Versione in Italiano QUI
Introduction
I absolutely love my Fujifilm GFX 100s, and I believe it’s one of the best purchases I’ve ever made because, with its incredible sensor and technology, it truly allows me to effortlessly transform what I see into images. Among the countless strengths of this system, I unfortunately struggle a lot with Fujifilm’s Color Science which tends to produce magenta-leaning tones, especially on neutrals and underexposed areas. On top of that, Adobe refuses to even decently handle Fuji’s RAW files, leaving us once again to suffer during post-production.
I then had the crazy idea of turning to social media (Threads) to connect with other Fuji owners, hoping to hear their thoughts and learn how they had solved the issue. However, after probably (definitely) asking the wrong provocative question, I inevitably found myself facing hordes of absolutists, ignorants, and denialist fanboys.
From this valley of tears and blows where I was starting to believe that I was the problem, a series of private messages brought some comfort. These messages confirmed that I wasn’t alone in struggling with this issue but was just the only one who was foolish enough to hope for a civil discussion on social media.
Among these messages, one stood out: a lifesaving angel named Mirko asked me if I had ever heard of Cobalt Images and their color profiles.
Initially very skeptical, the more I researched, the more intriguing it seemed. So, I decided to put this apparent black magic—crafted by some shaman in a remote village deep in the Amazon rainforest—to the test.
As you can guess, if I’m here talking about it today, it’s because it turned out to be far from black magic. The “shaman” was (is) actually a Master of Color. So, once again, let’s dive together to see just how deep the rabbit hole goes.
Premises
As usual, this article is probably longer than necessary, but you know me—these topics are a passion of mine. The subject at hand is fairly complex, but as always, I’ll do my best to simplify it as much as possible.
To fully enjoy it, it helps to have a basic understanding of Color Management concepts: we’ll talk about color spaces, profiles, and similar topics. You can find various articles on my blog to dive deeper, but don’t worry—you’ll make it to the end of this one without too much trouble.
If you choose to proceed, make sure to bring along plenty of patience, an open mind, and, as always, a good glass of wine.
Regarding the post-production software I used to install the profiles discussed in this article, my choice inevitably fell on Capture One. Every Fujifilm user knows that Adobe Camera Raw is still far from being able to handle Fuji RAW files in an acceptable way. However, it’s worth noting that Cobalt Image profiles are available for a wide range of software, including DxO products, RawTherapee, and Luminar. This means the insights shared here apply to any software. Just make sure you purchase the profile package specific to your editing software since each application processes RAW files differently, so profiles need to be tailored accordingly to work effectively.
We’ll discuss the methodologies for evaluating profiles later on, but it’s important to specify right away that to fully appreciate the visual analysis of the sample images used (or at least to see what I’m describing in the text), it’s absolutely essential that your monitor is calibrated and profiled (you can find dozens of guides on this topic on this blog). And in general, you know it: there’s a special circle of hell reserved for photographers who don’t calibrate and profile their monitors! 🙂
Difference Between Profiles and Presets
Let’s start from the very beginning.
One of the (abominable) things I kept reading in that Threads discussion was the utter confusion between Profiles and Presets. I believe this is a fundamental misconception that needs to be cleared up because it’s the cornerstone of everything that follows.
When we talk about Presets, we’re referring to a set of predefined adjustments that can be applied with a single click. These adjustments affect specific parameters (like brightness, contrast, and color balance) to achieve a particular look. A Preset can include multiple parameter changes that you can still tweak further after applying it. If you’ve ever seen those ads on Instagram from photographers selling Lightroom preset packs to make your photos look just like everyone else’s, you’re buying a Preset—and honestly, there’s no worse way to waste your money in photography.
Profiles, on the other hand, deal with the mathematical interpretation of colors in a RAW image. They operate on a deeper level than Presets, influencing how colors and tones are translated within the editing software. Unlike Presets, Profiles don’t directly modify the sliders in the software’s development panel and are instead used as a foundation for further adjustments. They allow you to simulate specific styles, like the look of a particular film stock or a specific color space, providing a consistent tonal and chromatic base for post-production.
A Profile, unlike a Preset, plays a critical role in “translating” the raw data of a RAW file (which isn’t an image) into a visible image.
This transformation involves two main stages:
- Demosaicing: A RAW file is essentially a collection of raw data from the sensor, where each pixel captures only one color component (red, green, or blue) through the Bayer filter (or an alternative depending on the camera). Demosaicing is the process that combines this data to create an RGB pixel matrix, rendering the image visible in full color.
- Profile Application: After demosaicing, the Profile comes into play to interpret the colors and contrast of the image. The Profile instructs the software on how to manage tones and colors, giving the photo its base “look.” This is why Profiles can drastically change an image’s appearance: each one interprets RAW data differently, enhancing or muting certain characteristics.
This is why we’re here today, putting high hopes in Cobalt Image profiles, while you start treating magical Presets with as much suspicion as investment offers from call centers.
It’s easy to spot the difference between someone selling Profiles and someone selling Presets: in before/after comparisons, Profiles don’t have that Instagram-style “wow” effect.
Why Choose Cobalt Image Color Profiles
At this point, it’s legitimate to ask: why would I choose Cobalt Image profiles over the ones we find in Lightroom, Capture One, and others, provided directly by the software developers and camera manufacturers?
It took me a while to answer this question, and I was able to do so both by reading the excellent technical documentation on the Cobalt Image website and by running some nerdy tests that I will share shortly. Ultimately, the essence is that Cobalt Image color profiles are based on a more accurate interpretation of the data, leveraging the original information from the camera’s sensor and then applying specific interpretations depending on the case.
The process to create these profiles is as follows:
- Sensor Response Analysis: The Cobalt Image team measures the specific response of each camera sensor (brand and model) to ensure a custom and accurate profile. This means that each profile is designed to match the specific behavior of the camera, not generically all cameras (like an Adobe profile does).
- Color Linearization: The first part of the profile converts raw information into a linear space, eliminating the chromatic distortions introduced by the camera. This ensures that colors are treated with the highest possible fidelity.
- Artistic Interpretation: After linearization, the profile optionally applies chromatic modifications designed to create a specific visual aesthetic. A unique color rendering is then developed to emphasize certain tones or contrasts while maintaining a realistic approach. Profiles are thus designed (and I underline this word) ranging from more neutral ones (such as Repro or Flat) to more specific ones, such as Landscape, Product, or Studio profiles.
This method of profile creation allows for several advantages.
One of these is undoubtedly that, at this point, our starting point is independent of the camera we use, as uniform looks are created across different camera models. This means that a photographer using multiple camera bodies will no longer have different starting points in post-production (or, for landscape photographers, the day we change brands will no longer be a nightmare).
Another important point, and perhaps the one that matters most to me, is that if the process of sensor response analysis and linearization is done properly, the profile can truly “squeeze” every “drop” of color available from the sensor.
Let’s now look at some images to see the results I’ve found.
The camera sensor response analysis is validated using the highly advanced HR-1 SuperChroma target, featuring 1,132 patches
The Base Pack and Cobalt Image Emulations
Cobalt Image offers several packages, but the first one to consider in this article (also because it is necessary for any further packages) is the one called Base Pack: as we have already said, this will contain color profiles designed for a specific camera (so no longer generic) and for specific software (in my case, as already mentioned, Capture One).
The package I purchased for Capture One contains three profiles:
1.Cobalt Standard: Designed particularly for natural light portrait photography, thanks to the optimization of skin tones, which ensures extraordinarily realistic complexions. The key feature is chromatic accuracy for skin tones, guaranteeing naturally vivid portraits without looking overly processed. It also maximizes the camera’s dynamic range, balancing details in shadows and highlights, thus preserving depth and contrast in the image.
2.Cobalt Neutral: Designed particularly for landscape and nature photography, it ensures accurate color representation, reflecting the real scene without embellishments or oversaturation while prioritizing the preservation of the subject’s colorimetry. It is optimized for natural light conditions, harmonizing the dynamic range of the camera sensor with that of the scene to produce well-balanced images with clear details at all levels of illumination.
3.Cobalt Repro: Offers a linear response that captures the entire dynamic range of the camera without any alteration. Practically speaking, this profile brings the raw image to the screen in its purest form. To a less experienced eye, it may seem like a loss of contrast and saturation, but it allows for the maximum editing latitude by starting from the RAW’s purest form.
If you purchase the package for Adobe, DxO, Luminar, or RawTherapee, you will also find an additional profile:
4.Cobalt Flat: Designed for commercial/product photography in controlled lighting (studio environments). It is optimized to display colors as perceived by the human eye, optimizing scenes for SDR monitors and reducing the need for tone mapping while preserving detail.
An absolutely essential thing to understand is that there is NO right or wrong profile; there is only the profile we prefer based on its design characteristics and the photography itself. If we remember how the profiles are created (i.e., by first analyzing the sensor’s response), we can agree that with all three (or four) profiles, we squeeze the maximum information from our sensor and then, if necessary, apply an artistic interpretation.
During my tests, I obviously thoroughly tested all three profiles, but since the choice is solely tied to artistic interpretation, I decided to focus the article on comparing them to the generic and standard profiles provided by camera and post-production software manufacturers. This allows for a comparison not just on the subjective artistic component but also on the objective qualitative component of the profile.
Extra Note for Hardcore Nerds: To be precise, if you download the profiles for Capture One, you will find six profiles, not three. This is because Capture One requires creating a profile for each illuminant, as it uses ICC profiles originally designed for scanners, which do not support dynamic adaptation to different lighting conditions. Unlike Adobe DCP profiles, which manage two illuminants and can interpolate automatically between them, ICC profiles for C1 work well only with a specific color temperature (e.g., daylight) and degrade rapidly with different lighting. This limits flexibility and requires separate profiles for each type of illuminant. For C1, you will therefore find “Daylight” and “Tungsten” profile variants.
But that’s not all. This Base Pack represents only the starting point of the Cobalt Image ecosystem. It is possible to go further and purchase the so-called Emulations: color profiles designed to faithfully replicate the chromatic aesthetic and tonal rendering of other photographic systems.
As mentioned earlier, however, we will discuss this in a dedicated future article.
The Cobalt Image Ecosystem
Methodology for Analyzing Cobalt Image Profiles
I spent a lot of time wondering what the most objective way to evaluate these Cobalt Image profiles could be.
Since these are Profiles and not Presets, the evaluation cannot simply be based on what we immediately see on the screen after applying the profile. This is because what we are applying is a starting point (profile) and not an end point (preset), and its appeal would be entirely subjective as it would only relate to the artistic interpretation applied.
I therefore decided to perform an analytical color analysis as usual, focusing on its three-dimensionality, meaning the combination of chromaticity and luminance.
The general idea is to analyze the colors in a set of images generated from “representative” RAW files, applying different test profiles (Cobalt and non-Cobalt) in a color space broad enough not to limit the RAW (so ProPhoto RGB will work perfectly for us). This will allow us to see, quite literally, “who can flex the most muscle” Does this sound complicated? It is, but we will approach it in the simplest way possible. However, as I mentioned earlier, I recommend a quick review of the basics of Color Management to better enjoy what we’ll see next.
First, let’s define our battlefield, which will be the ProPhoto RGB space, represented in three dimensions with yxY coordinates and projected onto the CIE 1931 chromaticity diagram (the mythical “colored bell”). As you surely remember (right? right??), the bell gives us a two-dimensional representation of what we, as humans, can typically see with our eyes: what is inside the bell is visible to us, while what is outside is not. The cage, on the other hand, represents the range of colors available to our editing software to transform the raw data of the RAW into what we call an image. This is just to show that profiles have access to plenty of colors—many more than we can see. Therefore, we will pit them against each other on a gigantic ring where they won’t be able to make any excuses.
It will be like a fight in the ring (or rather, in a cage), where the profiles have plenty of space to slap each other around
What we will do is take every single pixel of our representative images and place it within this gigantic grid. We’ll then visually observe how chromaticity and luminance develop in our image in volumetric terms.
By comparing different profiles on the same images, we’ll see which one is able to make the most of the information contained in the RAW file.
Generic example of image decomposition
The final theme concerns the selection of representative images, which has given me more than a few headaches.
My final decision was to narrow it down to just two types of images, and for each type, perform the comparative analysis mentioned earlier on at least ten images (for a total of 20 images with a large number of profiles applied—you can imagine how much time this required). The choice of image types stems from a mix of factors, including areas where I know cameras (or rather, their sensors) typically struggle the most, and, of course, what matters most to me as a landscape photographer (specializing in long-exposure seascapes). Thus, I focused on scenarios where I know my sensor will be pushed the hardest. However, I am confident that, in the end, you’ll agree it’s reasonable to expect similar results for any image, regardless of photographic genre.
The first type I selected is what I’ve called “Image R”, which is taken at the threshold of highlight clipping in warm tones. Yes, the classic fiery sunset (or sunrise), where we aim to squeeze every last drop from the sensor to capture every possible color transition in the sky.
The second type is what I’ve named “Image G B”, representing an image captured under more typical lighting conditions (if not slightly underexposed, due to my long-exposure shooting technique). These images include significant areas with green and blue tones.
With this approach, I believe I can assess how profiles perform across the entire dynamic range of the sensor.
Since the test results were consistent across all ten photos in each type, reporting on all 20 analyzed images would be unreasonably tedious and unnecessary. Therefore, we will now discuss only one image per type, focusing on the most significant profiles. Of course, feel free to contact me if you’d like to review the others, but the results are truly comparable.
Profile Analysis with Image R
For each RAW file in this category, I applied various profiles in Capture One without making any post-processing adjustments to the base image. This means the image was simply converted to JPEG from RAW with the applied profile in the ProPhoto RGB color space.
The profiles I applied were:
- Cobalt Neutral
- Cobalt Repro
- Fujifilm Pro Negative Standard
- Fujifilm PROVIA
Note for Capture One users: After installing the profile according to the provided instructions, go to the “Base Characteristics” tool, select the ICC profile you want, and set the Curve to “Film Standard” (this is important; otherwise, the Cobalt profile will not be activated correctly). In Adobe Lightroom, it’s simpler—just select it from the Profile dropdown menu.
Visually, here’s what applying the profiles entailed (notwithstanding the debatable image compression from the blog):
From a purely visual perspective, what I can already notice is that the Cobalt profiles seem to offer more color transitions in the warm tones, while unfortunately, Fuji’s tendency toward magenta in midtones and shadows persists. In this case, honestly, I don’t care much about Fuji’s magenta shift because my sole priority is to maximize the sky’s potential, even if that means “pushing” the sensor in less favorable ways for other parts of the image.
Analyzing this observation analytically, I can see that the data indeed confirms it. For example, let’s compare the two profiles that visually seem the most appealing as starting points: Fujifilm Pro Negative Standard and Cobalt Repro.
What you see is honestly surprising because it’s clear that using a Cobalt profile allows us to fully maximize our RAW file (and therefore our sensor). We are literally using more colors, both in terms of chromaticity (notice how, in the first 2D image, the green area is significantly larger than the yellow one) and luminance, where we have not only more information in the shadows but also— and most importantly in this case— more information in the highlights (observe how, in the 3D graph, the green “peak” surpasses the yellow one).
This was evident in this specific image, but the results were remarkably similar across the other nine images analyzed in the “R” category.
Not bad, huh?
Profile Analysis with Image G B
Now, let’s make similar observations on an image that isn’t as critical in terms of exposure and contains significant portions of green and blue to explore how the profiles behave near these primary tones.
This time, we’ll test the following profiles:
- Cobalt Neutral
- Cobalt Repro
- Fujifilm Pro Negative Standard
- CaptureOne Linear Response
I’ve decided to exclude Fujifilm PROVIA outright because the amount of magenta it introduces is frankly embarrassing. Additionally, this approach allows us to compare pairs of profiles designed with similar goals (e.g., Cobalt Neutral and Fujifilm Pro Negative Standard). In particular, I was curious to examine the behavior of the two linear profiles—those without any “artistic” adjustments applied—allowing us to evaluate the RAW file in its purest state. These are Cobalt Repro and CaptureOne Linear Response.
From a purely visual standpoint, here’s what we observe in one of the representative images (once again, no corrections applied, only the selected profile).
Visually, it’s quite clear that the interpretation of blues and greens is significantly different (and, in my opinion, much more pleasing), with the magenta madness being noticeably reduced. This difference is even more apparent when comparing the images side by side as thumbnails rather than viewing them full size.
At this point, one might ask, “But how can I be sure that Cobalt Image’s interpretation is the correct one?” This doubt is valid and does require a certain leap of faith. However, to support Cobalt Image’s case, let me remind you of the following:
- The sensor characterization used to create the profiles is validated using the HR-1 target with 1132 patches rather than the standard 24-patch targets.
- They achieved a four-star rating on the FADGI Color SG target.
- Their SMI (Metamerism Index) is 79, compared to Adobe profiles’ 65.
So, yes, I would say that Cobalt Image profiles are likely a more accurate interpretation. That said, as always, when evaluating visual results, subjective judgment is probably the only one that truly matters.
Now, as usual, let’s seek comfort in our pixel-by-pixel analysis.
Let’s start by comparing the Cobalt Repro and CaptureOne Linear Response profiles, which are both linear response profiles.
Honestly, what I see is stunning: it’s absolutely clear how the Cobalt Image profile once again pushes the sensor to its limits, utilizing a greater range of chromaticities (as clearly shown in the 2D graph). If we look at the 3D breakdown, the increased vertical development of our image is equally evident.
Once again, this is achieved by simply comparing linear response profiles, with no artistic intent—just maximizing the sensor’s information.
If we then compare the Cobalt Neutral profile (designed specifically for landscape, nature, and macro photography) with Fujifilm Pro Negative Standard (which I usually consider Fuji’s best profile), here’s what we observe:
Once again, the Cobalt profile appears to be significantly superior, particularly in the development of green and blue chromaticities.
Conclusions
At this point, I think I can draw my conclusions: Cobalt Image profiles are not only visually more pleasing than standard and generic profiles but also analytically demonstrate a better utilization of the information contained in RAW files.
After analyzing 20 RAW files with an average of 4 profiles each, resulting in a total of 80 images, the data incontrovertibly shows superior color development across the board with Cobalt Image profiles, both in chromaticity and luminance. This allows you to begin post-production from a genuine advantage in terms of data quantity and quality.
Not only do the Cobalt profiles fully unleash the potential of our sensor, but they also do so with exceptional precision: their quality is ensured by a characterization validated using the HR-1 target with 1132 patches, delivering superior color fidelity compared to standard camera manufacturer profiles or generic Adobe profiles. The accuracy is further evidenced by the profiles’ performance on the FADGI Color SG target and their SMI (Metamerism Index) score of 79.
It’s true that my tests were conducted solely with Fuji RAW files and on Capture One, so my findings are specific to this case. However, I have no doubt that similar results would be achieved with other camera systems and Adobe profiles. In fact, the results would likely be even better, considering Adobe profiles are entirely generic and not tailored to specific cameras. Unfortunately, I cannot test this because the profiles must be purchased for your specific camera, and I only bought those for my GFX 100s.
Regarding the “Fuji magenta” issue, while Cobalt Image profiles don’t completely eliminate it, they significantly reduce it, making it much easier to correct in post-production with a simple tint adjustment rather than a more advanced and complex color correction.
In summary, regardless of the camera you use, I truly believe you can benefit from starting your post-production with Cobalt Image profiles to maximize what your sensor captures. As of today, the price of the Cobalt Image Base Pack ranges from €27 to €54, depending on your specific camera and post-production software, but I think it’s been the best photography investment I’ve made recently.
