T 0708/21 (Matching foundation colours to a person's skin tone/CHANEL) 23-09-2024

1. The invention

The invention is about recommending make-up foundations that match a person's skin tone, [0002] and [0003] of the published application.

The skin tone is determined by averaging the lightness and hue of pixels in an image of the person's face, acquired in a controlled lighting environment. These average values are then correlated with the lightness and hue measurements of foundations stored in a database. The foundations with the closest match are presented to the person, [0017].

Each foundation's lightness and hue ("usage color map" in claim 1) are measured using a spectrophotometer under the same lighting conditions as those used to capture the person's face image. The measurements are taken on a film of the foundation, applied to a reference surface and dried at a temperature between 20°C and 45°C, [0038].

2. Second auxiliary request - inventive step

2.1 The Board finds it convenient to analyse the more specific second auxiliary request first, before turning to the higher-ranking requests. This request corresponds to the third auxiliary request in the decision under appeal.

2.2 It is common ground that D1 is a suitable starting point for assessing inventive step.

Similar to the present invention, D1 deals with recommending cosmetic products, such as foundations, that match a person's skin tone (e.g. abstract). Like claim 1, the method of D1 estimates the person's skin tone by analysing an image of their face, which may be taken in a controlled lighting environment (column 6, lines 54 to 59 and 65 to 67). In one embodiment (e.g. Figure 1), D1 suggests a two-step process. First, the estimated skin tone is assigned to a classification (or reference) colour, representing a subset of skin tones derived from test subjects (step 120, column 4, lines 35 to 44; column 19, lines 4 to 20). Second, the classification colour is correlated with one or more products (step 140, column 4, line 64 to column 5, line 9; column 18, lines 46 to 65). Another embodiment simplifies this to a single step, directly comparing the skin tone to cosmetic product colours (e.g. column 19, line 24 and column 20, lines 12 to 14). The Board takes this latter embodiment as basis for assessing inventive step.

2.3 The appellant argued that although D1 described acquiring an image of a person's face in a controlled lighting environment, it did so only in combination with product recommendations by a human consultant (column 8, line 63 to column 9, line 10). For automated recommendations, D1 required a "skin color estimate", which adjusted for variations in ambient lighting (e.g. column 18, third paragraph).

The Board acknowledges that D1 includes an embodiment involving recommendations by a human expert but disagrees that the use of a controlled lighting environment is limited to that embodiment. This is evident from the passages on column 6, lines 57 to 59 and column 11, lines 50 to 53, which state that embodiments of the invention may also be used in conjunction with controlled environments. Evidently, this possibility applies to both human and automated recommendations, as the lighting conditions affect only the skin tone estimate, not how it is compared to foundation shades. Using a controlled environment simply removes the need to correct for varying lighting conditions when determining the skin tone.

2.4 While the the Board agrees that D1 presents various methods for estimating skin tone and classification colours, as well as comparing the two, many of these methods are incompatible with the chosen embodiment. For example, capturing an image in a controlled lighting environment rules out skin tone estimates based on a "control reference color set" (column 18, lines 30 to 36), as this set is intended solely to adjust for varying lighting and image acquisition conditions. Furthermore, directly comparing the skin tone to product colours excludes aggregate estimates from test subjects (column 4, lines 42 to 44 and column 19, lines 17 to 22). Ruling out these incompatible alternatives in D1 logically leads to the use of individual colour values for both skin tone and product colours, such as those averaged from pixel colour descriptions (column 6, lines 65 to 68 and column 8, lines 17 to 25) in the HSV (hue-saturation-value) or RGB (red-green-blue) colour space (column 9, lines 40 to 45; column 17, lines 64 to 67).

2.5 Thus, the method of claim 1 differs from D1 in that:

1) it uses lightness and hue to represent skin tones and foundation shades;

2) the colour of a foundation is measured by spectrophotometry of a film spread on a reference medium and dried at a temperature between 20°C and 45°C, whereby the measurements are taken under the same lighting conditions as those used during the acquisition of the person's image;

3) matching foundation colours are those with a lightness difference equal to ± 10 and a hue difference between 0° and 90°.

2.6 The appellant argued that the distinguishing features interacted synergistically to solve the problem of providing simple and reliable foundation recommendations. Specifically, the use of hue and lightness (feature 1) simplified both the foundation colour measurements (feature 2) and the comparison between skin tones and foundation colours (feature 3). Moreover, measuring a foundation's colour under its conditions of use (feature 2) led to more precise shade measurements, which in turn improved the accuracy of the colour matching (feature 3).

The Board, however, considers that there is a lack of synergy between the differentiating features, apart from their inevitable chronological connection, namely that once a colour space has been chosen for representing the skin tone (feature 1), the relevant colour parameters have to be measured (feature 2), and the matching (feature 3) has to take place in this colour space. Analogously, the simplified measurements and processing are mere consequences of choosing a simplified colour space (feature 1). The remaining steps in features 2) and 3), e.g., spreading, drying, selecting shades within some range limits, even if contributing to simplicity or reliability individually, do not produce a combined effect greater than the sum of their individual contributions. It is not sufficient that the individual features solve the same technical problem or that their effects, though similar, merely add up to an increased but otherwise unchanged effect (Case Law of the Boards of Appeal, 10th edition, I.D.9.3.1).

Therefore, the Board considers that the differences may be assessed separately for inventive step, as the examining division did in the decision under appeal.

2.6.1 Concerning feature 1), lightness and hue define a 2D colour space that is simpler than the 3D colour spaces mentioned in D1 (HSV and RGB), yet still suitable for representing human skin tones and foundation shades (see also page 10, first paragraph of the published application). Therefore, the Board considers that the technical problem solved by this feature is to find a simpler colour space for representing skin tones and foundation shades.

D5, which also deals with matching foundation shades to skin tones, explicitly suggests a simplified 2D colour space as an alternative to other colour spaces like RGB and Munsell (column 9, line 49 to column 10, line 14). This space captures existing skin colours using two parameters: undertone and darkness. As illustrated in Figure 9 and explained in column 13, lines 39 to 41 and column 14, lines 42 to 57, these parameters correspond to hue (undertone) and lightness (darkness), with the undertone ranging from redder to yellower and the darkness from lighter to darker.

Hence, the skilled person seeking a simpler colour space would apply the teaching of D5 to the method of D1, and thereby arrive at feature 1) in an obvious manner.

2.6.2 The appellant argued that, in addition to providing a simplified colour scheme, using only lightness and hue to represent a skin tone improved the accuracy of the colour matching.

The appellant submitted that human skin appeared desaturated under illumination, while suitable foundations typically had higher saturation levels to create a more radiant effect. If saturation was taken into account in the matching process, the selected foundations would be less saturated, resulting in a duller appearance, which was undesirable. By ignoring the saturation from the colour matching, the invention eliminated unwanted noise and provided more relevant matches. During the oral proceedings, the appellant supported this argument with experimental data showing significant differences between the saturation levels of skin and foundation measurements, and the impact of saturation on the recommended products.

Since D5 did not mention that using only lightness and hue improved the foundation matching accuracy, the skilled person would not have considered this document when seeking a solution to the problem of providing a simple AND more accurate foundation matching.

2.6.3 However, according to the jurisprudence of the Boards of Appeal, an alleged technical effect cannot be considered for inventive step unless it is at least hinted at in the original application, or can be deduced by the skilled person from the application as filed considered in relation to the nearest prior art (Case Law of the Boards of Appeal, I.D.4.4.3 b, second paragraph). This is not the case here since the application as filed does not suggest that saturation is a noise factor and that ignoring it leads to more accurate results. It merely states that lightness and hue are chosen for simplicity, and that saturation can be disregarded since it is less important for foundation colours (page 10, top). Simplifying a process by removing irrelevant data does not imply or suggest an improvement in accuracy.

Therefore, the Board holds that the effect of improved accuracy cannot be taken into account in the assessment of inventive step.

2.6.4 Furthermore, it is established jurisprudence that if a claim achieves one effect derivable from the distinguishing features in an obvious manner, this cannot be redressed by an extra ("bonus") effect (see the Case Law of the Boards of Appeal, 10th edition, I.D.10.8).

As set out above, the Board considers that the skilled person looking for a simplified colour scheme for representing skin tones and foundation colours would arrive at a colour scheme comprising only lightness and hue in an obvious manner. Hence, the choice of these parameters does not involve an inventive step, regardless of any improved accuracy.

2.6.5 The appellant furthermore argued that the "undertone" in D5 was not equivalent to hue but also encompassed a notion of saturation. This is based on column 13, lines 49 to 51 of D5, which states "[s]kin tone 1Y07 would be the same depth in skin [as 5Y07] but the saturation in color is the least yellow and closer to neutral".

However, the Board notes that the term "saturation in color" in this passage is used to clarify the meaning of the first two characters in the colour code 1Y07. As explained earlier in the same paragraph, "the first two characters represent a color (Y corresponds to yellow) and how strongly hued the customer is (1 corresponds to very light hue)". Thus, in the Board's view, "saturation in color" refers informally to the visible prominence of a particular hue, e.g., whether it is more yellow or closer to neutral, rather than saturation in the technical sense used in colour space measurements. This interpretation aligns with the explanation of "undertone" in Figure 9 in column 14, lines 42 to 57.

2.6.6 Turning to feature 2), measuring a foundation's colour by spreading a film, drying it, and then using a spectrophotometer, is independent of the chosen colour space (i.e. lightness and hue) or the colour matching algorithm (features 1 and 3). The Board sees no technical effect achieved by this features beyond finding a suitable method for measuring the colour of a foundation. Therefore, the technical problem solved is how to measure the colour of a foundation.

The Board agrees with the examining division that the claimed measurement process is obvious in view of D2. D2 teaches spreading a foundation on a substrate ([0035]) and measuring its colour with a spectrophotometer under controlled lighting conditions (claim 2, [0061]). Although D2 does not state that the foundation should dry before measurement, this is already implied by D1, which mentions that the colour of a wet foundation may differ significantly from that of the dried product on the skin (column 1, lines 65 to 67). The claimed temperature range of 20°C and 45°C merely reflects typical room temperature, so, in the Board's view, this is merely common sense (see also the decision under appeal, point 2.10).

2.6.7 Finally, feature 3) addresses the problem of finding a suitable match between the person's skin tone and available foundations. D1 already teaches evaluating the similarity between a skin tone and foundation shades by calculating the differences in their colour values (e.g. column 20, lines 33 to 36). Since exact matches are rarely possible, the skilled person would establish acceptable ranges for these differences. The appellant has not demonstrated any specific technical effect resulting from the claimed limits of ±10 for the lightness difference and 0° to 90° for the hue difference, and the Board cannot identify any either. Therefore, the Board concludes that these ranges would have been obvious and, hence, feature 3) does not involve an inventive step.

2.7 For the reasons set out above, the Board judges that claim 1 of the second auxiliary request does not involve an inventive step (Article 56 EPC).

3. Since claim 1 of the main, first, and forth auxiliary requests is broader than claim 1 of the second auxiliary request, it lacks inventive step for the same reasons (Article 56 EPC).

4. Finally, the Board considers that the additional features in claim 1 of the third auxiliary request are already implicit in claim 1 of the second auxiliary request. The appellant did not argue that the third auxiliary request added any features of inventive significance. Therefore, claim 1 of the third auxiliary request lacks an inventive step (Article 56 EPC) for the same reasons.

5. As none of the appellant's requests is allowable, the appeal has to be dismissed.