Solenaja
Silver Member | Редактировать | Профиль | Сообщение | Цитировать | Сообщить модератору
stumbler
Type of quantification:
H-263: this filter tends to smooth the image, this is the type of quantification used by the DivX codec as well.
MPEG: this filter, which is more powerful than the H-263 filter, will give you improved image quality but encoding will take a little longer.
BiDirectional Encoding:
There are three possible types of image within a DivX video stream. These images are referred to as "I-frames" (Intra), "P-frames" (Predicted) and "B-frames" (Bi-directional).
Before the introduction of DivX 5.0, images were only types I and P. I images are encoded using information from the image itself only.
They do not use information originating from other images (like time-based compression). An I image in concept is equivalent to encoding a single image using the JPEG compression algorithm. P (Predicted) images are deduced in advance and can refer either to an I image or a P image. They are encoded from the image preceding them. Regardless of the video sequence, a group of images will always comprise the same images several times. For instance, if you look at a person standing still, you will notice that that person will move very slightly with the background remaining identical for each image.
Note: There are generally up to 30 images in a single second of video.
Thus, instead of encoding each of these 30 images independently as you would do in an image with JPEG, you can make use of the redundancy in each image using P images. Summing up, a P image is the next image that determines where the image blocks have moved since the preceding image.
Now instead of performing spatial encoding (JPEG), a P image says to itself "Hey, the block in the previous image has moved to co-ordinates (X,Y)" which requires far less data than encoding each image spatially. In essence, we output the difference between each image, which is more efficient than outputting the original I image.
DivX Pro 5.0 provides the capability of using "B images" as well.
B images enable the DivX codec to predict future images by selecting the best of 2 predicted images instead of a single one. B images are encoded using not only images predicted in advance but also images already predicted which may be I or P images. Using B images reduces the amount of data required to encode an image, thus improving quality, especially in zones where moving objects reveal hidden areas.
----------
Могу помочь, но только своими знаниями и ... |
|
или 
