Scanning negative 35mm: color accuracy, color balance, film bias

[finn21]

My question is about scanning color 35mm film. I use professional film (Portra 400) and a very good lab. I get consistent results and prints.

This year i have invested in semi-pro or pro scanner Reflecta 10T with expensive SilverFast Ai Studio 8. And I intend to scan 35mm film for printing purposes. My goal is to have very high quality 30 x 40 inch prints.

Scanner is giving me fantastic results: 230 MB, 7052 x 4715 pixels (33,5 MP) TIFFs with great natural/crisp film grain and overall great look.

BUT the colors are off. Always. I use NegaFix (piece of SilverFast) with the correct Portra 400 profile and I try to put accurate WHITE, BLACK and GREY points (where they are in picture) but I never get consistent and real results. Pics are never "right", always off, always to bright or to warm (to cold).

I read all forums about how scanning is difficult and about SilverFast and Scan Vue, Color perfect and Photoshop but everywhere people have problems getting colors right.

My questions are:

- Should I just continue this guessing game and try to somehow spot the right color everytime?
- Have you tried putting Grey Card 18% and set this as a correct exposure in scanning soft or Lightroom and color balance from there?
- Have you tried putting Color Balance Card (White, Black, Grey) to help scanning software determine exposure and color balance and color balance from there?
- Have you tried X-Rite's Color Checker Passport or other color calibration target to set full color correction this way?
- If yes, what about film bias. (Portra 160 more pastel, Ektar 100 a lot more vivid) Wouldn't it just get you to the POINT ZERO with film where you have no film specific properties?

The goal is to have as close to 35mm film scanned image as possible, without any interpretation. Just what you get from film processed neutral and printed neutral. To see real film color specific for it's kind. I just want to see good Portra 400 on my screen.

Thanks for your time.

 

[John_M_King]

Scanning

When I don't use my darkroom, I scan med format and 35mm using a quite elderly Epson V500 flatbed scanner. I find that most scanners don't take kindly to scanning over exposed or over developed negatives but they can be managed. Under exposed or underdeveloped negatives do not normally present a problem.

Also on the Epson there is a button 'restore colour' and this goes a long way to correcting errors.

I also find that if a negative has areas of great density difference such as bright sky (under exposed) and normally exposed forground. it is better to adjust the scanner so that it gives a good scan of the sky area and denser forground, then correct the difference using Photoshop etc. Go the other way and you will be working for ever and a day!

 

[Alan Klein]

With my V600, I scan flat and use Levels or black and white points in post processing to get the histogram straight. That tends to get 90% of the colors correct.

 

[John_M_King]

Use of levels - now thats interesting - I have never tried that for a lifetime. Thanks.

 

[John_M_King]

Use of levels - now thats interesting - I have never tried that in a lifetime. Thanks.

 

[WilmarcoImaging]

A big improvement for me was to use a color profile in the scanner and the same profile in the postprocessing software, and in the monitor. At the moment I am using Adobe RGB 1998. Images scanned from Portra 400 in 120 size look good on several devices. Have a look at the client gallery for STR on my website. These were done using this method.

Levels adjustment channel by channel is still needed for fine tuning. An overall levels adjustment is often needed to get the right feel.

wilmarcoimaging.com

 

[Adrian Bacon]

You always have to color correct film. The lab you where using before was doing this whether you knew it or not. The reason for this is that film (like a digital camera) has a native color temperature, or spectral response. For daylight film, the manufacturer generally makes it either 5000K or 5500-5600K, or somewhere in that range. A good film profile assumes the same color temperature as the film. If you shoot in the same light, you don't have to do anything.

I don't know about you, but just like shooting with a digital camera set to daylight WB outside during the day, I still find myself adjusting the color balance of most of the images because the color temperature is almost never exactly the same as the camera's response. It's the same with film, except you have the added complexity of variances in the development process, and any scanner introduced color bias stacked on top of that.

The good news is that white balance is just gain applied to individual color channels to get the desired response, it's just a matter of being as consistent as possible in developing and scanning, using a good film profile, and then doing basic color correction.

 

[John_M_King]

To add to my previous post above. I also use a Nikon Coolscan5 and have never had to correct any scns, perhaps the software has an auto correct feature I don't know about. I know the scanner allows me to scan in RAW which I always use but even then I don't have a problem

 

[holggger]

My question is about scanning color 35mm film. I use professional film (Portra 400) and a very good lab. I get consistent results and prints.
- If yes, what about film bias. (Portra 160 more pastel, Ektar 100 a lot more vivid) Wouldn't it just get you to the POINT ZERO with film where you have no film specific properties?

Wondering about the same thing, I came upon this thread – and I see the question has not really be answered yet.

Yes, as Adrian Bacon says, every film has a fixed color temperature – but how to get the look of that in your scans? (along with other typical colors, color casts, "looks" etc. that certain films have compared to others).
I do know how to color correct an image and I have no problem making my scanned color negatives look neutral, but I don't want neutral, I want the color cast that occurs because of the different light temperatures I was shooting at.

Now, if exposure was 100% consistent in every shot I assume you could just adjust for one image using the RGB levels and leave the settings there for the others, but change in exposure etc. of course screws with the colors, as well.

Does anyone here have a general workflow for color adjustments on color negative scans with particular focus on the above?
(as mentioned, I know how to get a neutral image from my negatives but this is not what I want)

At the moment I am scanning the new Portra 160 with a Flextight/Imacon scanner and let their software handle the negative inversion/getting rid of the orange mask while adjusting everything else with RGB levels before scanning to have an image file I use as basis for fine tuning later.

Thank you

 

[Kino]

You need to study color science and calibration. Calibrate your monitor, scanner and printer to the same profile and a known color space.

Just because it looks right on an uncalibrated monitor in an unknown color space doesn't mean it will print in those colors on your uncalibrated printer.

Otherwise, just futz around until it looks right...

 

[holggger]

You need to study color science and calibration. Calibrate your monitor, scanner and printer to the same profile and a known color space.

Just because it looks right on an uncalibrated monitor in an unknown color space doesn't mean it will print in those colors on your uncalibrated printer.

Otherwise, just futz around until it looks right...

I assume your post was not directed towards what I wrote above?
What you are saying is absolutely correct and common sense, of course, but misses the point of what I was trying to say a little bit.

 

[Kino]

I assume your post was not directed towards what I wrote above?
What you are saying is absolutely correct and common sense, of course, but misses the point of what I was trying to say a little bit.

No, it was not directed at your posting; it was just a general observation.

You'd need a color meter on location to record the color temperature to approximate the local conditions.

 

[Adrian Bacon]

Wondering about the same thing, I came upon this thread – and I see the question has not really be answered yet.

Yes, as Adrian Bacon says, every film has a fixed color temperature – but how to get the look of that in your scans? (along with other typical colors, color casts, "looks" etc. that certain films have compared to others).
I do know how to color correct an image and I have no problem making my scanned color negatives look neutral, but I don't want neutral, I want the color cast that occurs because of the different light temperatures I was shooting at.

Now, if exposure was 100% consistent in every shot I assume you could just adjust for one image using the RGB levels and leave the settings there for the others, but change in exposure etc. of course screws with the colors, as well.

Does anyone here have a general workflow for color adjustments on color negative scans with particular focus on the above?
(as mentioned, I know how to get a neutral image from my negatives but this is not what I want)

At the moment I am scanning the new Portra 160 with a Flextight/Imacon scanner and let their software handle the negative inversion/getting rid of the orange mask while adjusting everything else with RGB levels before scanning to have an image file I use as basis for fine tuning later.

Thank you

If you conform the raw scanned film to a color space (e.g. Adobe RGB) using a Macbeth color checker chart shot at a specific color temperature (e.g. 5000K, since that's the standard for ICC profiles), and embed an ICC profile of that color space, then yes, you do remove what makes that film unique as far as color response and color saturation are concerned. That step essentially conforms the films color response so that it is accurate for a given color temperature. You still keep the grain and dynamic range though.

That isn't really a problem though because even though you've "baselined" the films color, you now have a stable and consistent base to change the color to the look you want. If you want the portra look or the ektar look then go look at the tech sheet for that film and change the image to match using photoshops or Lightroom's color tools.

Keep in mind, profiling a film is all about getting to a consistent and stable workflow. If you don't do that, then you're going to spend a lot of time in PS or LR for each frame mucking around until it looks like what you want. Personally, I prefer a stable, fast, and consistent workflow where all scans are brought to a uniform base first. From there you can get to the look you want, save it as a preset, then it's just a matter of applying that preset to future scans. That's just me though. There isn't really any right or wrong way. If it looks good, then it doesn't really matter how you got there.

 

[Kino]

There are so many problems dealing with subjective color in any medium; film or digital. As Adrian notes, you can at least get a good, stable workflow to produce neutral images from which to start, but the "look" of the film image via digital is totally subjective and that is where the skill of recall and color correction come into play.

Of course, the "look" of any film based medium depends on the color of the light source illuminating it, so that look can differ from hour to hour, but the feeling it evokes the mind's eye of the photographer should be reproducible with sufficient manipulation in a color correction software package.

 

[jim10219]

What I do is try to remain consistent in my scanning methods. Then I take my scans into Photoshop, and reverse them using the curves tool. Generally what I'll do is adjust the curves for the first photo in my roll and save that curve as a preset. Then I'll apply that preset to begin the next photo, and make any edits to it to make it look even better. I do that for about half the roll or so. Then I go back and reapply the latest saved preset to the first several scans. Each time I edit the preset, I'm trying not to create an entirely new preset, but rather, tweak the old preset to work with all of the photos. Eventually, I have a preset that does a good job with any photo shot with that roll of film, developed in that method, and scanned in that process. From that point on, all I have to do the next time I shoot a roll of that film is scan it, apply the preset, crop it, and remove any dust. Each time you change something in your process, you'll have to change your preset, so try to keep everything consistent.

As for using the curves tool, it's pretty daunting at first, but as you get more experience with it, it becomes a lot easier and quicker to manage. My usual process is to invert the slope on each color channel (because we're dealing with negatives), set the high point and the low point for each color (red, green, and blue) in the histogram, and then drop one more points in the middle of the high and low point to push or pull the curve to get smoother blacks, grays, and whites. Keep applying to curve to new photos, and you'll start to see where each high and low point happen most frequently, and how much you need to push or pull each curve.

Now, occasionally I'll have to adjust the colors a bit more beyond what the preset does. That's not unusual. Sometimes it's to compensate for something that went wrong in the process (poor exposure or wrong color temp. of lighting), and sometimes it's just for artistic reasons. Sometimes you'll need more than one point in the curve to manipulate it how you want/need. The more time you spend with this tool, the easier it becomes to use. I much prefer it over any of those plugins or separate programs that try to automate this process. The curves tool does a much more reliable and accurate job, and once you get the hang of it, it doesn't take any longer to do. Each person's work flow is different, and that's why presets are often disappointing. But if you create a preset for your person workflow, you can have the best of both worlds.

 

[holggger]

That isn't really a problem though because even though you've "baselined" the films color, you now have a stable and consistent base to change the color to the look you want. If you want the portra look or the ektar look then go look at the tech sheet for that film and change the image to match using photoshops or Lightroom's color tools.

Maybe I am having a problem here with my way of thinking.
Taking slide film as an example things are pretty easy: You see what you get. If you shot your daylight balanced film at sunset time or in a tungsten lit room, then you get a certain color cast. These are extremes, but you also get certain more subtle color casts depending on what light you shot in, as well as getting a certain rendering of e.g. green or skin color, which all is an inherent quality of the film you are using.
When scanning this film, as long as you keep your settings constant you get pretty much the same qualities in your scanned images (let's not get into color management questions like "but is the green in the image exactly the same you see on your monitor and then the same as the green your printer spits out when you print the image" at this point)

And now the case of color negative film. I guess there is always a color correction process, even when printing color negs in a darkroom.
My problem is how to get these inherent qualities of the film into my scans – or is this simply not possible because there's always the choice/preference of the operator printing/scanning the film?

Starting from a neutral color corrected image, achieving the look of the above mentioned image shot in tungsten light and the look of the sunset image require some very different corrections. A preset will not do that.
In the same way I assume the less extreme color renderings inherent to the film cannot be simply emulated with a preset, unless we assume that all images were shot under exactly the same light conditions.

So to close the circle, is what I am asking for simply impossible because the way I am thinking about this is fundamentally wrong?

 

[Les Sarile]

So to close the circle, is what I am asking for simply impossible because the way I am thinking about this is fundamentally wrong?

Quest for the Holy Grail - getting "natural" colors from each specific color negative. Unfortunately even Kodak admitted that there is no standardization among scanners. So even if you used calibrated equipment, one film, one light, lens, exposure there are no guarantees they will come out the same from each scanner.

A couple of examples from my own archives.

This same frame of what I consider a perfectly exposed Kodak Gold 100 came out so differently you would not think they were the same.

Same frame of Kodak Ektar 100

I have scanned over 40K frames using my Coolscan+Nikonscan and few thousand more from various others (minilabs, dedicated, flatbeds) and I have found the Nikon combination to provide the best results - automatically.

 

[holggger]

Quest for the Holy Grail - getting "natural" colors from each specific color negative. Unfortunately even Kodak admitted that there is no standardization among scanners. So even if you used calibrated equipment, one film, one light, lens, exposure there are no guarantees they will come out the same from each scanner.

Very interesting, thank you so much for the example images. The top ones look so much better than the ones on the bottom.

Maybe the best way would be to do it like Michael Fox describes here on his site http://www.michaelfoxphoto.com/resources/tech_papers/imacon_neg_setup.html
I just found this today while further searching for ways to do it on the Hasselblad Flextight with its Flexcolor software, since that is the only software the scanner supports, as far as I know.
This could be applied to other scanners and software as well, of course, and in a nutshell it comes down to creating your own film profile by shooting a color chart at standardized light temperature (the color temperature the film you are using is calibrated for). Assuming, of course, that the film is calibrated for neutral white, black and grey at this temperature.
Using this profile without further (color!) adjustments you would get all the inherent color casts of the film in different light situations in your scans.

One question remains, though, and maybe anyone of you has the answer:
How much does over-/underexposing negative color film as well as variance of parameters during film development really influence the resulting colors?
If change in exposure and change in development parameters has only minor effects on the resulting color (and only on brightness, contrast, etc.) then above mentioned method might be a good way to go if you shoot a lot of a certain kind of film.

 

[Adrian Bacon]

Maybe I am having a problem here with my way of thinking.
Taking slide film as an example things are pretty easy: You see what you get. If you shot your daylight balanced film at sunset time or in a tungsten lit room, then you get a certain color cast. These are extremes, but you also get certain more subtle color casts depending on what light you shot in, as well as getting a certain rendering of e.g. green or skin color, which all is an inherent quality of the film you are using.
When scanning this film, as long as you keep your settings constant you get pretty much the same qualities in your scanned images (let's not get into color management questions like "but is the green in the image exactly the same you see on your monitor and then the same as the green your printer spits out when you print the image" at this point)

And now the case of color negative film. I guess there is always a color correction process, even when printing color negs in a darkroom.
My problem is how to get these inherent qualities of the film into my scans – or is this simply not possible because there's always the choice/preference of the operator printing/scanning the film?

Starting from a neutral color corrected image, achieving the look of the above mentioned image shot in tungsten light and the look of the sunset image require some very different corrections. A preset will not do that.
In the same way I assume the less extreme color renderings inherent to the film cannot be simply emulated with a preset, unless we assume that all images were shot under exactly the same light conditions.

So to close the circle, is what I am asking for simply impossible because the way I am thinking about this is fundamentally wrong?

You're not wrong, you're just not thinking about the whole picture, so to speak. The thing to keep in mind, that it's all relative. When we think about what makes a films color characteristics, it's relative to other films and relative to the "colorspace" of the paper on an analog print. This is why it's important to get the scan conformed to a known starting point. My workflow starting point is a floating point scene referred linear light space using the color XYZ primaries as defined in the identity color space from the nine degrees below website run by Elle Stone. It's pretty easy to figure out what RGB values a Macbeth chart is supposed to have in that color space, then conform the raw scan to that, assuming you shot a Macbeth chart at the 5000K that the ICC spec uses.

Once you're there, you assign/ embed that ICC profile and all the color tools suddenly magically start working the same way they do in digital land, meaning in Adobe Lightroom, you just set the color temperature like you would on a camera raw image. The saturation sliders and all the other color tools work the same way.

It's a little complex to get to that point, but once you get that worked out, it becomes really reliable and consistent.

 

[Les Sarile]

One question remains, though, and maybe anyone of you has the answer:
How much does over-/underexposing negative color film as well as variance of parameters during film development really influence the resulting colors?
If change in exposure and change in development parameters has only minor effects on the resulting color (and only on brightness, contrast, etc.) then above mentioned method might be a good way to go if you shoot a lot of a certain kind of film.

Goes along with Kodak's statement about no standardization across scanners and also how each scanner's software handles autoexposure and each film's latitude and color response to exposure and lighting temperatures.

This is the result from Kodak Portra 400 latitude test using Coolscan + Nikonscan with autoexposure turned on.

Larger version -> http://www.fototime.com/B1379B2FE749C83/orig.jpg

You can see that colors are reasonably very close over most of the range with variations of contrast and brightness as expected. With a little post work, you can probably use -1 through +7 and get very usable results.

 

[holggger]

Goes along with Kodak's statement about no standardization across scanners and also how each scanner's software handles autoexposure and each film's latitude and color response to exposure and lighting temperatures.

This is the result from Kodak Portra 400 latitude test using Coolscan + Nikonscan with autoexposure turned on.
[…|

You can see that colors are reasonably very close over most of the range with variations of contrast and brightness as expected. With a little post work, you can probably use -1 through +7 and get very usable results.

Thanks, this is great!
Assuming that auto-exposure did not mess with the colors these results are very reassuring.

 

[warden]

- If yes, what about film bias. (Portra 160 more pastel, Ektar 100 a lot more vivid) Wouldn't it just get you to the POINT ZERO with film where you have no film specific properties?

The goal is to have as close to 35mm film scanned image as possible, without any interpretation. Just what you get from film processed neutral and printed neutral. To see real film color specific for it's kind. I just want to see good Portra 400 on my screen.

Your last bullet point has been my experience (speaking as an amateur who has struggled with color using several of the approaches you listed). The act of scanning and color correcting to what you're calling point zero tends to strip the image of its film flavor, and then it's your job to put it back however you choose with sliders, graphs, buttons. For me this is a joyless pursuit and I've often considered keeping only black and white film, and using a digital camera for color.

Recently I've been scanning color negatives, correcting for the film base, finding black, white and grey assuming they're all present in the scene, and then screwing around until my uncalibrated eye likes what it sees on my imperfect screen in my poorly lit office on a slightly out of gamut street. ;-)

I think if you're a pro the only answer is to become an expert on the subject of color correction over time, or hire one any old time.

 

[alanrockwood]

Here's what I do when scanning negatives. I'm not saying that this is the best way. In fact, it probably is not, but it is the best I have come up with. I scan using Vuescan. I do not scan as negative (which would result in an automatic color inversion), so the scan looks like a negative. I then bring the scanned image into an image processing program, like Photoline or GIMP. I make sure it is a 16 bit image (converted to 16 bit if necessary). I then use levels to do an initial adjustment. I basically set the upper an lower level in each of red, green, blue spaces, setting the markers near the upper and lower edges of the histograms for each color. This scheme seems to do a reasonable job of removing the color mask. Then I invert the image. At this point contrast is weak, and sometimes there is a residual color cast, but usually the color cast is not too bad. Then I use the levels menu to adjust gamma (contrast) for the overall image (not the gamma for individual color channels). Then I go to curves and do some fine tuning until it looks good to my eye. (Sometimes I change the order of operations, such as inverting before setting the upper and lower cutoffs for each color, etc.)

There is a lot to criticize in this scheme, and it relies on some assumptions that are sometimes questionable, but works for me most of the time.

I would love to have some tutorials on how to do things much better.

Additional note added as an edit: As of the date of this post the stable version of GIMP is still not 16 bit. However, the development version can do 16 bits.

 

[Patrick Robert James]

I've never used Silverfast, but there should be a way for you to select a profile for your scanner and your film. The other option would be to use a program like Color Perfect and scan raw.