# Converting very short “motionless” video frames to single, better-resolved, images

I made a mistake with my newly bought compact camera, before a trip to Japan. Instead of taking single pictures, I ended up with a few short "MP4" video files (1 second, 2-3 frames). I could "easily enough" extract individual frames with ffmpeg, using help from:

I am unsatisfied with the quality of some frames, and as there is very little motion, I would like to restore or recreate a superresolution image from those two-three frames only, using the motion vector.

I would appreciate suggestions, software experience, etc. For readers interested in trying algorithms, I provided in the 22 MB file StackExchange-AVI2JPG.zip two examples with media information (obtained with MediaInfo) and camera reference (to avoid unwanted advertizing). One leaving Osaka, with a relatively nicely flat image:

the other from Nara, Todai-Ji:

where the wire fence is sharper, hence the individual frame quality is poorer. I enjoyed my trip, and would be glad to share crisper images of the first day (a view on the Fuji in the early for the finest eyes).

• Have you seen stackoverflow.com/q/10225403/4386900 ? – MBaz Mar 12 '19 at 16:12
• None, thanks for the hint. I'm going to check and add it – Laurent Duval Mar 12 '19 at 16:40
• Does the camera have Video Stabilization feature? Is it digital or mechanical? Does it have any information about the motion sensors? – Royi Jun 28 '20 at 18:00
• Yes is does. I think it was set, but I bought the camera just before the trip, and I did not read the manual soon enough. It is not appropriate to advertize, but let's say it is called TZOO (not O, but zeroes) by pnsnc – Laurent Duval Jun 29 '20 at 17:02
• @LaurentDuval, Probably better extract frames as a PNG (Or just arrays if using MATLAB for example) and then try Multiple Image Super Resolution. – Royi Oct 17 '20 at 8:23