Very few image processing algorithms require the whole image. Most work locally (blurring and sharpening) or on a pixel level (curves, color adjustments, etc.). Global algorithms like histogram can be programmed to take individual images (tiles) as input. So theoretically this is possible and must have been done before in other areas like cartography and microscopy. I don't think there is a practical need to adopt these methods in photography or to optimise software for working with very large images.
"Scanning" negatives with digital camera on a light table
-
H
- Thread starter J N
- Start date
Recent Classifieds
-
For Sale Nagaoka 4X5 View Camera w/ Lens Boards
- Started by Beevo
-
For Sale FS: Fuji GF670 - Silver, recently serviced.
- Started by 0x001688936CA08
-
Found Mamiya RZ67 120 film back
- Started by blee1996
-
For Sale Kodak 125PX Plus-X 35mm Film Lot (And More)
- Started by chuckroast
-
For Sale Large format Kodak 1443 Aerochrome color infrared film
- Started by MCB18
Forum statistics
Very few image processing algorithms require the whole image. Most work locally (blurring and sharpening) or on a pixel level (curves, color adjustments, etc.). Global algorithms like histogram can be programmed to take individual images (tiles) as input. So theoretically this is possible and must have been done before in other areas like cartography and microscopy. I don't think there is a practical need to adopt these methods in photography or to optimise software for working with very large images.
Thank you. In a few words, what was the conclusion? There are two schools of thought here. Some propose the use of a source with high CRI, while the others suggest illumination with three narrow bands centered around the peaks of Status-M (or Status-A) spectral curves.
My personal conclusion was, and remains: it doesn't matter all that much in practice. I'd expect a slight advantage for the discrete R, G and B illumination, provided this is properly matched with the transmission spectra of the dyes in the film as well as the sensor in the camera. I imagine the high CRI approach is easier and therefore less likely to be f** up because there's very little you can do wrong with it. Whatever little crosstalk you'll end up, you'll just have to live with, but in the grand scheme of things, it's no biggie anyway.
Micro 43 is just fine (20MP). 20MP scans of 35mm are more than adequate. I use my a7iii or EM1-II depending which is handy. I've been known to use the Oly + 60/2.8 Macro handheld just to get sample exposures to work with. If you have MFT with pixel shift high res then you can do that with a tripod....
I have a Meike MF 85/2.8 that I use for this purpose with the Sony, although I do have a Canon MP-E. Almost all of the glass I use with my Sony is EF (with a Metabones adapter), as I still have EOS equipment, including an EOS-3 that I use for most 35mm film.
I do all of this with raw files, certainly not JPEG. Again, if 10-12 bits isn't enough then you can always go with bracket and HDR.
As for illumination, as long as it's uniform and high CRI, you should be fine. You can always use your illumination source to light a color checker and photograph that to see if there are any obvious issues to resolve.
Now. In regard to color ... If you have good color negatives and a flatbed, use the flatbed software, as there is no particular reason to "camera scan" if your negatives are properly exposed and the film is in good shape and the flatbed process gives you good results.
If you are like me and shooting decade+ expired film and often cross-processing it, your scanner tuned to "normal" color processing is not going to work well and if you are in the mood, you can get better controlled results with "camera scanning."
You may have to make a surprising number of adjustments to get color looking the way you want. Your scanner's software is doing that ... you'll need to also. Color balance and mono curves alone probably won't do the trick. You should consider fiddling with just about every color control you have (in Lightroom or whatever you're using). Although using color curves can get you most or all of the adjustments you need, they can be very fiddly, and using the grading wheels is simpler and also effective. You may also want to use the color band controls.
Getting a photo of a Color Checker on each roll could be very helpful although the color corrections are still likely to be tricky.
The somewhat obscure ACR profile generator has the ability to generate a profile based on a photograph of a Color Checker, but it won't really fit into this workflow and it's definitely not set up to work with inverted color. (!) However, always something to bear in mind.
Anyway, this process is IMO most helpful for truly problematic negatives, especially very dense ones. For example, you can coax a usable or at least interesting image out of color film that is so fogged you can't scan it at all. I've found old rolls of undeveloped film that had after processing had profound levels of base fog, but recognizable and valuable images. I have negatives like this where I have no other alternative, which is why I started doing this a while ago, and since then I've decided that I'm going to try to use it for most of my scanning except where the negatives are "ordinary." Even my ordinary negatives tend not to be that ordinary, as 100% of my color film photography is using 10-20+ year old film, most of it cross-processed E100, and meanwhile I also tend to create a lot of very contrasty or dense B+W negatives because of the lighting and subjects I shoot and the fact that I generally can't be bothered to use anything other than Diafine.
Last edited:
I just read the ENTIRE thread that was linked, and the only things I could conclude, which I did not already know, were
- One could use Vuescan softward to convert neg to positive images with reasonable result (not much different here than using the scanner software that Canon provided with my 8800F flatbed scanner
- It really does not seem to matter about filtered vs unfiltered light source using dichroic head as the source...not much visible difference, not enough to matter unless you are being reaaaaaaly picky
- Using standard postprocessing software techniques for inversion really is quite time consumting to get color balance right
And now in post 28 we come back once again, the conclusion about light source 'it all really does not matter'.
...it all makes me wonder why I bothered to read that quoted thread, because it seems that whether the 'special conversion sofware' is Vuescan or Negative Lab Pro or something else, you still NEED IT...or else you need to invest a lot of time using run of the mill postprocessing software. Someone please correct a false conclusion here?!
This applies up to the point where you have a negative that doesn't scan well, and then "you still NEED IT!!" becomes "it doesn't work."
I can come up with a literal pound of negatives that have produced excellent images and nevertheless don't scan well on a flatbed.
Alternative techniques absolutely work for challenging negatives. I will have no further comment on this statement of yours.
MattKing
Moderator
I'm afraid you misunderstand.
I was suggesting that it is better to first deal with and negate the orange mask and overall red colour, and only then attend to the inversion.
It really depends on your requirements and the intended use of your images. With camera scanning you will be shooting in RAW format and you need software to convert RAW to other formats for post processing. Likewise, you need scanner software (Vuescan) if you go the flatbed scanner or dedicated film scanner route. In that sense, yes you still need this software.
That said, nothing stops you from using a different workflow. There are programs for converting RAW images to TIFF. You can then open the TIFF file in a photo editing program of your choice and use whatever tools you like to convert the negatives to positive images. Some prefer this workflow to Negative Lab Pro or Negadoctor as it gives them more control over the process. These folks usually have advanced image-editing skills.
There is also specialised software that does the conversion from RAW capture of the negative to a positive image, like Filmomat SmartConvert
Filmomat — SmartConvert Negative Conversion Software
Convert DSLR-scanned negatives easy and fast with Filmomat SmartConvert. The new standalone tool for negative conversion.
www.filmomat.eu
There are ways to recover images from severely faded color negatives. They involve taking multiple exposures using several carefully selected narrowband filters and reconstructing dye densities using known (or measured) transmittance spectra of the film dyes. This goes way beyound our normal image processing but it can be done.
Of course. There's no magic light source or filter pack that makes a digital camera behave like an RA4 paper from 1995. And even if there was, you'd still have to make judgements on color balance, so you'd still be pulling sliders.
OTOH, per my own experience with using scanner to scan neg (rather than using digital camera to shoot an image of the neg) there was little to no slider manipulation to obtain the last image which I posted.
Wow, that is really severe dye fade! The image posted in my earlier camera vs. scanner comparison was 42 year old neg (Kodacolor II), and this one (Kodacolor) is 61 years old...
In both cases, I took no extraordinary steps in storage...in glassine which was then kept in a cardboard box (similar to what you get for gifts from a department store). This shot was rediscovered when prepping my mother's home for sale this year.
Last edited:
Because the software "magic" is built into the scanning software. After all, that's what it's made for. However, many people are not happy with the colors they get straight from the scanning software when scanning color negative. So post processing often remains necessary.
Thanks for sharing the image. The storage conditions were not too bad. The box gave some protection from moisture and light. Also, processed negatives are more stable than undeveloped film.
The mask affects colors differently depending on their intensity and overall image exposure, which is why in most cases just "canceling" it with WB doesn't work well. Removing the bias of the mask as a first step significantly uncomplicates the process.
I wish LR had a straightforward way to "make a new DNG with LR adjustments out of this image and stick it right here in the catalog without doing an export" but still, unbelievably, after almost two decades, it doesn't. It would hugely simplify two-step processes where doing everything all at once requires really fiddly contrast adjustments, inversion, etc.
3d color profiles can replicate a lot of process accurately. Of course, you have to create them, and they have to be applied to something that they'll work on.
Yes, for color negative film, this would involve constructing a profile for the specific film used and preferably for specific lighting conditions, and it requires that the scanner allows a straight scan without adjustments to be made. The latter is a bit of a challenge; for instance, Epson flatbeds even when set to a scan without any adjustments exhibit scan to scan variations. Other scanners don't offer any option at all in this direction. I suppose something like Vuescan may be helpful here.
I believe others have mentioned it but it is so important to use an appropiate negative correction software, doing it by hand with curve inversion is too complex and takes too much time, best to use either Negative Lab Pro or Grain2Pixel (if you dont mind tinkering a bit more and prefer not to pay anything, like me), but this way you can get the best possible results. I also believe that scanners are easier to work with, and possibly more consistent, but digitalization with a camera allows for greater flexibility: you can do multiple formats with the same set up, you can exchange parts to improve the end result (better body, better lens, a better light source and holder) and its more feasible for people who travel often to carry this "system" than a scanner.
MattKing
Moderator
But that differential effect is designed to exactly counteract the non-linear colour behaviors of the dye sets. The sum of those non-linear colour behaviors and the differential response of the mask gives you a corrected total result, which is then available to be compensated for by the subsequent inversion process - whether that is supplied by RA4 paper or digital schools.
Essentially, the mask works as a colour "leveling" compound.
Honestly it's not that difficult and I really need to whip up some simple PS actions, as I have a bunch of these I'd like to redo (like, a couple dozen rolls at least).
* Cancel mask with complementary photo filter layer
* Invert
* Somewhat drastic levels or curves to get to the ballpark
* ACR filter for WB and then whatever
I get results that are reasonable for my modest expectations when dealing withproblem atypical negatives, and that are much better than anything I get with any of my scanning software options (VueScan, Silverfast, Epson).
Again if you have a negative that works great with your scanner, use that, obviously. If you have something that just isn't working, you're not out of options.
This also takes much less time than endless iterations with Silverfast that just don't work.
For b+w I will probably always use "camera scanning" in the future, as it eliminates all potential issues with density and/or fog.
* Cancel mask with complementary photo filter layer
* Invert
* Somewhat drastic levels or curves to get to the ballpark
* ACR filter for WB and then whatever
I get results that are reasonable for my modest expectations when dealing with
Again if you have a negative that works great with your scanner, use that, obviously. If you have something that just isn't working, you're not out of options.
This also takes much less time than endless iterations with Silverfast that just don't work.
For b+w I will probably always use "camera scanning" in the future, as it eliminates all potential issues with density and/or fog.
I tested more light sources that I can remember (halogen with different filters, different CRI white LED, different narrow spectrum RGB LED, RGB OLED) and looked at the design of many commercial scanners and can safely say that there is no magic solution that is "best".
It really depends on the combination of film dyes, spectrum of the light source, color filters on the sensor, and signal processing. It's a very complex interaction and each light source has it's strength and weaknesses, depending on the design of the whole unit. (there is even one motion picture film scanner that uses an RGB light source with two different Red sourced, one specially for certain archival footage which had different dyes).
If we limit the question to scanning with a normal digital photo camera and commercial software, then a cool white LED light source with a reasonably high CRI is a good, practical choice. Not all cameras will react the same to the same light source though, and a 99 CRI will not always work better than a 95 CRI, and the software further influences the result, so it's impossible to give general advise about a certain light.
The main problem is that we need different profiles for different films to get an accurate representation of how the film looks when printed optically (and even there there is some differences depending on the paper). It's really incredibly complex (I've been at it over three years until I felt good about my scanner) and none of the current camera scanning solutions really managed to solve that yet (actually, as other mentioned, commercial scanners sometimes struggle as well).
It really depends on the combination of film dyes, spectrum of the light source, color filters on the sensor, and signal processing. It's a very complex interaction and each light source has it's strength and weaknesses, depending on the design of the whole unit. (there is even one motion picture film scanner that uses an RGB light source with two different Red sourced, one specially for certain archival footage which had different dyes).
If we limit the question to scanning with a normal digital photo camera and commercial software, then a cool white LED light source with a reasonably high CRI is a good, practical choice. Not all cameras will react the same to the same light source though, and a 99 CRI will not always work better than a 95 CRI, and the software further influences the result, so it's impossible to give general advise about a certain light.
The main problem is that we need different profiles for different films to get an accurate representation of how the film looks when printed optically (and even there there is some differences depending on the paper). It's really incredibly complex (I've been at it over three years until I felt good about my scanner) and none of the current camera scanning solutions really managed to solve that yet (actually, as other mentioned, commercial scanners sometimes struggle as well).
I don't think anyone who has ever taken a photograph has been concerned with the photograph precisely representing reality. Or at least not twice.
Photographs ideally represent the experience of the photographer, or a third party, or an imagined entity.
No one cares whether "these are the right colors" or "these are the power lines in my viewfinder." What people want are "this looks like I remember it."
| Photrio.com contains affiliate links to products. We may receive a commission for purchases made through these links. To read our full affiliate disclosure statement please click Here. |
PHOTRIO PARTNERS EQUALLY FUNDING OUR COMMUNITY:  |