For a continuum soft robot with an end camera, the visual servo of the end camera-robot system and the overall shape control of the robot must be considered when performing a task, due to various factors such as obstacle avoidance or security. To solve this problem, a hybrid eye-in-hand visual/shape control method for the soft robot is proposed. This method doesn't require the three-dimensional coordinates of the spatial feature points. The control objective can be achieved given the desired pixel coordinates of the feature points in the end camera plane and the desired shape of the soft robot. The kinematic model of the soft robot is established. Combined with the adaptive eye-in-hand visual control based on the depth-independent interactive matrix and the shape control for the soft robot, a hybrid control algorithm is proposed using the aforementioned kinematic model, and its stability is proved via Lyapunov stability theory. The convergence to the desired value of the pixel coordinates of the feature points in the end camera and the shape is verified through simulations and experiments.