Different Textile Materials as Light Shaping Attachments in Studio Photography and Their Influence on Colour Reproduction Različni tekstilni materiali kot nastavki za oblikovanje svetlobe v studijski fotografiji in njihov vpliv na reprodukcijo barv

The research focuses on the quality of colour reproduction when using different light sources, often used to illuminate scenes in a photo studio, and different types of fabrics as lighting shapers. With the latter, the light can be converted into softer and more diffuse light, but the properties of the fabrics used affect the colour impression and thus the quality of the reproduced colours. This was evaluated by analysing the colour differences which were calculated from the colorimetric values of the colour patches of the X-Rite ColorChecker Passport test chart. Test chart was photographed in a controlled environment and illuminated with different combinations of light sources and tested fabrics. The results confirmed that not all combinations of variables are suitable for use if the goal of the photograph is to achieve high quality colour reproduction.


Introduction
For photography, light is of crucial importance as it is needed for the ability to represent a subject/object. In a photo studio, the photographer himself controls the various properties of the light emitted by the lights on stage. These are usually the intensity and colour temperature, but we can also control other properties, such as softness and dispersion of the light beam, which is achieved by using light shaping attachments.
These attachments can be of various shapes, which affect the angle of the dispersed light. The consequence of the attachment's material is the quality of light, it being soft or hard [1]. Often used attachments are softboxes, which soften the original light by transmitting it through a white fabric. Studio lighting continues to evolve with technological progress. Today, xenon studio flashes (stroboscopic lights) and continuous LED lights are increasingly used in studio photography. Unlike halogen lamps, they do not generate high temperatures between 400 °C and 1000 °C [2]. By reducing the operating temperatures of the lights, it is no longer necessary to use special fabrics with a high melting point to soften the light. Instead, fabrics of organic origin can be used, whose structure begins to change at temperatures around 150 °C [3]. We can also use polymer fabrics with lower melting points, since the working temperatures of available lights are now lower and will not melt the fabrics. With a wider range of light shaping attachments, the amount of final light shapes also increases. By using light shaping attachments we transform also the colour of the original light emitted by the light source. Therefore, the light illuminating the scene cannot be described precisely. This research focuses on the quality of colour rendering and correlation between five different lighting conditions and the use of five different types of fabrics as light shaping attachments. The findings of this research will benefit the understanding and describing light used in researches, since no prior research has been found addressing the problem of the effects of fabrics as light shaping objects on the quality of colour reproduction. The study included three types of lights that are commonly used to illuminate the photographic scene, but are very different from each other, since they emit light of different intensity and colour temperature while producing different amounts of heat. A photographic studio flash that illuminates the scene with a xenon lamp was used. Its light flashes are extremely short therefore this type of lighting does not involve high operating temperatures. The second type of light source that does not produce high temperatures are LED lights (light-emitting diodes), with which colour temperature and light intensity can be adjusted. The third type of lighting is the illumination with a tungsten halogen lamp (hereinafter referred to as halogen lamp). This type of light source generates higher temperatures during illumination, so when staging the scene, we must pay attention to the distance of the flammable materials from the light head [2]. Textile fibres rarely melt, as their change at high temperatures does not necessarily mean a change in physical state from solid to liquid. Exposure to a suitable temperature ensures a change in the crystalline structure, which is also referred to as melting in the literature [4]. Synthetic fibres such as polyamide and polyester therefore have a melting point at temperatures between 223 °C and 270 °C, depending on the properties of the fibre polymers. Natural fibres, such as cotton, linen and wool, do not melt but carbonize at a certain temperature, i.e. such changes begin to show at temperatures around 150 °C [4,3]. In order to compare the influence of natural and synthetic fabrics on the quality of colour reproductions when used as light shaping attachments, five types of fabric were tested, made entirely from one of the five fibre types, cotton, linen, wool, polyamide or polyester. A series of natural and synthetic fibres has been chosen, since it has been studied that light reflects differently depending on the fibre structure [5]. A colour test chart has been illuminated using five different lighting conditions with alternating use of five fabrics as light shapers. Photographs of the scene were measured and the quality of colour reproduction assessed.

Experimental
For research purposes, a special cube that allows control over the tested light and limits the influence of its reflections has been created. With five different lights, the colour chart has been photographed and then the process repeated five more times, changing the fabric that served as the light shaping attachment. The combinations of variables and their labels are shown in Table 1. Each combination of variables was photographed three times, making a total of 90 photos. By adjusting the whiteness and brightness of the photos, the variables that would affect the colour differences caused by using different light shaping attachments, in this case fabrics, were eliminated. After these adjustments, the RGB values of all the colour patches were measured. The average values measured from three photos of the same combination of variables were converted to CIELAB values, and then the colour differences between the values without and with fabric as the light shaping attachment were calculated.

Scene
The test cube The photographs were taken in a photo studio where there is no influence of outside light. A 60 cm cube was made, lined on the inside with black felt (Figures 1 and 2), in order to prevent the reflectance of the light, which could affect the reproduction of the colour of the test chart. There are two round holes 20 cm in diameter in the centres of the two opposite sides of the cube, which serve as openings for illuminating the test chart ( Figure 1a). This ensures that the angle of the incident light on the test plate is always constant. Test chart was mounted on the holder in such a way that it was always evenly illuminated from the left and right at an angle of 45° (Figure 1b). The tested light source was 30 cm from the sides of the cube and positioned so that the centres of the light source were aligned with the centres of the round cut-outs on opposites sides. If a light shaping attachment was used, it was placed on the side of the outer part of the cube so that it completely and evenly covered the opening in the side (Figure 1c).
At the level of the illumination holes, there is an additional hole with a diameter of 10 cm which allows the camera to be mounted so that the lens reaches into the cube (Figure 1d). This limits the possibility of additional reflections in the lens and prevents the capture of light outside the test area.

Colour test chart
To evaluate the quality of colour reproduction with different light shaping attachments, an X-Rite ColorChecker Passport Photo 2 test chart was photographed, which consists of twenty-four colour patches from the ColorChecker Classic colour plate [6,7].
The RGB values of the colour patches on the test chart were measured with an X-Rite i1 Pro spectrophotometer and Argyll software, using the spotread command to perform the measurement. A two-degree observer and a standard D50 illumination were used.
The CIELAB values obtained were converted to RGB values of the sRGB colour space, whose white point is D65. This is the same conversion method that is stated in the X-Rite manufacturer's specification for the values in the test chart used [8].
The measured RGB values of the colour patches deviate from the values given in the specification of the test plate [8], which has no influence on the final results of the study, since the values under different lighting conditions were compared, and not the colour differences between reproduction and original. The measured and reference RGB values are shown in Table 2.

Photographic equipment and settings
Nikon D850 DSLR camera and a Nikkor AF-S 50 mm 1:1.4G lens with a HOYA Pro1Digital 58 mm MC UV(0) filter were used to photograph the ColorChecker Passport Photo 2 test chart. The filter serves only as a protection for the lens and does not affect the quality of the colour reproduction. Photographs were captured in the RAW output format (file type NEF) in order to obtain as much data as possible. The ISO value or sensor sensitivity setting was set at 100 for all the photographs in order to cause as little noise as possible. The white balance setting was changed according to exposure using camera presets and manual white balance settings as shown in Table 3 [9]. An aperture value of f/4.5 was used, due to the sharpness of the lens is at its best at this aperture and the degree of colour distortion (chromatic aberration) is among the lowest [10]. The shutter speed was adjusted according to other settings so that the photo exposition was ideal. Shutter speed was monitored using a camera light meter and histogram analysis of each photo. The exposure time required was influenced by the type of lighting, as not all lights provide the same intensity of light emitted, and by the tested fabric, as each allows a different amount of light to pass through, thus affecting the brightness of the scene.

Method of analysis
All photos were imported into Adobe Lightroom Classic CC 2019 where we used the Crop tool to crop the edges of the photographed test charts (Figure 3). This reduced the size of the photos per area, resulting in a smaller file size. However, the data important for analysis is retained. The dimensions of the final photos are approximately 1500 px × 1000 px (pixels), with minor size variations, and a resolution of 300 ppi (pixels per inch). For all cropped photos the whiteness was set to colour field no. 22, in order to colour match the photographs. The white balance adjustment was made in the largest possible range of 15 px × 15 px. The NEF photo format is no longer suitable for further processing of photos and the information they contain. Therefore, they were exported to a 16-bit TIFF format, which is the least lossy export format available. The exported photos were opened in the ImageJ program. Since the photos are 16-bit TIFF, the program displays them as grayscale values of each colour channel (red, green and blue). Therefore, they were converted to a 32-bit RGB colour image using the Image -Type -RGB Colour command [11,12]. It was also necessary to unify the brightness of the photos, as it was not possible to ensure exactly the same light intensities due to the lighting conditions. For the control area the colour field no. 22 was chosen, which has RGB values of 120/120/119. With the RGB Measure plug-in [13] average RGB values in the area of 100 px × 100 px in the colour field were read. Since the colour adjustment was already provided for when the white balance was set, the ratios of the read RGB values match, but in some photos, it was necessary to lighten or darken by adjusting the brightness. The measurement of the RGB value was repeated and the procedure was repeated until a sufficiently good adjustment was made to reach the reference RGB values of 120/120/119. The corresponding brightness was checked on all 90 photos and adjusted it if necessary. All 24 colour patches in 90 photos were then converted to RGB values, measuring areas of 100 px × 100 px. The same combination of variables was photographed three times, so in the next step we calculated the average RGB values of the individual colour fields in three photographs of the same combination of variables. CIELAB values were needed to calculate the colour differences and thus determine the impact of using fabrics as light shaping attachments. The conversion was first done between RGB and CIEXYZ then between CIEXYZ and CIELAB [14,15]. The average R, G and B values of the colour patches were normalized and the inverse sRGB compression function (Equation 1) was used to obtain the linear values of r, g and b. These were multiplied by the matrix M (Equation 2) with respect to equation (Equation 3). The matrix M considers the sRGB colour space and the white point D65 [16].  [18]. According to Equation 17 [19], the colour purity C was calculated for each colour field under different test conditions and then the differences between the illumination without and with fabrics as light shaping attachments were determined.
= 200( ' − ) ) (15) ∆Ε = $( ' − ) ) ) + ( ' − ) ) ) + ( ' − ) ) ) (16) where (L 1 , a 1 , b 1 ) are first and (L 2 , a 2 , b 2 ) second colour The colour differences for each colour field of the test chart were calculated, describing the influence of the use of different fabrics and lighting types on the colour reproduction. The colour difference value is described by the perception of a standard observer, where: 0 < ΔE * ab < 1 -the observer does not notice the difference between the colours, 1 < ΔE * ab < 2 -only an experienced observer notices the difference between the two colours, 2 < ΔE * ab < 3,5 -even an inexperienced observer notices the difference between the colours, 3,5 < ΔE * ab < 5 -an obvious difference between the colours is observed, 5 < ΔE * ab -the observer perceives the colour as two different [20]. To make the results easier to interpret, the mean values of the colour differences for each test combination from 24 colour differences of individual colour patches were calculated.  device. The normalized emission spectra of the emitted lights from halogen source (H) and three LED sources with colour temperatures 3230 K, 5000 K and 6260 K (L3, L5 and L6, respectively) are shown in Figure 4 as well. Spectral distributions differ depending on the used light source (Figure 4). A light source X with a xenon bulb emits the most uniform spectrum, because its maximum is indistinct in the blue part of the spectrum but shows less emitted light in the purple part. A light source H with a tungsten halogen bulb emits the greatest amount of light in the orange-red part of the spectrum, while LED light sources vary according to colour temperature. L3 shows a smaller fall in the blue part of the spectrum, while the curve rises sharply in the yellow-orange spectrum. This indicates the warm light from the light source. Standard light at 5000 K represents the light source L5, which has a maximum in the blue part of the spectrum and corresponds to the maximum of the light source L6. They differ from each other by the emitted light between the wavelengths 530 nm and 780 nm, with white light having higher values.

Fabrics properties
The values of fabrics' variables are shown in Table 4. In Figure 5 the images of fabrics are represented, showing the differences in fabric density and weave. The calculated colour properties in CIELAB are pictured in Figure 6, while in Figure 7 the differences can be observed visually.
The differences in the measured colorimetric values are mainly due to the yarns used, which are derived from the raw material composition. Twisted yarns are used in natural fiber fabrics, i.e. CO, LI and WO, while synthetic fabrics, i.e. PA and PES, are made from multifilament yarn. The uniformity of the yarn structure has a significant influence on the perception of light passing through it. In the case of wool and polyester fabrics, where the yarns used are the most voluminous ( Figure 5), the pores are the smallest and consequently the differences in light colour are the greatest (Figure 6). The passage of photons through the pores has no effect on their transformation, whereas the passage of photons through the yarn significantly transforms them, resulting in a greater change in colour ( Figure 6). The degree of whiteness varies significantly, where wool has the lowest level of whiteness and poliester the highest, the latter being a consequence of high amount of optical brightening agents.

Colour differences
Colour differences ΔE * ab of individual colour patches are presented in Figure 8. A red line indicates the value 2, which illustrates the boundary between the colour differences that can be recognized by any observer. Figure 9 shows the average values of all measurements according to the tested light sources and fabrics. The average colour differences (Figure 8) show that the greatest differences occur with the halogen  light source (marked H) and independently of the fabric used. The values are between 2.86 and 3.37, indicating a recognizable difference between the colours. The other four tested light sources give better average results, as the average colour differences according to Figure 8 are between 1.99 and 2.27. Xenon light (marked with an X) causes the smallest colour differences when using different fabrics, however, we observe a large difference in the illumination through polyester (PES). The LED light sources L3, L5 and L6 give similar results, but individual differences are noticeable, i.e. in the combinations L5/LI and L3/WO. The smallest colour differences were measured when using linen (LI) and polyamide (PA), followed by cotton (CO). Wool (WO) is the least suitable, due to the large colour difference reflected by all tested light sources. The same applies to PES, which however shows completely different deviation tendencies than WO under different light conditions. When comparing the colour differences in xenon flash illumination, it was noticed that PES causes a large colour difference, which is also perceived by an inexperienced observer. The use of wool also results in a colour difference greater than 2, but an excess of 0.02 is neglected. From Figure 10 we can see that the PES fabric in combination with xenon light X is different from the other four combinations of this light with other fabrics. It shows a greater colour shift towards green and blue and a lower brightness. The difference in colour hue can be related to the whiteness of the fabric (Figures 6 and 7), as the only fabric tested did not show any yellowing. The PES fabric is also sparsely weaved, which means that a lot of transmitted light is undeformed, so that the colour reproduction is affected by the cold light from light source X. ΔC * also shows the deviation of X/PES from other combinations with light X, as it gives the lowest value, which means that the colours reproduced are less pure. We conclude that the fabrics CO, LI, WO and PA in the tested composition and weave density are suitable for use as light shaping attachments when using xenon light.