A method and system are provided in which a current block of pixels of a video frame may be encoded and a direction associated with a next block of pixels of the video frame may be estimated. The direction estimation of the next block may be concurrent with the encoding of the current block. The current block and the next block may each be a sub-block in the same macroblock of the video frame. The current block and the next block may both be 4×4 or 8×8 pixel blocks. The encoding may comprise predicting values for the current block based on an estimated direction associated with the current block and values of previously reconstructed pixels adjacent to the current block. The encoding may comprise the generation of reconstructed values of the current block of pixels that may be fed back for the direction estimation associated with the next block of pixels.