Abstract: The coding method of 3D videos has been greatly changed on the basis of high-efficiency video coding. Aiming at these changes, we propose a 3D video watermarking algorithm based on texture complexity and motion direction. First, we use the gray-level co-occurrence matrix and the gradient matrix respectively to analyze the texture complexity of the coding tree unit and the 8 CU×8 CU in the I frame and use the analytical result as the threshold to elect the 8 CU×8 CU with higher texture complexity in the P frame and B frame. Second, we select the embedded block whose inter-prediction mode is symmetric segmentation and combine the motion vector component of the co-located texture block to determine the embedded watermark. Finally, we modulate the search range of the current block or modify the regional optimal motion vector size to embed the watermark. The average embedding capacity of different video sequences reaches 613 bit/frame, and the average bit error rate under the re-encoding attack of different quantization parameters (QPs:25, 30, 35, 40) values is 13.31%. In addition, the average peak signal-to-noise ratio (PSNR) decreases by only 0.007 5 dB, which has almost no effect on the quality of the synthetic viewpoint. The above shows that the algorithm has good robustness.