Following the dominant editorial line on the Web, that is, fiercely comparing products with each other in order to expose the shortcomings of each manufacturer, I have undertaken to evaluate two tools that, as in politics, fight to win the heart, because as for the head, we lost it a long time ago, of the users. Focusing exclusively on fidelity in color reproduction between both tools, since it is one of the most praised aspects of the emerging Capture One.
What do we compare?
Like any experiment, we are going to establish its foundations, what we also like to call “methodology,” that is, the rules of the game by which we will obtain two processed images in TIFF format ready to be analyzed against their colorimetric reference and against each other.
One of the problems of analyzing processing tools lies in “how we configure them“, since each of these tools has thousands of possible combinations, from the default one that appears as soon as the tool is “opened” to the most absolute button tweaking to achieve the desired look.
From my conception, it is only possible to compare these two tools if we try to reach a neutral state between them, a highly complex and controversial topic. Even so, I have adjusted each tool based on the following criteria:
- Perform a precise tonal adjustment, placing each sample of the tonal scale in its respective place, thus preventing exposure errors and curve manipulation transparently to the user by the manufacturer, a situation very typical of Adobe.
- A “white balance,” that is, seeking neutrality by eliminating possible scene dominants.
- Select a relevant camera profile. At this point, I considered it appropriate to perform two processes, one with the profile provided by the manufacturer (manufacturer profile or default profile), since this is a very common way of working, and another with a profile custom-made for the scene (custom profile) which, in the case of ACR, used a camera profile created with the X-Rite tool “Colorchecker Camera Calibration” and the ICC profile for Capture One was also created with this tool thanks to the new function we discussed in previous articles.
These adjustments are validated against other possible adjustments, ensuring that they are those that triggered the lowest delta-e error. In this way, we understand that at least, I have not found a more neutral or precise way to make such an adjustment with either tool. Understanding neutral as the absence of parameters that introduce aspects beyond the strictest colorimetric or tonal adjustment.
After these adjustments, the resulting image is compared with its colorimetric reference and the delta-e error is extracted for each sample, as well as compared between them to estimate the difference or distance between them, also expressed by the delta-e metric using the CIE76 equation, so we can easily understand the average error and the error for each sample.
Results
The results of such adjustments in terms of Delta-e are quite conclusive:
Capture One with default camera profile:
- Average ∆e: 6.4
- Maximum ∆e: 17.9 (Cyan)
- Minimum ∆e: 3 (Purple)
Capture One with custom camera profile:
- Average ∆e: 3.3
- Maximum ∆e: 6.8 (Red)
- Minimum ∆e: 1.7 (Foliage)
Adobe Camera Raw with default camera profile:
- Average ∆e: 3.3
- Maximum ∆e: 8.5 (Red)
- Minimum ∆e: 1.1 (Blue)
Adobe Camera Raw with custom camera profile:
- Average ∆e: 2.8
- Maximum ∆e: 6.4 (Cyan)
- Minimum ∆e: 1.08 (Dark Skin)
Difference between both processes using custom camera profile:
- Average ∆e: 1.9
- Maximum ∆e: 5.4 (Blue)
- Minimum ∆e: 0.4 (Foliage)
In the minimum errors I have omitted the densitometric scale, which accumulates less error due to barely having chromaticity, since I find it more relevant to know the errors in more critical colors.
In the following graphs, we can discern the detailed errors for each color and the differences between both charts, that is, the distance in terms of CIE76 between one chart and the other.



Conclusions
Based on the collected data, we conclude two quite evident situations: On the one hand, when using “default” color profiles, that is, those provided by the tool manufacturer, Adobe Camera Raw shows a significant advantage over Capture One, which will require us to create a custom camera profile if we need additional precision. It is worth remembering that this error has been accumulated for the particular profile of my camera (Nikon D610), which is not extensible to all camera models accumulating such error. Thus, the use of custom camera profiles is proposed as necessary in Capture One, and as we discussed in other articles, if conditions are adequate, in Adobe Camera Raw they do not make a significant difference.
However, after proper color management with Capture One, the average results are very close to Camera Raw, with an average error of 0.5 units between the averages of both processed images, so we are facing very similar images. Although, for example, if we look at specific colors like Red where Capture One accumulates a delta of 6.8 and Camera Raw 3.7, we are facing a difference of 6 units, which does constitute a clearly appreciable difference.
Visually, the differences between color samples can be appreciated, besides in the previous graphs, in the following visual comparison (left image), where the patches marked in yellow have a delta-e error between 1 and 2 units, or in other words, barely noticeable differences, and those marked in red have an error greater than 3 units, that is, particularly evident differences. The samples not marked have an error below 1, so they are considered equal.
Thus, we can observe how most samples are situated at a distance or delta-e error between images very low, that is, below 3, so we consider them barely noticeable color differences, which leaves us an average distance between images for CIE76 of about 1.9 delta-e units, a situation that evidences two processed images, with two tools, and with almost identical results.
However, even proposing a more or less sensible method, we are not free from the subjectivity of the parameters introduced in each tool, which do not exempt us from a certain arbitrariness in the results. Thus, if it is complex to establish a comparison from a neutral point of view, more or less easy to establish, proposing a comparison from another point of view is a highly arbitrary task lacking any utility.
How can we explain the miracle of Capture One in terms of color reproduction?
The answer to this statement cannot be found among the colorimetric data obviously, but rather based on the argumentum ad populum, that is, if many users say it is better, it probably is!
On the other hand, there is an important bias between what achieving a raw processing work that satisfies our expectations more or less immediately entails, versus a tool that implies an additional investment of time and complexity to reach the same point. That is, achieving certain results with Capture One does not imply that this is not possible with Camera Raw or Lightroom, but this is a mere ergonomic issue, and not so much colorimetric as many authors want to see.
