Abstract: During encoding, blank frames may be substituted for text images in video content, encoding the blank frames instead of the text images. The text images may be any kind of text images such as opening credits, ending credits, and so on. The selection of text images to substitute may be performed by optical character recognition, user selection and so on. The text associated with the text images may be embedded in the encoded video. If the text is already embedded, an indicator of the location may be added. If not, the text may be derived from the text image using optical character recognition and then embedded. When decoded, the encoded video may be analyzed to determine whether blank frames were substituted for text images. Embedded text associated with the text images may then be located, obtained, and added to the decoded video. Thus, the original text images are essentially reconstructed.