Abstract: In a method of encoding/decoding a video sequence, which is composed of video frames, a first video frame and a second video frame are divided into a set of coding blocks, and at least one of the coding blocks of the first video frame is encoded by intra-coding. Then a first reliable region in the first video frame that comprises at least one intra-coded coding block and a second reliable region in the second video frame are determined. At least one coding block of the second reliable region is predicted from said first reliable region, and said second reliable region in said video sequence is encoded such that an inter-prediction information dependency on coding blocks outside said first reliable region is prevented.