Getting good exposure is cruicial for repro photos of negatives (and may often not be intuitive)

[C.Oldendorf]

Attached are two versions of the same color negative captured on Canon EOS 5D Mk II but 4EV apart.
That exposure difference is all it takes to make it or break it...

These are from a set of five exposures of the same color negative kindly given to me by our user Wendelin Bottländer on July 27th 2018. To whom we extend a big thank you. It's been a while and the image has become quite dear to me as I have seen it so much in testing various builds.

If it could tell of all the things we have seen together that would be fun, For example the longer work on post inversion re-interpolation. Fascinating stuff which did not amount to much as an interpolation method was added to libRAW that works as well when used before inversion.

[C.Oldendorf]

Ok next let's see how different sizes of JPEG behave with different RGB working space profiles embedded.

"Place thumbnail inline" IMG_1634 +2 EW (Adobe RGB 1998 full).jpg:

[Full image link - opens in new tab]

"Place thumbnail inline" 1,920px sRGB version:

[Full image link - opens in new tab]

The term “thumbnail” is a bit misleading in our setup. I’ve set the inline thumbnail display size to 1,000 px on the longer side so that such images are large enough to view meaningfully, while still being clickable for enlargement. With quality and conversion set high, these are essentially full images inline rather than small attachment previews.

I’ve added a Lightbox extension to the forum. Without Lightbox, a thumbnail linked straight to the full-size file. I thus re-programmed the board's template to embed a direct to image link below what is placed via "Place thumbnail inline",

[C.Oldendorf]

One more thing to try, does the lightbox stuff work if we do not inline at all?

Non inlined attachments get listed below and are the 1000px Thumbnails in case of images.

I hoped to get to those and add the direct links I did for the inlined items... Well, I tried and nothing worked thus far. The image links will thus stay exclusive to actively inlined Thumbnails, they will not be part of the Attachments box for now. Whatever is in the Attachments box gets included in the lightbox gallery (where full size images get used, not the 1000px so called thumbnails).

[robyferrero]

Image uploading works perfectly.

I tried processing the file you provided.
As you can see, it came out very differently.
Your color cast is much better.
The quality of my file, in terms of noise and artifacts, is rather poor. I don't know why; perhaps it was generated by LibRAW, and perhaps I boosted the highlights a bit too much.
Other post-production flaws/errors are out of the question; I'm still not very good with CP, and my experience with scanning or reproducing from negatives is still limited.
I also applied a slight sharpening, but the sky is reduced by a further 50%.

[C.Oldendorf]

Two versions of the same color negative captured on Canon EOS 5D Mk II but 4EV apart. That is all it takes to make it or break it...

What I wanted to emphasize is that one exposure will work out beautifully while the other will fail miserably. Thank you Roberto for proving that point by deliberately picking the –2 EV version — it shows very clearly what happens when the signal-to-noise ratio collapses. With only about 8 bits (≈256 tones) available in the brightest channel and even fewer in the darker ones, color integrity suffers beyond repair.

Because repro-photography usually allows for constant illumination, the most reliable results come from working in full manual camera mode with a fixed set of parameters:

• Use the lowest available ISO setting to minimize digital noise.
• Set the aperture to a value like f/5.6 or f/8 that suits the lens and task and does not introduce refraction as a problem.
• Control exposure only by adjusting time (or, if using flash, by altering light intensity or distance).
• As a last resort, open the aperture further — but avoid this if possible.
• If you have a diffusor close to the film keep it squeaky clean so DOF does not make dust on it visible in pictures.
• For the repro shots use a light source that is somewhat close to daylight rather than tungsten.

Evaluating exposure
The camera’s JPEG histogram is only a rough indicator. Once you think you’re close, convert a first shot into a linear TIFF via MakeTiff and check the histogram in Photoshop. This shows the true sensor data. The TIFF will look dark before inversion in ColorPerfect, which is normal and expected. The histogram, however, tells you what matters.

Guidelines

• Watch the brightest channel first.
• If its data ends well before mid-histogram, add +1 EV.
• If less than a quarter of bins are populated, add +2 EV.
• In practice, including unexposed film base in your frame is very helpful: nothing on the negative can be brighter, so it acts as a reference.
• Never allow clipping at the right edge — even a small spike means overexposure and lost information.

Bottom line: Maximize data captured
Exposure is critical with color negatives. A seemingly small deviation of a couple of stops can be the difference between a file that grades beautifully in ColorPerfect DC and one that is beyond salvage.

Camera JPEGs will most likely be deceiving and will not help in achieving this goal. I'll attach relevant examples from this set. In camera regular exposure looks fine but captures only 10 bit of data where we can get 12 with +2EV compensation which looks totally blown in camera.

[robyferrero]

Very interesting.
At first glance, I'd personally choose the +-0EV nominal exposure.
I'll pay more attention to this data with my next reproductions.

[C.Oldendorf]

At first glance, I'd personally choose the +-0EV nominal exposure.

Indeed, we must get used to the fact that while the JPEG engine’s output can give us an idea of what gets captured, it does not relate well to the raw data — particularly not in off–color-balance scenes like color negatives with their filter layers.

Daylight-like illumination rather than tungsten-like will typically work best for color negatives — be that LED, flash, overcast sky, or special tube lights. Early on I thought it might be good to use a color head to filter the color masks out, but results are not better, for reasons I understand today. Not totally impossible to do, but also not recommended.

Both our posted images in one gallery:
+2EV:

[Full image link - opens in new tab]

-2EV:

[Full image link - opens in new tab]

[robyferrero]

Well, now I understand more clearly how the color negative is treated during reproduction.

[ethanshen]

Guidelines

• Watch the brightest channel first.
• If its data ends well before mid-histogram, add +1 EV.
• If less than a quarter of bins are populated, add +2 EV.
• In practice, including unexposed film base in your frame is very helpful: nothing on the negative can be brighter, so it acts as a reference.
• Never allow clipping at the right edge — even a small spike means overexposure and lost information.

Bottom line: Maximize data captured
Exposure is critical with color negatives. A seemingly small deviation of a couple of stops can be the difference between a file that grades beautifully in ColorPerfect DC and one that is beyond salvage.

Sorry to brought this discussion up once again.

I am considering which is the best exposure for the DC scans for color negative films.
From your guideline, it says, "Never allow clipping at the right edge — even a small spike means overexposure and lost information. "
If I understand correctly, it refers to only the content of the shot, rather than the base of the film, right?

What I want to mention is, after the conversion by MakeTiff:
The histogram in Photoshop of the tiff file WITH film base is clipping on the right side,but when I cropping out only the content of the photo, it is not clipping.
By inputing it into the CP3, the title part shows that this photo still have some headroom left (say, 0.3R, 0.3G, 1.0B).
Then, by my current understanding, this is a valid DC scan without any clipping of the image (even though the film base part is clipping).

Am I understand it right? Please correct my if I was wrong on any details.
Thanks!

[C.Oldendorf]

If I understand correctly, it refers to only the content of the shot, rather than the base of the film, right?

Exactly. Don't over expose the unexposed base of the film. Nothing can be brighter than it in a color negative and you're totally safe.

What I want to mention is, after the conversion by MakeTiff:
The histogram in Photoshop of the tiff file WITH film base is clipping on the right side,but when I cropping out only the content of the photo, it is not clipping.
By inputing it into the CP3, the title part shows that this photo still have some headroom left (say, 0.3R, 0.3G, 1.0B).
Then, by my current understanding, this is a valid DC scan without any clipping of the image (even though the film base part is clipping).

Yes, what is outside of the negative makes no difference to us. It can be the sun for all we care as it will be cropped away anyhow.
It actually is best to crop before calling ColorPerfect. If you want a version with sprocket holes etc. crop anyway and after you OK out of ColorPerfect, use Photoshop's history to go back before the crop and then call ColorPerfect again to quickly bring the adjustments back to the entire image using the Exact carry over choice as detailed here: https://www.colorperfect.com/restoring/ ... orPerfect/

[ethanshen]

Exactly. Don't over expose the unexposed base of the film. Nothing can be brighter than it in a color negative and you're totally safe.

I see. So basically try to expose the unexposed base of film to its limit but not over exposed, is the nearly best exposure for DC scans. Thank you Christoph for the explanation. Now I can safely conduct the exposure.

What about white balance setting in the digital camera, does it matter?
From my shallow knowledge of raw file interpretation, the white balance data is not directly written into the raw file but instead as a "sidecar" data. The raw-interpretation software can choose whether to use it or not. From the output of Maketiff, clearly it does not include that data, so my guess is the white balance setting in the camera does not matter the output of the DC scan anyway. Am I correct?

Also, what about the choices of working color space in Photoshop?
I read somewhere else from Christoph that "the Prophoto color space is quite tricky". I can't recall where I read it but always remembered that. So what I did with CP3 is I set Adobe1998 to both the working color space of Photoshop and the targetting color space for Maketiff transformation. Will it be the best practice for DC scans?


Thanks in advance!
Ethan.

[C.Oldendorf]

What about white balance setting in the digital camera, does it matter?

For reproductive photos of color negatives, the white balance information stored as metadata by the camera is fully irrelevant. That is because we do not strive to create a color-correct version of the still-negative representation with the orange mask in place, but rather invert the frame and aim to achieve a natural rendition of the scene captured.

Of course, the light source makes a difference, and I’ve said before that at present I feel that anything more or less resembling daylight is going to be fine, whereas anything resembling tungsten lighting will require more processing. That’s not to say that it’s necessarily impossible; I just feel that the extra mile introduced by it is not one we must travel.

To clarify, I should also say that, of course, MakeTiff reads the white balance information included by cameras as metadata and transfers it into the TIFF file in a way that ColorPerfect can deal with, because in PerfectRAW mode — where the camera image represents the scene and is not an intermediate of a negative in turn representing the scene — camera white balance can indeed be used.

So the summary is: camera white balance recording for ColorNeg DC is irrelevant, while for PerfectRAW it can totally be used.

Also, what about the choices of working color space in Photoshop?
I read somewhere else from Christoph that "the Pro Photo RGB color space is quite tricky"…

In the past, the whole RGB working space debate was something of a hold-up, let’s say. Folks wanted to use wide-gamut color spaces in ways that they weren’t really fit to be used. We tried to explain what they were, how they related to scenes captured on film, and why, particularly when working with scans, it made sense to use narrower gamut color spaces with ColorPerfect 2 instead.

All of that was one of the most difficult things for photographers to accept, and I suppose also to understand — to really get to the bottom of. Years ago, Dave said to me, “We’re not the color police, you know.” Meaning that while we hoped to make our users understand their choices, we didn’t want to orchestrate them. We weren’t trying to create some color dogma — and I’m not today either.

I thus decided to take this whole field out of the equation in ColorPerfect 3 by including suitable steps so that anybody can use any supported color space anywhere without it affecting the general operability of the plug-in. I also extended support a lot but not in MakeTiff yet.

For scans, that means a distinction between narrow-gamut and wide-gamut assignments still exists. In ColorNeg SC mode, you can assign a wide-gamut color space such as ProPhoto RGB to a scan, and ColorPerfect 3 will perform internal color processing so that the output becomes meaningful and consistent. When a narrow-gamut color space is assigned instead, the plug-in behaves just like ColorPerfect 2 did: the RGB primaries of the narrow-gamut color space directly form the final image’s RGB, as encoded in the negative itself — just as they would appear in a projection of that negative onto color photographic paper.

In contrast, when working from digital camera reproductions of color negatives — that is, in ColorNeg DC mode — the situation is different. Here, considerable color processing is always required to take the digital camera out of the equation, so the distinction between narrow- and wide-gamut assignments that exists for scans no longer applies.

We are no longer using the ProPhoto RGB primaries as stand-ins for the color photographic papers’ red, green, and blue dyes — which never would have made sense and is not something we ever suggested doing, but rather something that resulted from assigning ProPhoto RGB to images of linear origin and using ColorPerfect 2 on them. The challenges of ProPhoto RGB follow downstream. It encompasses such a large color gamut that your screen will not be able to display it. And if you can’t display it, you may introduce colors that only show up in print — for example, certain cyan hues generated in computer graphics. That can be a good thing because you can print them and you don’t need to display them faithfully if you know what you want. So that’s something to look out for.

Furthermore, because ProPhoto RGB encompasses what we can call hypothetical colors — that is, those lying outside of human vision — the portion of the container volume actually representing visible colors is smaller than it would be in narrower-gamut encodings. For 16-bit-per-channel images, that has no negative implications at all. But if anybody decides to convert to 8-bit in Photoshop without first converting to a narrower-gamut RGB working space, that would lead to problems. Suddenly we have a small container volume, and then only a portion of it is being used for actually visible colors.

As long as folks know what they’re doing, any color encoding will be fine — and even if they don’t, any detriments arising from such an RGB working space choice are outside of ColorPerfect itself. That said, for ColorNeg DC, I think Adobe RGB (1998) is a useful choice. But you can also experiment with wider settings and see whether the math then behaves in a way you like. Of course, you’ll need to convert the final result to something practical for downstream use.

[ethanshen]

Thank you so much for such a comprehensive and detailed explanation.

In particular, the part about color spaces was extremely enlightening for me. I’ve long lacked the theoretical understanding behind it — my observations came only from repeated practical experiments. I often noticed that when working with color negative film, using ProPhoto RGB tends to produce colors that look less natural (regardless of which method I use to remove the orange mask), even though in theory it’s supposed to be the widest color space. Thanks to your description, I now have at least a partial understanding of the principles behind this phenomenon. Of course, there’s still a lot to digest and fully grasp, but I now have some theoretical foundation — and a clear guideline — that makes me feel more confident when choosing a color space.

Christoph, have you ever thought about compiling this information — including aspects like the proper exposure level when scanning negatives (whether with a scanner or a digital camera), the choice of color space, white balance, and so on — into a systematic guideline that could serve as a go-to reference for users before using ColorPerfect? Personally, I’ve found that different choices in workflow have a huge impact on the final results. Exposure, in particular, plays a decisive role. I believe many users, like myself, have been “misled” by histograms in cameras or other editing software, and as a result, haven’t fully captured all the information in the film. Whether the exposure is correct makes an enormous difference to the outcome in CP3. Since I started exposing correctly, I’ve truly appreciated the full potential of CP3.

Of course, there’s no absolute “right” choice when it comes to exposure or color space, but I believe having a reasonably structured guideline would give users a much stronger starting point.

So, have you considered writing a comprehensive guideline and placing it somewhere easy to find — for example, pinned or featured on the ColorPerfect website? I think it would be of great help to many users.

Once again, Christoph, thank you so very much for your patience and detailed explanations — I’ve learned a tremendous amount from them.

[C.Oldendorf]

For ColorNeg DC, I’d like to create a sibling documentation to the existing one for black and white film.

One more upside to the forum is that it gives me an opportunity to discuss much of the information I’ll eventually need to document — without first having to formally write and shape it into an authoritative article that covers everything. The whole ColorNeg DC topic goes much deeper than one might think, as over the past couple of years I’ve explored it quite extensively. On the hardware side of digitizing 35 mm color negative film, I even developed prototype solutions and produced small runs of custom hardware.

In the end, though, I decided not to get sidetracked by all that and instead to see the creation of ColorPerfect 3 through to a form fit for public release before getting distracted by other fascinating projects.

What remains most challenging for me is living up to my original hope of being an artist and photographer — the very motivation that drove me to do all this for so many years. But that, too, is beginning to return. So I can honestly say that, overall, 2025 is shaping up to be a pretty good year.

[ethanshen]

For ColorNeg DC, I’d like to create a sibling documentation to the existing one for black and white film.

This is so great. Really looking forward to it!

The whole ColorNeg DC topic goes much deeper than one might think, as over the past couple of years I’ve explored it quite extensively.

Yeah I know 100% what you mean.
I’ve read dozens of theoretical and non-theoretical discussions about methods for removing the color mask over the last couple years, and experimented with countless different approaches. Therefore I know how challenging this process can be — far beyond what most people might imagine.

[C.Oldendorf]

For scans of color-negative film made with devices well suited to the task, such as Nikon’s long-discontinued medium-format CoolScan series, the funny thing to me has always been that everyone kept talking about removal of the orange mask. Yet that removal is actually a trivial matter — the real problems lie elsewhere.

However, when you say this today, and when I look at what needs to be done to get anywhere close to having ColorNeg DC, these things blend together more. If we acknowledge that removal of the orange mask is part of the complexity that ColorNeg DC must handle, then indeed, for the first time I can appreciate that orange-mask removal is far from trivial for digital-camera reproduction photographs — at least not if the goal is to successfully transform the color-negative film’s densities back into an intensity-accurate representation of the original scene.

[robyferrero]

I can say that, from the little I have tried ColorNegDC — it was a test I had to do — it has already given me the best results I have ever had on that image, an image that was not so perfect and had mixed lighting. So I think that with more experience, the results will be astonishing.

viewtopic.php?t=938

Speaking of surprises, I will now upload something on the topic of my presentation.

[ethanshen]

I can say that, from the little I have tried ColorNegDC — it was a test I had to do — it has already given me the best results I have ever had on that image, an image that was not so perfect and had mixed lighting. So I think that with more experience, the results will be astonishing.

Hi Robyferrero,

Same for me. I have tested quite several color negatives I digitally scanned before and, it turnes out, the CP3 always give me the best results, which is like you said, astonishing.

[robyferrero]

Hi Ethanshen,
Yes, the great thing is that the same goes for all CP modules; ColorNEG, ColorPOS, and even PerfectRAW are amazing.