Proper operation sequences to correct color cast.

[ethanshen]

After reading the following post and article, I have a deeper understanding of how the color calibration of CP3 works.
viewtopic.php?t=964
https://www.colorperfect.com/filmtype_s ... ml?lang=en


If I understand correctly, when everything is set up properly — and after importing a negative into ColorNeg and selecting the corresponding film type — if there’s still a noticeable color cast, it could be due to one or more of the following reasons:

Film processing or film condition issues: The film might not have been developed according to standard procedures, the negative itself might have defects (expired film, improper storage), or there could be batch-related variations. In this case, the color cast can be corrected using FilmType / SubType / FilmGamma.

Non-neutral highlights in the photo: For example, a concert scene with colorful laser lights or a close-up of a vividly colored flower. In such cases, you can use AutoColor or other color correction functions to adjust the color balance.

Mismatch between film type and lighting conditions: For instance, shooting daylight film under indoor or cloudy lighting, or using tungsten film in daylight. In such situations, you can partially correct the color cast by clicking on a neutral gray area in the image — if your goal is to restore the white balance to how it appeared to the eye at the time of shooting (though of course, you may choose not to do so).

My question is: when I import a photo into ColorNeg and notice a color cast, how can I determine which of the above factors (or combination of them) is causing it?

Is there a recommended workflow — for example, should I start by trying AutoColor first, and if that doesn’t correct it, then move on to adjustments via FilmType or Click-Gray? In other words, what’s the best sequences of operations for handling color casts?


Additionally, I have a small question about Click-Gray: is its neutral gray correction based on the exact pixel that’s clicked, or on a small area (e.g., several pixels around the point) surrounding the click?


Thank you very much!
Ethan.

[C.Oldendorf]

This sounds straightforward at first, but it is actually a fairly nuanced question. We can broadly differentiate between calibration and per-image adjustment.
FilmType, SubType, and FilmGamma represent the characterization of a film stock and processing chain for a whole roll, often multiple rolls under comparable development. Correct characterization is the prerequisite for effortless color balancing. At the same time, per-image balancing can help reveal that characterization needs refinement. These two aspects inform one another.

There is also a third factor that strongly influences color and that sits between characterization and per-image work: black point and the three black-point colors. These are automatically chosen while scrolling BPTails, and for color-negative film it is usually best to address them early. If parts of the frame have no density above film base, then by definition those pixels represent pure black and form a natural anchor. Establishing black point is therefore not cosmetic — it determines what is black, what is shadow detail, and, importantly, how color is interpreted in the deepest tones.

So, a practical sequence for most color-negative work begins with a quick initial neutral balance. There is never harm in doing this first. If you have a neutral surface, click it. If you are unsure, choose a medium-brightness region. Once you have that initial landing point, evaluate black point via BPTails. Do not focus first on detail loss — focus on what happens to color. For color-negative film the correct BPTails position will often reveal itself very clearly in color behavior.

If the point where BP colors look best sits at a position where overall depth becomes too heavy and shadows collapse, there is an advanced option: switch to the B Point slider (plain black point), lower it slightly, and keep the BP colors discovered by BPTails. This is not needed often, but it is occasionally very useful.

Once black point and BP colors are established, we can evaluate whether the FilmType / SubType / FilmGamma characterization is right. A very good diagnostic is: if there are multiple neutral tones under the same lighting and after a well-placed gray click they disagree that is a strong signal that characterization would benefit from refinement. That is not a per-image step. It is a calibration task, done once for a roll or batch, as covered in the article linked above.

Now a clarification regarding one part of the question: for color-negative film there is a common misconception about illumination versus film designation. Shooting daylight film in tungsten light or tungsten film in daylight does not leave an inherent residual cast in ColorPerfect. The initial normalization step is independent of the scanner’s raw channel ratios. In other words, “twice as much red” or “half as much blue” is automatically normalized away. The same is true for gentle CC filters used during shooting. Only if filtration radically removes parts of the spectrum (hard cut-offs) does information loss occur. Otherwise, for color-negative film, illumination mismatch is not a color-balance concern — balancing is handled by design.

From here, per-image work is about choosing a CC filter pack. All balancing approaches in ColorPerfect — gray click, AutoColor, Ring CC, or in PerfectRAW mode the camera’s metadata / CC master presets — implement a pair of CC filters. One channel among RGB will always remain zero, because equal RGB is neutral mathematically. You can also type values directly; for example:

+10y increases yellow
-10y increases blue
(equivalents exist for R, G, B and C, M, Y)

If you establish more than one CC pack for different scene areas (whether via gray clicks, AutoColor, or numeric entry), you can fix one as reference with CC Ref, establish the second, and then hold Shift and scroll in the field with the largest absolute numeric value to gang-balance between them and find a sensible compromise for mixed light.

Regarding Click-Gray sampling: in Photoshop / Photoshop Elements, unless zoom is exactly 100%, the preview uses pixel-skipping, so even when zoomed you effectively click individual source pixels. In Photoline, previews are down-sampled, so multiple original pixels influence the click. That behavior has been stable for more than a decade.

To summarize the working order:

1. Initial neutral balance (gray click or AutoColor).
2. Set black point and BP colors using BPTails.
3. Evaluate FilmType / SubType / FilmGamma — refine only if necessary (not per-image).
4. Finalize CC filter pack using gray click, AutoColor, Ring CC, numeric entry, CC Ref, and Shift-scroll gang adjustment if appropriate.
5. Proceed to other tonal and creative adjustments.

Once this process becomes familiar, diagnosing the source of any cast — calibration, lighting, or per-image balance — becomes more intuitive and faster.

[ethanshen]

Thank you so much for your detailed and clear explanation, Christoph. Clearly once again a lot of new information and techniques for me to slowly understand and digest.

I tried following the steps you described and noticed something curious: whenever I use Gray-Click and then click AutoColor (or vice versa), the latter action seems to cancel out the effect of the previous one. Was this intentional in the design?

Another question — which might sound a bit silly, but I’m genuinely curious: why does every image start with an AutoColor value of 333.0? I mean, since AutoColor is computed based on each image’s unique black, white, and gray points, why do all images begin with the same initial value? Or does that number simply represent different color balances depending on the individual photo?

One more question: for color negatives, is the out-of-the-box exposure level shown in ColorNeg correspond to the actual exposure level of the shot? For example, if the out-of-the-box result from ColorNeg appears about one stop underexposed, does that mean the original capture was indeed underexposed by one stop?

Sorry for asking so many questions. That said, I want to make it clear (so that other users don’t misunderstand) that in my experience so far, for most color negatives I’ve tested, the results from CP3 are the best among all the methods I’ve ever tried. Many times I find myself surprised by realizing “So this is what my negative was actually meant to look like”. I’ll start gradually uploading some of the photos personally think turned out pretty well to the forum.

[C.Oldendorf]

You are very welcome to ask all these questions — in fact, this is exactly what I hoped would happen on this forum. As time goes on, many of these items will already have been discussed and can simply be referenced, which is why I try to write things in a reusable, generally helpful way.

In ColorPerfect, it is useful to distinguish between two fundamental classes of color-balance operations.

The first group consists of controls that establish an entirely new CC filter pack — meaning they compute a fresh, complete set of color-correction filters.

GrayClick is one of these. When you click a point, the plugin makes that color neutral and constructs a new CC filter pack. If the color clicked is very far off, we intentionally require you to hold Shift while clicking in order to force neutrality. This prevents accidental clicks from causing extreme color casts. If you want to enforce neutrality, hold Shift while Gray-Clicking and, if physically possible, the plugin will honor it.

Another tool in this “full new CC pack” group is AutoColor. It scans a range of tonal relationships and establishes an internally computed balance. If you find a setting you like on that scale, that becomes your new CC pack.

In PerfectRAW mode, using the button to cycle through Working / Best / Camera CC also swaps entire CC packs. You can go back to values derived from camera metadata, return to your working state, or “Save as Best” to lock the best result for future comparison. This is a PerfectRAW-specific mechanism. The older CC-Master panel can also save CC information in files — that feature has been around a long time. I have not actively used it myself for years, and although I am not aware of issues, it may get a review for CP3.

The second group consists of tools that modify the existing CC filter pack gradually. The most visual example is Ring CC, where each click shifts balance slightly toward green, red, blue, yellow, cyan, or magenta, or brightens/darkens the image in small steps.

When entering values directly, anything prefixed with “+” or “−” modifies the current filter pack additively. For example, +10R means “add a 10R filter to whatever is already in place.” After this addition, the neutral component is automatically removed so you end up with a valid filter pair. The underlying CC filter system is quite elegant.

So if Gray-Click and AutoColor seem to “cancel” one another, that is expected: both create new CC packs, and the last one used becomes the new baseline.

About the default AutoColor value of 333 — this comes from discussion on the forum. In CP2, the initial balance depended on the brightness threshold in Options. Some users prefer setting that threshold extremely low so they almost never see clipped specular highlights initially. That unfortunately destroys the chance of a correct initial color balance on many images. So in CP3, the plugin simply starts using the AutoColor algorithm from the moment it opens. I had to choose a position on that tool’s range; 333, about one-third up, proved robust across many images. It is not a literal “value” — it is an index telling AutoColor where to begin searching. If testing shows a better starting point, it will change.

Exposure and “out-of-the-box” appearance in ColorNeg: the initial rendering does not represent the actual exposure offset of the negative. Color negative film has a wide straight-line region on its characteristic curves. You can expose the same scene at +1, +2, or even +3 EV and, as long as you remain within the linear part of the curve, digitized frames will look very similar when first opened — differences mainly appear as changes in noise distribution the farther you move from base density.

If we wanted to show three exposure differences explicitly, we would need to digitize all frames under identical illumination, aperture, ISO, and shutter speed, then either composite them or process one and apply settings to the others via Optimized Carry-Over. Only then would you see the expected one-stop differences in the rendered results.

But the intended mental model in ColorNeg is that we are effectively printing onto imagined color photographic paper. As long as the negative was exposed within film’s usable limits, the output can be adjusted to look correct. Therefore, a darker initial display does not necessarily mean the original capture was underexposed by that amount.

What controls the initial image brightness is the according threshold set on ColorNeg’s Options.

[ethanshen]

To summarize the working order:
1. Initial neutral balance (gray click or AutoColor).
2. Set black point and BP colors using BPTails.
3. Evaluate FilmType / SubType / FilmGamma — refine only if necessary (not per-image).
4. Finalize CC filter pack using gray click, AutoColor, Ring CC, numeric entry, CC Ref, and Shift-scroll gang adjustment if appropriate.
Proceed to other tonal and creative adjustments.

I’m gradually beginning to understand what you meant by this workflow (if I’m interpreting it correctly):

Premise: If an image already looks nearly perfect color-wise, there’s no need for major color adjustments. The following steps apply when a noticeable color cast is present.

Step 1: Use GrayClick or AutoColor to establish an overall target result — or as you said, a landing point. Since the effect of GrayClick will later be overwritten (true?) by adjustments to BPTail, this step mainly serves as a preview of the desired final outcome.

Step 2: Adjust BPTail and B Point. This is because part of the color cast might come from the black point values, and it’s best to correct those early in the process.

Step 3: If needed, modify FilmType / SubType / FilmGamma.

Step 4: Finalize the CC filter pack.

Regarding the earlier issue I mentioned — where GrayClick and AutoColor seem to cancel each other’s effects — I now understand why that happens. However, I’ve noticed something else: as described in Step 1, modifying BPTail also appears to cancel previous adjustments from either AutoColor or GrayClick. Is this behavior intentional as well?

Another thought: since GrayClick and AutoColor both belong to the same pack, would it be possible for the AutoColor index to change correspondingly when performing a GrayClick? What I mean is that GrayClick is by nature somewhat imprecise — so if I could see the resulting AutoColor index after performing a GrayClick, I could fine-tune the color more precisely by adjusting within a nearby range.

For example: after gray-clicking on a roughly neutral-gray object in the photo, the AutoColor index shifts from 333 to 741, presumably. If I still feel the result isn’t quite “neutral,” I could explore values in AutoColor slightly above or below 741 to find a better balance.

I understand that this feature isn’t currently viable — but would it be feasible to link those two in a future version? Or have I completely misunderstood how they work internally?

But the intended mental model in ColorNeg is that we are effectively printing onto imagined color photographic paper. As long as the negative was exposed within film’s usable limits, the output can be adjusted to look correct.

I’m truly glad to hear this, because that’s exactly the ultimate goal I (and I believe many other users) have — to reproduce what the negative would have looked like when printed onto photographic paper.

[C.Oldendorf]

The Blackpoint, the CC filter pack, and the Black implementation go all the way back to the very beginning of the ColorPerfect project, so we are easily looking at ~15 years of history there. I would need to check specific edge-case behavior, but one thing is certain: changing the selected color-negative film from the maker and film type lists will reset a large portion of this system. That is expected. It is also expected that black point must be established before refining color balance.

I had believed that a gray click would survive even after adjusting BPTails, but as you observed, there may be scenarios where this is not the case. I will need to look at the underlying logic again to confirm. There can certainly be edge cases.

Regarding your question about linking GrayClick and the AutoColor index: it is important to remember that AutoColor is an algorithmic method to derive plausible CC filter packs based on the statistical distribution of tonal values in the brighter parts of the image. It works well, but it intentionally produces only a limited set of candidates. A gray click can absolutely produce CC filter packs that are nowhere near the AutoColor range — and that is by design, because a gray click anchors to actual scene content rather than statistical inference.

[robyferrero]

Speaking of PerfectRAW, which shouldn't be any different from other modules, I personally haven't experimented much with gray click, but I can say that when you find the right pixel, it's a sight to behold.
It's also very interesting to perform a dual gray/shadow/highlight reading with CCRef.

But ultimately, I'm happy with AutoColor.
I was already happy with CP2's AutoColor, but I'm even better with CP3's AutoColor 333.

It seems to me that often, or quite often, AutoColor click doesn't deviate from the base curve with AutoColor 333, which already provides excellent color balancing.

In any case, with AutoColor + related adjustments with RingAround, you can find balanced, or extremely adequate, color.

After all, you basically need to balance with a monitor calibrated for photographic printing.
If you plan on achieving a balance that works across all the computers you share the image with, then you'll never stop balancing.

If you then also wear anti-blue computer glasses, which remove blue light, then everyone sees color differently.
I make this observation because I recently purchased anti-blue computer glasses and discovered this huge discrepancy between yellow and blue.

[C.Oldendorf]

AutoColor is a very versatile tool as long as the main light source — in terms of specular reflections or the brightest regions of the image — is essentially the governing brightest sensation in the frame. As soon as that is not the case, which can very well happen in a macro shot of a vividly colored flower (say, a buttercup or similar), AutoColor will fail. In those situations we have plenty of other means to arrive at a beautiful CC filter pack.

Regarding specialized glasses for computer workspaces: yes, I know they exist, and I understand that they filter out blue light. I never got a pair because I figured that for color-critical work — which is the core of what I do — that would be detrimental. They might be useful for coding with white UIs, but on macOS with Dark Mode there is basically no advantage to specialized “computer glasses.”

[robyferrero]

In practice, if there are no specular highlights or brighter areas—that is, essentially, when the images are low-contrast, underexposed (?), or in any case flat—isn't AutoColor the best solution? Is it better to try to balance the color with Click Gray?

I'm not sure I understand correctly when you say "the brightest areas should be the areas with the brightest impression in the frame."

By definition, the brightest area is the area with the brightest impression in the frame.

Is the problem with brightness, or with a more or less vibrant color?

Here's an example of flowers adjusted with AutoColor + RingAround.
It's true that there aren't any bright colors, but overall, there's no strong brightness either; there are a few brighter spots on some petals.

2011-105-4484-AutoColor_+_RingAround-roby_ferrero.jpg (335.43 KiB) Viewed 289 times

[Full image link - opens in new tab]

[robyferrero]

Glasses are great, but in some ways they're a bummer, like the color.

After all, they're prescription glasses.
When a photographer needs glasses, he has to decide whether to see color well or keep his eyes more rested by attenuating the blue.

This variation in glasses color takes a bit of getting used to, but it seems that it's basically difficult to mess up the color balance because of the glasses; it's strange that when you take them off, your photo looks decidedly bluer, but after a while, the photo seems less blue.

It doesn't make sense because of what I said previously, but how do you see this photo: normal, a little blue, too blue?

[ethanshen]

Here's an example of flowers adjusted with AutoColor + RingAround.
It's true that there aren't any bright colors, but overall, there's no strong brightness either; there are a few brighter spots on some petals.

I think it is these brighter spots on the petals which are considered to be the brightest part of the photo, though not many, but enough for the CP system to consider them as "true bright". (Correct me if I am wrong)

By the way, is the flower photo taken by digital camera and went through PerfectRaw or is it a film negative through ColorNeg? It looks fantastic.

[robyferrero]

Of course, the bright spots are those of the petals.

It looks fantastic to me too.
The camera is a Canon 30D that went through MakeTiff and PerfectRAW, and what other way would there be?

[robyferrero]

I'm trying Gray Click instead of AutoColor
In a situation like this, where there are bright flowers, even though it's not a macro image.

You can see how Gray Click has achieved better color balance than AutoColor.

2017-002-0336-AutoColor_vs_Gray_Click-roby_ferrero.jpg (312.08 KiB) Viewed 260 times

[Full image link - opens in new tab]

In both cases, RingAround was necessary, but with Gray Click, the optimal result was achieved.

Then again, if we take small steps in the color package, it is possible to achieve the same result with AutoColor + RingAround. But here, we achieved the balance rather quickly.

I'm not saying it's the best balance achievable, it could be better, but it already seems like a pretty acceptable result.

So, although AutoColor can do an excellent job in the best circumstances, as Christoph says, this is not always the case.
There are more complicated situations, such as those described in previous posts.

Gray Click, on the other hand, can solve any situation, with the help of RingAround if necessary, but it is important to find the right gray, the right pixel.

Now, a personal reflection on the search for the neutral zone.

In a photograph like this, it is not easy to find the gray point, the neutral point. There are photographs that are even more difficult than these, such as macro shots, for example, where finding gray amid all the green and bright colors is a challenge. When you do find it, it is likely to be a very small point to click on.

So, while my old 2008 computer's low resolution allowed me to zoom in to 100%, which was already large enough to be inspected carefully to find the gray point amid the foliage, the same cannot be said for my 4K computer, which, despite having a default standard resolution of 2048x1152 and not 2560x1440, its 100% is still a little too small to comfortably find and click on the gray dots in the image.

I don't know if this is my problem, or if it's a problem that everyone has. But if so, with computers becoming increasingly high-resolution, might it be useful to increase the CP magnification percentage to something like 200/300%?

[robyferrero]

The test continues and convinces me to use Gray Click. Then, when it's easy to see, it's even more accurate. In this case, RingAround doesn't help.

If I were to add a yellow dot to AutoColor with RingAround, the yellow would go beyond Gray Click. So I'd have to find a half-dot, a quarter-dot, etc.
At this point, the AutoColor result would be more yellow, and perhaps even a little red, especially on neutral, bright clouds.

2012-116-3712-AutoColor_vs_Gray_Click-roby_ferrero.jpg (186.53 KiB) Viewed 254 times

[Full image link - opens in new tab]

[ethanshen]

The test continues and convinces me to use Gray Click. Then, when it's easy to see, it's even more accurate.

Hi robyferrero,

You’re absolutely right — in certain situations, GrayClick definitely has advantages over AutoColor, especially in photos where there's a clearly visible neutral gray that can be identified by eye. What I mean is, finding a truly neutral gray requires some editing experience, and it’s also a trial-and-error process. Since GrayClick operates on the exact pixel we click on, the noise and grain present in film scans/digitizations might influence the outcome. That’s why it often requires clicking back and forth within a small area to test which spot gives the most optimal result (I usually save different results in an A, B, C… list to compare). With digital RAW files, this process might be slightly easier, because the noise pattern is usually less random compared to film grain.

AutoColor provides a useful level of automation. While it may not always produce the best or final result, it does offer several viable starting points for reference. Personally, I think it would be fantastic if AutoColor could be used as a fine-tuning aid after GrayClick, rather than replacing its effect entirely. That combination would be extremely powerful.

All the best,
Ethan.

[robyferrero]

Hi Ethanshen,

It's true, I was happy with AutoColor, but when Gray Click works well, it's unbeatable.

Autocolor 333 is more accurate than the previous version, so you don't even need to use AutoColor Clic anymore; it works very well by default.
You can adjust the shot with RingAround.
You don't need to use RingAround with Gray Clic.
With Gray Clic, as you said, if you find the right Clic, it's perfect.
You have to look for it, and sometimes it's difficult to find, sometimes it takes a lot of patience, but if you don't give up, you'll always find it, even in unexpected places, you'll even find it in the middle of colors.
With the native digital file, it's a little easier than scanning with grain and noise, but there is also noise in the native digital file.

An AutoColor to support Gray Clic could be a good idea.
I don't know how feasible it is, as I believe AutoColor does its own calculation based on linear file data, rather than on changes already made. A second AutoColor by Gray Clic would be needed. But only Christoph can say if this is true.

Bear in mind that CC Ref does not act as AutoColor, but it can help in that sense; it can adjust a color temperature from warm to cool and vice versa.
It is true that saying warm and cool is reductive, because there is not only yellow and blue, but the concept is the same: adjusting temperatures to make them relevant to the scene.

The point is that gray is not just neutral gray, it is warm gray, it is cool gray, that's how it is in nature.
That's how it is in a sunset.
In a sunset with gray clouds, the clouds are not neutral at all, or rather, there are neutral ones too, but they are warm or cool gray clouds; if you click on cool, it becomes warm and vice versa.
You can't make a warm gray neutral, otherwise you ruin the whole photo.
So there are a lot of factors to consider when looking for color balance; it's not just a matter of clicking.

In any case, I always find everything surprising.
As I like to say, CP color is beautiful even when it's ugly or incorrect.

Look at what came out in this scene; this is what I call surprising. It doesn't matter if it's not entirely accurate, it just has to work.
Here the fire is incandescent, and the whole thing seems to work.

All the best!

2024-005-1489-roby_ferrero.jpg (310.94 KiB) Viewed 160 times

[Full image link - opens in new tab]