This invention generates encoded data within a required buffer size with minimum deterioration of image quality by finely adjusting a code data size to be generated so that an encoding condition can be changed to one of multiple levels on the basis of a cumulative encoded data size which is being generated during an encoding process of one frame which forms a moving picture. To this end, a VBV buffer zone controller compares a cumulative encoded data size during encoding of one frame with a plurality of thresholds, and outputs the comparison result to a vector processing controller, coefficient processing controller, and quantization controller. The vector processing controller controls a motion detection/compensation unit to fix/not to fix vectors for motion compensation. The coefficient processing controller determines an area size of a plurality of DCT coefficients to be masked, which are obtained by a DCT unit. The quantization controller determines the quantization step size of a quantizer. These vector processing controller, coefficient processing controller, and quantization controller adjust an encoded data size to be generated in multiple levels.