Hi! Back again with some more deinterlacing problems. I have been attempting to deinterlace some footage which I believe is MiniDV, but have struggled to get footage I am happy with. Applying Deinterlacing using QTGMC causes a jagged effect with the bars becoming bigger, so I attempted to halve the height of the image, deinterlace, and resize. However, doing this still leaves a lot of ghosting and jagged edges. The effect is particularly pronounced on this footage of a mouse running on a wheel. I was wondering if there was any way to get an improvement on this image. I am not expecting a perfectly clear image without ghosting, but anything to reduce that effect would be great! I have attached the original file to this post.
Those are not deinterlacing problems, but problems caused by someone taking interlaced content, resizing it (without taking into account that the content is interlaced) and then saving it as progresive content.
Don't think there is anything that really makes such content better.
Not creating such content with proper capturing and processing.
Other than that the usual approaches are:
a. resize to the original resolution and try to deinterlace
b. use convolution/blur/resize to get rid of the combing
both methods leave, depending on the source serve, ghosting, but there is no way around it.
Sometimes separating the fields and filtering them separately can also help, but in general there is no solution for this with conventional filtering.
=> Unless someone develops a machine learning method/filter for this I don't see any good way to 'restore' anything that is near the original.
