This roll of ColorPlus 400 has some negatives that look fine and some that are unusually noisy, like above. Almost all the negatives that contain sky are noisy. I scanned/converted all negs on the roll in the exact same way, with Epson v550, iSRD-enabled Silverfast, LR classic, NLP.
Help!
Look at the negative. Is it underexposed? If so, and NLP tries to pull the highlights back up from barely any information on the negative, that would create noise.
Dense parts of a negative have more grain which translates to grainy appearances of bright areas after the conversion. This can further be made worse due to digital noise as @ArnoG proposes.
It looks exposed correctly, so do others on the roll with the same problem. I can try to convert with SF or Epson Scan 2 instead of NLP and see if the same noise occurs. Also can try the Grain/Noise Reduction feature in SF and see if that helps.
This is good info! Did not know that. it makes total sense. Iâm going to destatic and clean the neg a little more, then try another scan, then convert with SF or Epson Scan 2 and see if those conversions create less noise
The negative looks like its histogram should be fairly narrow. This can also lead to issues as yours.
Starting from the jpeg you share, I got this manual conversion that makes me think that the ânoiseâ is rather the filmâs grainâŠ
âŠwhich is part of the original image and should not be eliminated imo.
Have you converted your image with an intermediary JPEG scan or a RAW (positive, 3 x 16 bit)?
happy to try. How do I do that?
Please check the guide (find it at the top of the forum page) and see how that is done with epson scan. It usually takes the following âmain ingrdientsâ
As for resolution: Whatever the DPI or PPI values are, try with less resolution to speed up your tests. You can always increase resolution once you found how to convert and set things. Needs another scanning session though.
Again: 400 iso film does have visible grain which is easy to see on copies from 135 film. Get rid of the grain means âget rid of detailâ. And weâre using film today because of its âcharacterâ , i.e. characteristic properties like grain, colour shifts and imbalances etc..
This is great! the negative i posted above was actually a test i did with epson scan, scanning a positive JPEG (to convert with NLP) to see if it was any better than my SF scan. did not try with a 16bit/channel TIFF file tho, will try that.
I hear you about the grain! We love some good grain. But I think what Iâm experiencing is digital noise/coloration by some step Iâm either doing incorrectly or not doing at all.
at the same time, i think underexposure/narrow histogram is also at play. so my tests may come up empty, but iâll see if i can close the gap even a little.
Epson 550 likely doesnât have sufficient resolution to see grain in much detail so it will be hard to distinguish between grain and digital noise. Regarding the not so wide histogram: This is either due to scanning or lack of dynamic range in the negative. For scanning, it is important to scan so that the right part of the histogram is close to, but not beyond, clipping. The right part of the histogram during scanning is a high signal level, while the left part is a low signal level. By âexposing to the rightâ the low signal levels are pulled up as high as possible above the noise floor, which helps to reduce noise. If the dynamic range of the negative is small, then keep in mind that this ânot so wideâ histogram will be spread out across the available dynamic range of the digital environment during inversion. In other words, the signals will be amplified and reduced to stretch the histogram out to set black and white points. Amplification of low signal levels introduces noise, so less amplification is better, hence ETTR. All this can introduce noise in some conversions, while it might be less in others on the same roll.
I agree with Digitizer (as usual!) - this is primarily so-called grain inherent to the film. I say âso-called grainâ because in fact there are no sharply defined particulates (such as silver halide particles) in developed colour negative film - what we are seeing is actually âdye-cloudsâ, which are soft-edged blobs. Dyes cluster around silver halides and then those silver particles are washed out during film development leaving the dye clouds. Typically these dye clouds range in size from 6 to 10 ” (but depending on the film could be smaller or larger) and become more visible the greater the enlargement. They are more visible on smooth light areas such as sky and skin than on darker areas, but they are they all over the image nonetheless. These dye clouds are in fact the structure of the developed image, so the more you do to smooth them, the softer the photo will be - i.e. you lose detail. Where there is no detail worth preserving itâs fine to do that, but otherwise often best left alone. The visibility of these dye clouds also depends a lot on the film ISO (lower ISO films tend to have smaller dye clouds than higher ones) and how it was developed. I doubt this has anything to do with the scanning software you are using. The illumination system of the scanner may accentuate or smooth the delineation of these features, but that is a hardware variable over which you have no control as long as you are using that particular scanner.
This is very good info!! Iâve never thought of scanning this way and I will do so going forward.
I think Iâm understanding it now. This is great info! Basically what Iâm hearing from all this is Iâm doing things right (I think?) but am limited by the tools in hand.
I did a few tests:
Scanned as positive (did not convert with scanner software)/Set 16 bit/channel and save as TIFF based on @Digitizer âs instructions above
converted in SF with digital ICE and noise reduction enabled. âexposed to the rightâ based on @ArnoG instructions above.
converted in Epson Scan 2
All had pretty much the same noise problem. (epson had the least digi noise, but it also had the worst looking conversion out of the bunch)
I will try to move forward with the original batch of SF scans into NLP conversions. going forward if i want the least amount of noise, itâs gonna have to be back to the basics: low speed film, processed at box speed, and scanned by a lab with a nice scanner.
Chemical photography hasnât changed that much since I used 400 ASA film ⊠120 format rolls though. 
I have âcamera copiedâ 35mm colour negatives to 24MP that many years ago I printed commercially in the darkroom, just as a test, Iâd kept some of those prints. Back then I had to print these in both A4 (12âx8) and A3 (16âx12â). I canât show it on here but this noise/grain was barely persceptible in the contemporary prints, much more so in the scans. My best explanation for this is that when processing colour negatives digitally there is a massive increase in contrast required as one of the stages and that inevitably shows up this âgrainâ. This is not true with B&W negatives, if anything contrast has to be reduced, but seemingly it is not present when printing chemically on to RA4 paper either, at least not unduly, though the paper itself must be inherently high contrast. Perhaps I should add that they were printed with a diffuse colour head on a Durst M605.
Open to other explanations?
It looks mostly like film grain, as opposed to digital noise. 400 ISO is FAST for film, and made that way by using coarse-grained silver halides in the light sensitive layers. UNDEREXPOSURE on the film narrows dynamic range and makes grain look more obvious. The same phenomenon happens with B&W film. I have a lot of 55+ year old negatives I made in high school on Tri-X exposed at 1280 and push-processed (developed) in Acufine. Some of them are quite grainy. Judicious use of Lightroom Classicâs noise reduction features can reduce the grain to more tolerable levels. Play around with sharpening, unsharp masking, and noise reduction settings until the image looks better at 100% magnification.
