philipus Posted March 24, 2012 Share #1 Posted March 24, 2012 Advertisement (gone after registration) I'm curious about how many other people use Photoshop's or ACR's noise reduction capabilities on scanned images. Or is it perhaps advisable not to use this on scanned colour images because it degrades them somehow? Cheers and thanks for sharing Philip Link to post Share on other sites More sharing options...
Advertisement Posted March 24, 2012 Posted March 24, 2012 Hi philipus, Take a look here Post-processing q: removing colour noise from scanned image. I'm sure you'll find what you were looking for!
jaapv Posted March 24, 2012 Share #2 Posted March 24, 2012 It does degrade but you have no choice in reducing grain aliasing. Link to post Share on other sites More sharing options...
dave_d Posted March 24, 2012 Share #3 Posted March 24, 2012 I use the non destructive noise reduction in Lightroom so I can undo it if necessary. Works good for me. Link to post Share on other sites More sharing options...
tobey bilek Posted March 25, 2012 Share #4 Posted March 25, 2012 There was a program called Grain Surgery which suited FILM. Kodak has one called GEM. The others mentioned are for digital images and the requirements are different. Link to post Share on other sites More sharing options...
StS Posted March 25, 2012 Share #5 Posted March 25, 2012 Are you using a film scanner? There should be the option to scan the image several times and average over the single scans. This should be much more effective compared to post-processing the noise out. However, the scans from my scanner appear to be noise-free from the first scan. Stefan Link to post Share on other sites More sharing options...
philipus Posted March 25, 2012 Author Share #6 Posted March 25, 2012 Thanks for the software tips, Tobey. I'll look into them. Yes, I am using a Nikon Coolscan V. Perhaps I should have been clearer in my initial post. What I'm describing as "colour noise" is this: Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Not a particularly appealing photo but it is just an example to show what I call colour noise (clearly the reduce noise filter degrades but I wasn't at all careful in applying it here). It is a bit like colour bleeding almost and looks like the noise in some digital camera files. Stefan, which scanner do you use? This is a crop of the same area after having put "Number of passes" to 3. I see very little difference to the above scan. I previously tried scanning as DNG and reducing noise in ACR but that's not an option any longer because I use the Color Perfect plugin (and ACR's noise reduction produced results similar to CS5's filter). cheers philip Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! Not a particularly appealing photo but it is just an example to show what I call colour noise (clearly the reduce noise filter degrades but I wasn't at all careful in applying it here). It is a bit like colour bleeding almost and looks like the noise in some digital camera files. Stefan, which scanner do you use? This is a crop of the same area after having put "Number of passes" to 3. I see very little difference to the above scan. I previously tried scanning as DNG and reducing noise in ACR but that's not an option any longer because I use the Color Perfect plugin (and ACR's noise reduction produced results similar to CS5's filter). cheers philip ' data-webShareUrl='https://www.l-camera-forum.com/topic/175572-post-processing-q-removing-colour-noise-from-scanned-image/?do=findComment&comment=1963892'>More sharing options...
StS Posted March 25, 2012 Share #7 Posted March 25, 2012 Advertisement (gone after registration) I'm using a Quato 5000 (Plustek 7500 with Mac driver), controlled by Silverfast 6.6 Some members reported a similar issue in the lengthy Plustek 7600 thread. If I remember correctly, a member resolved this by using vuescan (no experience here from my side). Stefan Link to post Share on other sites More sharing options...
hoppyman Posted March 25, 2012 Share #8 Posted March 25, 2012 Hi Philip. Are you scanning with the original Nikon software or with Vuescan? I aim to just capture a flat scan with no clipping of white or black points initially and all editing in Photoshop/Lightroom. If you are still shooting your favourite Portra the results are different of course to scanning transparencies or BW film but a scan after all is just another digital file and there's no reason why the same tools won't work. In Lightroom, take a look at the masking options when you do your sharpening. Smooth sky areas are notorious for showing grain and noise (from scanned film). You can readily apply no sharpening or some noise reduction to just parts of the image. Remember that the sharpening and noise reduction functions are inter-related. I'll take a look and send you some specific suggestions. Are you running LR3 still I guess? Link to post Share on other sites More sharing options...
plasticman Posted March 31, 2012 Share #9 Posted March 31, 2012 Coming from a digital background, I don't see any need to treat scanned noise any differently than sensor noise. Although film doesn't suffer from the blotchiness of low-frequency chroma noise (the purple-yellow patches of smearing that are characteristic of high-digital ISO), the fine-grained noise you're talking about here reminds me of what I get out of the M8 when I underexpose at ISO320 then push up the exposure in post. Ps plug-ins like Noise Ninja allow for extremely customizable fine-tuning of how the noise reduction is performed, and I find it can totally clean the chroma noise without affecting the luminance detail, at all (I like grain incidentally). Another tactic is to use multiple samples to reduce random noise in the captured scan - but I found this lengthened the scan times inordinately, and gave no better results compared to post-processing with NN (I also worried about wearing-out my scanner unnecessarily). Link to post Share on other sites More sharing options...
adan Posted March 31, 2012 Share #10 Posted March 31, 2012 Philip - out of curiosity, have you tried using slide film (E6) for your color work instead of negatives? For whatever technical reason(s), slide films usually scan with less color noise, and even less grain overall, than comparable-speed color negative films. I spent 5 years trying to get skies shot on color neg to scan as simply blue, rather than red/green/purple porridge - and never found a way to get even close to slide films. Slides do have their own problems - DR and shadow noise, as well as more complex and less available processing - but net/net I preferred them. Scan below is from 35mm Provia 100 with Nikon 5000ED, 4000 ppi. Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/175572-post-processing-q-removing-colour-noise-from-scanned-image/?do=findComment&comment=1969581'>More sharing options...
philipus Posted April 1, 2012 Author Share #11 Posted April 1, 2012 Thanks for sharing your experiences, I appreciate it a lot. I realise now I forgot to attach the image: Below is the 3-pass version of Portra 160. Shows little difference to the one-pass scan. Hi Philip. Are you scanning with the original Nikon software or with Vuescan? I aim to just capture a flat scan with no clipping of white or black points initially and all editing in Photoshop/Lightroom. If you are still shooting your favourite Portra the results are different of course to scanning transparencies or BW film but a scan after all is just another digital file and there's no reason why the same tools won't work. In Lightroom, take a look at the masking options when you do your sharpening. Smooth sky areas are notorious for showing grain and noise (from scanned film). You can readily apply no sharpening or some noise reduction to just parts of the image. Remember that the sharpening and noise reduction functions are inter-related. I'll take a look and send you some specific suggestions. Are you running LR3 still I guess? Hi Geoff, thanks very much for the tips, esp about selective masking for the adjustments. That's a great idea which I will explore. I'm still mainly using Portra, for instance all my Oz colour photos are taken with that film. I scan in Vuescan with a straight linear graph, curve low 0.25, curve high 0.75 etc. So I get a flat looking scan. I actually don't have Lightroom, though I've considered it now since the price for LR4 appears to have dropped. I do all my editing in CS5 having first run the photos through ColorPerfect which I like a lot. Philip - out of curiosity, have you tried using slide film (E6) for your color work instead of negatives? This is really interesting, Andy. I did a quick scan below and the grain difference really surprises me even though it is grainier than your example (Sensia 100, also at 4000). I used Velvia for 20 years until I went digital. Then, when I moved to Leica and went back to film, I switched to Portra because I found slides a bit difficult to scan. I realise this was because I wasn't very good at estimating exposure with my M3 but since I now use an M6 I will definitely give slide film a go again. So thanks for pointing me in the E6 direction! Ps plug-ins like Noise Ninja allow for extremely customizable fine-tuning of how the noise reduction is performed, and I find it can totally clean the chroma noise without affecting the luminance detail, at all (I like grain incidentally). Another tactic is to use multiple samples to reduce random noise in the captured scan - but I found this lengthened the scan times inordinately, and gave no better results compared to post-processing with NN (I also worried about wearing-out my scanner unnecessarily). Thanks for the Noise Ninja tip, I will look into that. I wasn't aware it worked for film also but that's good news. I also feel that multi-passes (not talking 5th Element here ) on my Coolscan V do not give the quality increase they might on better scanners. Cheers Philip Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/175572-post-processing-q-removing-colour-noise-from-scanned-image/?do=findComment&comment=1969953'>More sharing options...
tgray Posted April 1, 2012 Share #12 Posted April 1, 2012 This isn't the answer you want to hear, but I just leave grain alone with my film scans. Then again, it doesn't really bother me. Some times I even seek it out - it's what my stash of Ektachrome 400X is for. One thing to keep in mind, kind of pointed out already, is that in color film, the color 'noise' is actually dye clouds that form the image. It has a different character than the blotchy chroma noise from digital cameras, and is more like the luminosity noise you find on them. Multiscanning really doesn't help that much, particularly in sky areas, because the color 'noise' IS the image. As I understand it, multiscanning is useful for reducing noise in deep shadows in slide scans where the scanner sensor introduces noise, and not in reducing actual grain. So, in order to reduce the color noise you have two options in my mind. One, blur the chroma channels, which can be done by a quick trip to Lab color space and doing a gaussian blur or noise reduction step on the a and b channels. This is probably very similar to what LR does with chroma noise reduction. However, this will leave you with only luminance grain in your image - it will still be just as grainy, but more monochrome now. The second option is to use noise reduction on the luminosity channel too - again, like what LR does. I personally don't like the way that looks because it gives you oddly smooth areas in an image with a sudden increase in detail (and noise) when a feature in the image pops up. I realize the tools in LR are pretty sophisticated with respect to this, and they can do a lot, but I prefer them used with a more subtle touch. I can see using the color noise reduction feature to knock some of the chroma out of the grain, but if you print big, you are still going to have grainy unsmooth skies. And I just don't like what film looks like when you crank up the luminance noise reduction. Other than shooting lower grained films (Ektar, slide film, etc.) if you want to stick with film, you best bet in reducing grain is probably to move up to medium format. If you need a lower grained 400 speed film, Fuji Provia 400X is pretty amazing and scans easily on a Coolscan V. Here's a link to a 4000 dpi scan on a Coolscan V from 400X: Post #1122073 You can find a link to the full picture as well in that post. Then again, keep in mind with all the above that I really only print to 8x10 or 11x14. I don't have the desire to work with big prints, nor do I want to buy expensive frames, nor do I have the wall space. I like smaller prints. Link to post Share on other sites More sharing options...
adan Posted April 1, 2012 Share #13 Posted April 1, 2012 One, blur the chroma channels, which can be done by a quick trip to Lab color space and doing a gaussian blur or noise reduction step on the a and b channels. This is probably very similar to what LR does with chroma noise reduction. However, this will leave you with only luminance grain in your image - it will still be just as grainy, but more monochrome now. Right. You can even bypass the trip to L*a*b color mode by simple running any type of blur filter on the RGB image, and then "fading" the blur to limit it only to color and not luminance. E.G., in any recent version of Photoshop: Filter > Noise > Median > 3 pixels, then Edit > Fade Median filter > Color > 100%. Gaussian Blur can be substituted for Noise/Median filter, as tgray says. Link to post Share on other sites More sharing options...
tgray Posted April 1, 2012 Share #14 Posted April 1, 2012 Good point Andy. There are a couple ways of doing the same/similar things. The way you mentioned is a good way to go about it if you don't need Lab for other things. Link to post Share on other sites More sharing options...
CalArts 99 Posted April 2, 2012 Share #15 Posted April 2, 2012 Aliasing in CCD scanners is a fairly well documented phenomenon. As we know, this is the aliasing between the randomized organic film grain and the structured grid of pixels on the scanner. Here's a general primer. It's a bit dated but is fairly explanatory Grain aliasing I realize some people will yell at me for this, but if somebody needs a big file for a big print output, color negative film tends to be better drum scanned at a very high sample rate and then down sampled. Color negatives seem to need a very high sampling rate to help correct the visual noise issue. The effect is virtually eliminated by using a 8000 spi (samples per inch) and then downsampled. The high resolution will accurately describe the grain/dye clouds, and avoid the aliasing. After downsampling it will appear to have a normal amount of grain compared to the amplified “graininess” of a direct scan at 4000 (or 2000.) Unfortunately the grain aliasing appears to be most apparent at 4000 with 35mm film which happens to be the max optical scan sampling rate of the Nikon Coolscan. Multi-sampling, exposure, noise reduction filters , sharpening filters, and the PS despeckling filter can help, but I don't know of any real answer using the CCD Coolscan. I think it's pretty much just the way it is with small color negative film. Sometimes try to 'fix' it is like a dog chasing its tail. Apparently color reversal film doesn't have the same sort of visual issues because the grain tends to be monochromatic. You will primarily have 'black dots' rather than heavy color dots as noise. Although since scanner noise is stronger in the blue channel (blue filter allows more light to come through), the grain is much more apparent in places like blue sky with transparency film. But overall, I think it makes transparency scanning a better choice when using consumer CCD scanners at their highest sampling rate such as the 4000 found on the Coolscan. Here is a scan from my old Coolscan V that I shot originally with reversal film. The 100% crop is from the scan. The original size was for a publication and so the scan wasn't huge (119 MB) but it was scanned at 4000 ppi. At 240 dpi output it would make a 15x22 inch print. Anyway, it still pretty much shows the difference in grain aliasing with transparency film. And you can see the blue coming through in the scan crop. I personally don't scan on these scanners when I'm using color negative film for large prints. But transparency film is not nearly as much of a problem. The other advantage with transparency is that you can calibrate the scanner for it by using an IT8 file. I think one has to think about final output, cost, and the pros and cons of using each film type (negative film has certain advantages as does reversal film, and especially when it comes to exposing the film in camera.) But I personally don't think there's 'one size fits all' for every kind of output. Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/175572-post-processing-q-removing-colour-noise-from-scanned-image/?do=findComment&comment=1970535'>More sharing options...
CalArts 99 Posted April 2, 2012 Share #16 Posted April 2, 2012 Here are some more transparency scans off the Coolscan V. As I mentioned I've had pretty decent results with transparency film at the 4000 dpi level with the CCD of the Coolscan, but color negative film is tougher I think. And all the noise reduction filters eventually eat into the image and then you try to correct that, and it seems to go around in somewhat of a vicious circle. So I personally try to stick to transparency film and drum scan any 35mm color negative film. But again, I believe the final output of the work should really be dictating the entire process. (fwiw, small white thing in the man's hand is a piece of crack cocaine.) Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/175572-post-processing-q-removing-colour-noise-from-scanned-image/?do=findComment&comment=1970723'>More sharing options...
philipus Posted April 3, 2012 Author Share #17 Posted April 3, 2012 CalArts, thanks very much for these examples of scanned transparencies. The 100% crop of the lady with the yellow hat is really quite impressive. Btw the last photo looks like an undercover agent - I hope his cover isn't blown Re LAB - I usually work a lot in LAB since I found Margulis's truly excellent "Photoshop LAB color". I tried the blurring/noise reduction on the a and b channels but on my monitor at least the results are very similar to running the ordinary Reduce Noise filter on a photo in RGB. From a practical point of view it isn't very easy to apply gaussian blur or the Reduce Noise filter because the a and b channels don't show much detail in the 100% window. I agree with CalArts that it becomes a vicious circle. When artefacts appear - to me, they look a bit as if very tiny parts of the image have been ripped off from the overall "smoothness" of the photo - one has to work around that. Or better, start over... Edit: Re IT8 calibration, does anyone of you have experience with the targets offered by Wolf Faust (http://www.targets.coloraid.de)? The N3 target seems interesting from a slide scanner point of view. cheerio p Link to post Share on other sites More sharing options...
cbretteville Posted April 3, 2012 Share #18 Posted April 3, 2012 Finding a lab that does E6 is getting increasingly difficult, but as long as you're not in a hurry and can use mailers I guess you'll be fine with slide film. I have bought it8 targets from Wolf Faust and made profiles for my Minolta scanners. His service is quick and the price is reasonable. Carl Link to post Share on other sites More sharing options...
tgray Posted April 3, 2012 Share #19 Posted April 3, 2012 I tried the blurring/noise reduction on the a and b channels but on my monitor at least the results are very similar to running the ordinary Reduce Noise filter on a photo in RGB. It's quite possibly the Reduce Noise filter does something very similar to blurring the a/b channels in Lab. I don't know. Personally I find that if you do need to do a bit of chroma noise reduction, a radius of 2-3 pixels is general enough. Edit:Re IT8 calibration, does anyone of you have experience with the targets offered by Wolf Faust (Affordable IT 8.7 (ISO 12641) Scanner Color Calibration Targets)? The N3 target seems interesting from a slide scanner point of view. I have one and it works pretty well. I also have a similar profile for Kodachrome, from Silverfast, which I've used a bit more. Again, it does what it's supposed to. Link to post Share on other sites More sharing options...
CalArts 99 Posted April 3, 2012 Share #20 Posted April 3, 2012 CalArts, thanks very much for these examples of scanned transparencies. The 100% crop of the lady with the yellow hat is really quite impressive. Btw the last photo looks like an undercover agent - I hope his cover isn't blown Philipus, although I find the Nikon Coolscan models to be very good with scanning transparency film, I think that negative film is probably their weak spot. Part of what makes the Coolscans good overall are the high quality Nikkor ED optics (the lenses are excellent and that makes a big difference.) In addition their density range (with high Dmax) is also good. And their spi ratings are true and not interpolated like the Epson flatbed's advertised spi. But the Nikons do use LED as their light source. That has the advantage of 'instant-on' and good contrast direct from a scan (for scanning without post processing.) But it is a harsh light. The light accentuates the grain and any film defects (scratches, etc..) Because of this I don't find them to be the ideal scanner for negative film (and especailly with B+W film.) It's also why I mentioned that I prefer to drum scan color neg film instead of using the Coolscan. However, the choice depends on output. And if you aren't making large prints, then it's fine. Plus you can always try scanning at a lower spi to help reduce visual grain if you don't need the bigger file size. Transparency film has its issues, too. You get the pepper grain (black dots) with transparency film. But overall it's not nearly as obvious with the LED lighting as is the visual color grittiness of negative film, and it ends up looking pretty smooth overall yet still very sharp. Of course there are the considerations such as film costs, limited processing facilities for E6, and less latitude with in-camera exposure. But it does scan much better when using the LED illuminated Coolscans. I think all scanners have their good points and bad points with different materials. But these days there is very little choice in affordable hardware. Diffusing the LED light source can reduce the visual grain issue with color negative film quite a bit. There was (still is?) a Dutch company that made a piece of diffusing material that sits between the film and the light source. It was made for the Minolta Dimage scanners but people were adapting it to use in the Coolscan with the MA-21 slide adapter (with neg film just cut the frame and mount it in a Gepe plastic slide mount.) But the downside is it can sometimes affect color with the Coolscans. Diffusing the light of the Coolscans isn't really an optimal solution, either. btw, he's not an undercover cop but an actor in a film production on the streets of Los Angeles. Those images were part of a larger project of mine (5 years in the making) that was funded by the Getty Museum and the J. Paul Getty Trust, and the Andy Warhol Foundation for the Visual Arts. It was primarily about Los Angeles attempting to find its 'cultural center' with the building of the Frank Gehry designed concert hall in downtown. It consisted of several 'chapters' and that particular chapter was about how Hollywood produces its make-believe constructed violence on top of the city's own backdrop of a very real physical and economic violence. There was a lot more to it than that, but I'll spare you for now (fyi, the entire project was done with one camera and one lens: M6 with 50mm Summicron and with a pocket full of fresh film each day. ) Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.