That's right!
I recorded the videos on greenscreen, rendered them into image sequences, resized them and cut characters out from photos frame by frame. Each enemy has its own 4 animations - walking, getting shot, attacking and dying. To keep the illusion of the video being played on engine, I never skipped a frame from the original video. This in turn left me with A LOT of sprites and time spent in photo editing software. For example, basic enemy (such as The Masked Mutant) is made out of 237 frames in total.