Noise in 48 bit scans

[DaveO]

I have come across a problem when scanning to 48 but colour depth. This as expected produces large files so I ran LZW compression on some of then and instead of getting compressed they actually increased in size.

A query to the software developers forum (I use Picture Window Pro which can edit and compress 48 bit files) concluded that the higher bits probably contain nothing but noise and by the nature of the LZW algorithm this will cause the increase in size as it is adding compression data for each pixel of noise (as they are dissimilar to adjacent ones) rather than describing how a number of similar pixels are compressed. It could also be the film grain if that has been captured that is causing this.

If you change the bit depth to of one of these 48 bit files to 24 bits and then compress it, the compression works as expected and reduced the file size quite dramatically so this bares out the theory there is a lot of rubbish in the high order bits.

This begs a number of questions I hope people can help with.

First of all - how do I reduce any noise in the scans? I can't see any obvious noise when I look at them but something non linear is in the files.

Does my experience mean it's pointless to scan at 48 bit colour depth?

I scan at 3200 or 4800 and am scanning colour negatives so is this just too high and so the grain is causing the problem?

Thanks,

Dave

[LSI_Morales]

Dear Dave

There is always noise when working with scanners and the film grain is also there. There is much more noise when scanning negatives than slides.
A good explanation can be found here: http://tinyurl.com/6jxf6y

First of all - how do I reduce any noise in the scans? I can't see any obvious noise when I look at them but something non linear is in the files.

Noise can be reduced by using Multi-Sampling in SilverFast and grain can be reduced by using the GANE filter.

Does my experience mean it's pointless to scan at 48 bit colour depth?

Of course not, It is perfectly valid to use 48 bit color depth because you will simply have more information to recompose and correct the image later

I scan at 3200 or 4800 and am scanning colour negatives so is this just too high and so the grain is causing the problem?

It is hard to say, it might be the grain, it might be objects texture, it might be noise, it might be all combined....

Cheers

[DaveO]

OK thanks for the reply. I think the noise I am experiencing may be related to exposure problems I have outlined in a separate post (not yet approved for publication as I type this) in the NegaFix forum.

I have managed to get much less noise/grain by scanning the negative as a positive and converting it in my image editing software but I will not repeat myself here given I have gone into detail in my post on the NegaFix forum. Suffice to say scanning this way is giving a level of film grain I would expect from a 100 ASA film (very low) but scanning as normal using NegaFix is producing far too grainy a result.

Dave

[LSI_Muenier]

Dear Dave,

when your are scanning a negative as a positive and then invert it manually (after orange mask elimination), you are missing the NegaFix profile's curves that perform the mid tone correction, and you are missing NegaFix's color balancing.

Try to edit a NegaFix profile to a custom one by changing its curve manually and make is less steep, so that grain and noise will become less prominent.

Please also check your USM settings to see if too much sharpening is the problem.

Best regards
Martin

[DaveO]

Dear Dave,

when your are scanning a negative as a positive and then invert it manually (after orange mask elimination), you are missing the NegaFix profile's curves that perform the mid tone correction, and you are missing NegaFix's color balancing.

Yes I know but the software I use has a specific negative transformation that attempts to deal with the orange mask but of course it takes a generic view of the masks actual color and is not specific like NegaFix. As a result the scan can sometimes have a slight green cast but what it does not do is apply any mid tone correction so I suspect this is why they look very clean compared to what I am getting with the NegaFix settings.

Try to edit a NegaFix profile to a custom one by changing its curve manually and make is less steep, so that grain and noise will become less prominent.

I will try this if I can't get it to work with the canned profile for the type of film I am using.

Please also check your USM settings to see if too much sharpening is the problem.

This is set to none as I like to sharpen in my image editor.

Regards,

Dave