The usage of SLAM technology in underwater environments presents unique challenges due to its dynamic nature. Because of the selective absorption of light by the sea, greater depths equate to a further loss of color. This thesis aims to study the creation of underwater 3D models using visual SLAM technology by splitting the image into separate R, G, B channels. In addition to studying the construction of 3D models, it is necessary to describe the main features of video SLAM technology and the advantages and disadvantages of using it in underwater environments. Due to the selective absorption of the sea, the red color is lost first (already at 5 meters), and as such, it cannot be used for creating 3D models. Both green and blue channels can be used, but blue is preferred. Because of the amount of blue colors under the sea, most details of the image are found in that channel. Compared to a full RGB image, the biggest drawback of the blue channel lies in consistency. The blue channel alone cannot achieve as much detail as a full RGB image, which corrupts the localization.