Graphics and Media Lab

MSU Graphics & Media Lab (Video Group)

This filter magnifies video sequence while saving and improving details and reducing artifacts. Its work is similar to MSU Video2Photo filter effect, applied to entire video sequence.

The main problems of current approach:

• Fixed zooming factor. Later it will be generalized.
• No scene change detection.

Examples

The primary examples can be seen on the MSU Video2Photo filter page, here only some video sequences are showed. The DivX 4.12 codec in the Quality-based mode was used for compression.
Since the output resolution of video is very large only parts of sequences are shown.

The first example is from the "foreman" sequence:

AviSynth Bicubic	AviSynth Lanczos4
Topaz Enhance SR	Proposed SR

The next example is from the "news" sequence:

source frame of 'news' movie and frame processed with MSU Super-Resolution filter

A frame of the "toy" sequence:

source frame of ''toy'' movie and frame processed with MSU Super-Resolution filter

Download

"foreman", source video and video processed with MSU Super-Resolution filter (340 KB, AVI)
"news", source video and video processed with MSU Super-Resolution filter ( 974 KB, AVI)
"toy", source video and video processed with MSU Super-Resolution filter ( 660 KB, AVI)

Team

• Dmitriy S. Vatolin, Lomonosov Moscow State University
• Artem Titarenko