Abstract: The invention discloses a method and apparatus for an improved digital film recorder. The improved recorder is capable of quantizing and recording high resolution images, in order to reduce recording time, without adding, or displaying appreciable quantization error artifacts. An image processor is provided which takes an image value at a select one of M input levels and quantizes the image value down to a select one of N output levels where M>N. The image processor of the invention then generates an error value which is related to the deviation between an idealized input signal and the quantized output signal. This error signal is then added back into the next input signal in order to smooth, or diffuse the error across a large area. Finally, in order to further reduce visible artifacts associated with the repetitive errors accompanying image areas having little or no change in image value, noise may be intentionally injected in order to break up any "bunching" of errors.