Noise vs resolution when scanning

[whitestone]

Is there an explanation why I get more noise with higher resolution? Can

anything be done to mitigate the noise as scanning resolution increases? It's

bad enough that (in the crop below) it seems the higher resolution scan doesn't

actually add any information. (The 4000 ppi scan is shown at 100%.)<div></div>

[geoff_worboys]

You dont mention what you are scanning (reflective or transparency). What software are you using for the scan? Have you turned off all of its automation (sharpening etc) before comparing the results? Did you reduce your high-res scan down, or raise your low-res scan up for the comparison? [Your example looks a lot like the 4000ppi scan has been sharpened.]

 

With more information in the scan any noise can certainly be exaggerated by some image processing (eg: sharpening or resizing) performed by some image software.

[erik scanhancer]

Well, what do you expect? At higher resolutions you scan into the grain of your film. Your

1000ppi scan is obviously unsharp due to low resolution. Note that grain is not noise.

Grain is inherently part of the the look of film, while noise is an electronic defect of the

scanning process. Noise should be avoided by multi-sampling and grain should simply be

accepted as an important image forming factor. (Grain often has a sharpness-increasing

effect, because it acts as a random raster.)

 

Your image does seem to show actual noise, BTW. Your 1000ppi scan is probably an

internal downsample of the native 4000ppi signal, which also eliminates the noise pattern.

Many scanners scan at a native resolution and then downsample to the resolution you

specified in the driver.

[whitestone]

The scan was done on a Nikon Coolscan 5000 with the software that came with the scanner (Nikon Scan 4.0.2). The film is 640T slide film. There was no sharpening with the scanning software and no post processing in Photoshop, however, digital ICE, GEM, and ROC were on. The 4000 ppi scan is shown 100%, and the 1000 ppi scan was simply scaled up 4x in each direction in Photoshop (I don't know what this does in terms of resampling).

 

Erik - I did do a resample of the 4000 ppi scan down to 1000 ppi in Photoshop , but it still looked very noisy -- nothing like the actual 1000 ppi scan. Also, I've tried the multi-sampling feature, but it seems to have almost zero effect.

[ellis_vener_photography]

One excellent way to address this is to get better scanning software. Right now the best

available is SilverFast Ai6.5.5. and use the multiple exposure option. This makes two

exposures and combines them (And earlier version --6.5.4 -made either 2 or 4 but I never

saw any advantage with the 4 exposure option

 

I've been working with SilverFast v6.5.5 for about a week now with a Nikon Coolscan 5000

and the scans I'm making have the most fidelity to the original of any scans I've ever made

or had made --noise free and color accurate and clean detail up into the highlights and

down into the deep shadows. The scans are so good that I have started rescanning the

best

of what I've scanned before as well as a lot of my film archive material. I profiled with the

scanner using the SilverFast IT-8 target and am scanning at 16bpc into Joseph Holmes

Dcam 3 ((www.josephholmes.com/profiles.html) but his free EktaSpace 5 workspace works

very well too. The scans are saved as an approximately 112mb TIFFs. Each scan is

effectively customized to the dynamic range of each image -- meaning : I set the black

point at zero (0) and then drag in the white point to just a couple of points beyond the end

of the histogram. If I think a slide benefits , I'll adjust that white point setting in each of

the Red, Green , and Blue channels. If a slide needs it I'll use the mid point (gamma) slider

 

I also just discovered the Kodachrome setting in SilverFast Ai6.5.5 and this really helps

with

Koddachrome color balance. I use Digital ICE set to standard, even for Kodachromes

(someone on this forum suggested that and I thank you for the tip.)

 

 

Tell us more about your workflow and post screen shots of your settings.

[Ed_Ingold]

This "noise" is in the film, not the scanner. It is called "grain", or more specifically, "dye clouds", which constitute the image.

 

Choose your options more carefully. ICE is useful in nearly every case, reducing the effect of inevitable dust on the film, and has little effect on sharpness in the NORMAL setting. ICE cannot be used for silver-based images (which block infrared). GEM (grain reduction) reduces sharpness and is seldom needed IMO. If grain bothers you, use a slower film or enlarge less. ROC (color restoration) is intended for old slides which have taken a color cast due to aging. There's no reason to use it routinely, and it may affect your ability to get accurate colors.

[photosnafu]

Hi,

Note that some scanners (mine Epson 3170 does) produce much more noise after 1h beeing switched on than "cold" just after switch on.

Try it.

For 4000ppi scans, you could try NoiseNinja, and if you want to do it the best, you can make a dedicated noise profile of your scanner.

As I don't want to have a long workflow (or make only one scan each day to let the scanner get cold!), I select the minimum resolution at which all optical information of the original image is caught. Anyway, all scanned files need post-accentuation.

[stevecs]

Another thing to try is multi-sampling. I had the LS4000 and now the LS9000 and you do notice a difference when doing the 16x multi-sample (besides taking 16x as long to scan that is). :)

[upscan]

Yes, as Edward I. points out, the problem is not the scanner. One file in www.scanscience.com shows a scan with, and without the identical problem with Kodachrome to the image in your post. The images in the link were scanned at 4000 ppi in a Canon FS 4000. The scanner gets the blame for its higher fidelity to the original but as others point out the problem is not the scanner. I am not sure that Nikon has explained this issue as well as it might but in reality, the fact that the Nikon 9000 can show this effect is all to its credit. Dry scanning inevitably emphasizes grain due to light scattering. Running at lower resolution as your posted images show will blur the grain.

[juan_f._navarro_staicos]

I have a CoolScan 9000ED and have found the same problem. I can agree with the experts that there's nothing wrong with the scanner; is just the grain in the film. The question I would like to ask is how the scans can be improved? Is there a combination of resolution and output size that gives better results? I use NikoScan with the default features checked. ICE is also check when scanning not too recent negatives.

Thank you in advance