Masoomeh Golabkesh is doing her master’s degree in graphic communication at Clemson University. She also has a master’s in graphic design and a bachelor’s degree in the same major. She has experience working as an art director in a commercial company. Her past research has focused on Measuring the visual learning capacity of people over 60 years old: A comparison study between static and animated infographics. She also worked on Color Correction in Video: Determining the Best Combination of Equipment and Settings to Capture Brand Color. She is interested in research topics related to branding, Photography and videography technology, and Web design.

Color Correction In Video: 
Determining the Best Combination of Equipment 
and Settings to Capture Brand Color

Masy Golabkesh and Dr. Erica Walker, Clemson University

Color is an essential tool that delivers a brand image with more substantial persuasive power than shape, and it appeals to emotions rather than rationality. Color is also closely related to consumer behavior. As a result, color plays an influential communication role in situations where direct and specific verbal expression is complex. Moreover, color serves as a visual language that provides information to the sensory mode – sight – with the highest information capacity of the five human senses (Rubio, Oubiña, Gómez-Suárez, 2015).

In live sports event broadcasting, the team color can look different from the intended color when displayed on the screen. Accurately recreating the team brand color is very important. Companies spend a lot of money on developing and protecting their brand color pallet. They choose specific colors for their brand to identify between crowded marketplaces, and they want to see accurate brand colors portrayed in live events. Sport team fans recognize their team color and expect to see the consistent color in different games (Mayes, Lineberger, Mayer, Sanborn, Smith & Walker, 2021). Displaying the wrong color on screens may confuse fans.

Multiple variables play a role in capturing and displaying colors, such as camera body, lens, camera settings, ambient light, and camera positions to the field and lights. Current systems of ensuring the accuracy of the brand color of sports teams onscreen require manual adjustments of the camera footage in real-time or in post-production. To reduce this manual effort, we need to understand the variables that impact the colors.

In this study, we test the camera-level capture of Clemson orange by changing multiple variables. We have examined the following variables in this experiment: five camera models, six lenses, five camera positions with the player, two sets of camera Picture Profile settings, and two types of lighting (inside lighting using LED lights and outside lighting with ambient light). This resulted in 600 unique samples. These camera, settings, and lens combinations were chosen to mimic the on-field situations encountered by content creators for Clemson Athletics.

We shot 4-second clips of the still life which included a helmet, a jersey, a Pantone book open to uncoated Pantone 152 U and coated Pantone 1595 C, and an X-Rite Color Checker for reference with the different combinations of camera bodies, lenses, positions, and lighting. Using a program written in Python leveraging the strengths of Artificial Intelligence and previous work in this area on ColorNet (Walker, Smith, Mayes, Lineberger, Mayer & Sanborne, 2020), we sampled frames from the footage to create the data set. In the next step, we measure pixels from each frame. To increase the pixel sampling randomness, we extract multiple pixels in different areas such as shadow, mid-tone, and highlight areas of the frame. Then we calculated the RGB values for each pixel and averaged the RGBs values for each clip.

Then we compared the produced RGB values with Clemson brand orange RGB values in the different permutations to better understand the variables that impact brand color capture on video and determine the ideal settings and situations to reproduce brand color accurately at the camera level. We used the T-test to compare produced RGB values with Clemson Brand orange RGB values and determined the difference between the two values. Knowing to use the proper parameter settings during camera capture for broadcast will help to reduce the timely post-production process and create accurate brand colors in real-time.

References

Rubio, N., Oubiña, J., & Gómez-Suárez, M. (2015). Understanding brand loyalty of the store brand’s customer base. The Journal of Product & Brand Management, 24(7), 679–692. https://doi.org/10.1108/JPBM-03-2015-0822

Mayes, E., Lineberger, J.P., Mayer, M., Sanborne, A., Smith, D.H., & Walker, E.B., “Automated Brand Color Accuracy for Real-Time Video.” SMPTE Motion Imaging Journal. https://ieeexplore.ieee.org/document/9395672

Walker, E.B., Smith, D.H., Mayes, E., Lineberger, J.P., Mayer, M., & Sanborne, A., “Consistent display of Clemson brand colors using artificial intelligence.” Technical Association of the Graphic Arts (TAGA), June 2020, Oklahoma City, OK.