Animation, there are two-dimensional animation and three-dimensional animation, of course, when it comes to animation now, it can no longer be simply said to be two-dimensional animation or three-dimensional animation. There are many mixed functions now, but for the compression direction, it is mainly because there are a large number of models in 3D animation, and effective coding optimization is carried out for these models.
Speaking of three dimensions, I must add a word to my mind: stereoscopic image. Let's talk about the evaluation method of stereoscopic image first. In a paper I read, the stereoscopic image was evaluated from two aspects: image clarity and stereoscopic effect. I feel that it is enough to evaluate stereoscopic images from these two aspects! Like one here!
Image clarity: As the name implies, it is the degree of image clarity, which is quantified from four aspects: nonlinear characteristics of brightness amplitude, multi-channel characteristics, sensitive characteristics of contrast and masking effect, and finally the definition index Q is obtained.
The nonlinear change of brightness amplitude refers to the logarithmic linear relationship between the brightness perceived by human eyes and the actual brightness of the image, that is, the absolute brightness of each pixel is imperceptible to human eyes, so the nonlinear change can simulate the actual feeling of human eyes on the pixel.
Multi-channel visual psychological and physiological experiments show that HVS has an independent spatial frequency processing unit, which transmits observed information through multiple parallel visual channels. A set of bandpass filters can be used to simulate the multichannel characteristics of human eyes.
Contrast sensitivity characteristics: Experiments show that human eyes are very interested in distinguishing contrast components. Combined with CSF, a minimum contrast recognizable function is constructed to define the minimum difference between two contrasts (from two images respectively) that can be perceived by human eyes.
Masking effect: a phenomenon that an originally perceptible stimulus becomes less perceptible because of the existence of another perceptual stimulus. The delay effect can be simulated by threshold.
Combining these four points, we can calculate the signal-to-noise ratio, and then get the evaluation index of image clarity.
Objective evaluation of stereoscopic effect: binocular parallax is the main external factor affecting people's strict stereoscopic vision. A large number of experiments show that the left and right viewpoint images contain a lot of similar information, and binocular parallax is the most obvious at the edge of the object. So calculate the absolute difference map between the original image and the distorted image, then reconstruct the error signal and quantify the evaluation index of stereoscopic impression!