drum scans and they

came back so grainy they were unusable. The Art Director was beside

himself. I made a scan on a Epson 1240 flat bed using the film scan

accessory and got way better (read no grain) scans. Same thing happens

when you make a conventional print, the enlargers' diffused light softens

the effect of the grain.

Oh, I forgot...did you try making a print? Sometimes what shows on the

monitor doesn't in the print (especially in the darks).

You said the scanner resolution is twice the resolution of the film. I doubt that. 4000 ppi is 157 pixels per mm. The maximal theoretical resolution in the usual sense (as in film resolution) obtainable from that is half it or about 78 lp/mm. But no scanner is going to actually deliver its theoretical maxium. Color films often have resolutions about 50 lp/mm. So the two are roughly comparable.

 

But you may very well be right that you are seeing grain and not grain aliasing with your 4000 ppi scanner.

 

See

 

www.normankoren.com/Tutorials/MTF8.html

 

for further discussion of grain and grain aliasing.

I have experienced significant noise in dark/shadow areas. I do not scan myself (4 x 5s) but have them done by a guy with a pro UMAX flatbed. I silently question his operating skills. I have read the specs on his machine and others. Unfortunately none of these scanners capture the density ranges they claim. Most of my B & W negs have a net density of 1.3 and his scanner boasts 2.4 and deep shadow detail is lost. By comparison I have seen drum scanners do better. Since you scan yourself, you have much control, and I would suggest 2 scans each film (if its a problematic film), one for shadow detail to complement one for overall or favoring mids and highlights. When a scanner has to "bite" through blacks in a transparency I have seen noise result many times.

I re-read your question and also wonder why the low end of your histogram is as you describe. Do you let the machine scan using defaults or do you select the black point yourself? just another thought - good luck

Hi,

 

I really appreciate the comments so far.

 

I'd like to draw out a few more points.

 

The highest resolution film is probably something like Kodak Techpan. I understand it can resolve 220 lp/mm under ideal conditions. This means it can see a gap under very high contrast circumstances of about 45 microns. I also understand that when they measure rms values, they use a photo cell that is also 45 microns. Maybe this is no coincidence, maybe the sensor is sized to look at the gap of the highest resolution film.

 

If this is true then grain is always significantly smaller than the reolution. On another photo.net thread it was suggested that grain was 3 microns and grain clumps (3-6 grains) were 6-9 microns, roughtly 1/0th the size of the gap between line pairs.

 

If this is true, however, then it should be very difficult to ever see grain clumps and yet we see something. Maybe the only time grain becomes visible is due to aliasing. Is this possible? Aliasing is probably progressive not appearing just suddenly.

 

So, just trying to understand what's happening. Scott Eaton, Norman Koren and Struan Gray can you guys comment?

 

Thanks.

 

Mike

The theroy on grain aliasing is an interesting one. Saying that, because of the irregularly spaced grain and other snot on film, light passing through it may strick the sensor at an odd location. Enough grain lined up in the right way and it will not line up with the sensor cells, causing the "graininess" of some scans as the sensor tries to make sense of the strangely aligned light rays. There are some thorough discussions on it, but nobody really seems to know the right or best answer.