In this mode, the decoder initially decodes only I frames. This mode is intended for quick generation of thumbnail images. When this method is called, the decoder outputs only frames that fall within the range of time stamps specified by the caller. For precise seeking, call the IMFTransform::SetOutputBounds method. To enable low latency mode, set the CODECAPI_AVLowLatencyMode attribute. However, the decoder buffers fewer samples in this mode, which can potentially lead to glitches, because the decoder does not decode as many frames in advance. It reduces start-up latency, so the decoder produces the first output sample sooner. This mode is appropriate for scenarios such as real-time communications. The decoder supports DirectX Video Acceleration (DXVA) using either Microsoft Direct3D 9 or Microsoft Direct3D 11. The maximum supported resolution is 1920 × 1088 pixels. The input to the decoder must be an elementary stream. The MPEG-2 video decoder exposes the following interfaces: MFVideoFormat_I420 MFVideoFormat_IYUV MFVideoFormat_NV12 MFVideoFormat_YUY2 MFVideoFormat_YV12 The decoder supports the following output types. The decoder supports the following input media types. The decoder supports MPEG-2 Simple and Main profile video (H.262, ISO/IEC 13818-2) and MPEG-1 video (ISO/IEC 11172-2). Acoustics, Speech, Signal Proc., 32, 6, 1243-1245, Dec.84.The MPEG-2 Video Decoder is a Media Foundation transform that decodes MPEG-1 and MPEG-2 video. "A New Algorithm to Compute the Discrete Cosine Transform," "A Fast Computational Algorithm for the Discrete Cosine Transform," IEEE Trans. Knuth, The Art of Computer Programming, vol.3, Addison-Wesley,Ħ W. "Fifth Generation Wafer Architecture", Prentice Hall, 1988.ĥ D.E. Alcatel-Telettra (Milan-IT), PrivateĤ M.J. ACM, 34,4,47-58, April 1991.Ģ Draft, Recommendation H.262 – ISO/IEC 13818-2, "Generic Coding of Moving Pictures and Associated Audio: Video," May 1994.ģ N. "The MPEG Video Compression Algorithm," Comm. Milan-IT) for their valuable support and cooperation.ġ D.J. Telettra, Milan-IT) and Massimo Martelli (Texas Instruments, Processing Laboratory: Nicola Griggio and Fabio Valente (Alcatel. The authors would like to thank the "old" students of the Image In any case, this analysis makes us more confident that the complete decoder will properly work up to 6 Mbit/s. At this bit-rate, all the constraints are satisfied with some margin with the exception of the PB couples, which do not satisfy the constraint T p+T B<2T q with "probability" about 1/4. The corresponding simulations show an increase of the maximum bit-rate, which, however, is not sufficient to go to 8 Mbit/s. Where T q=40 ms i.e., the frame period, while in the steady state, it must be: is transmitted as I 1, P 4, B 2, B 3, P 7, B 5, B 6 This implies that, at the start, the constraints for real time decoding become: This fact suggests there is room for some improvement by exploiting the difference between the transmission and visualization orders of the different pictures a sequence to be visualized as I 1, B 2, B 3, P 4, B 5, B 6, P 7. Bit-Rateįrom Figure 6b it results that there is a (small) margin for I and B pictures.
Actually, the preliminary version of the MVP decoder has been tested on I pictures and its performance was found to agree with expectations.įigure 7. It allows one to predict the decoder will properly operate at bit-rates up to about 6 Mbit/s. This is particularly true for P pictures, that at 8 Mbit/s (Figure 6a and b) become the only pictures whose decoding time is greater than 40 ms, the value required for real-time operation on a frame by frame basis.įigure 7 summarizes the expected per-frame decoding times vs. The figures clearly show that VLC decoding becomes more important in determining the processing time as the bit-rate increases.
0 kommentar(er)
