Removing banding noise from video still
October 3, 2014 11:29 PM Subscribe
I have a video recording of a computer screen, and so the resulting video includes some wavy banding artifacts (the kind you often see when you film a monitor or TV). Small sample image. What is this artifact/noise called, and is there any way to remove it?
Actually, in the end I don't need a video, I only need a still image captured from the video. So I hoped there would be some way to remove it with Gimp. I have followed a few guides on removing Moire patterns, but I don't think they work for this kind of artifact. If anyone knows how to remove this from still images, or video, I'd really appreciate some help. Thanks!
Actually, in the end I don't need a video, I only need a still image captured from the video. So I hoped there would be some way to remove it with Gimp. I have followed a few guides on removing Moire patterns, but I don't think they work for this kind of artifact. If anyone knows how to remove this from still images, or video, I'd really appreciate some help. Thanks!
Response by poster: Sorry I should have mentioned that I'm not able to reshoot the video, unfortunately. PM'd Gyan to see if they're able to work any magic.
posted by theyexpectresults at 2:38 AM on October 4, 2014
posted by theyexpectresults at 2:38 AM on October 4, 2014
That type of noise is also called moiré. In your case, the two conflicting grids are the RGB sub-pixel grid of your LCD and the pixel array of your camera, so it appears as color distortion.
There's no solution that's both easy and universal. There's a possibility that you could recover a clean still by aligning and averaging multiple adjacent frames, but that would only work if the two grids were in motion relative to one another. But I infer you were shooting a screen from a stable tripod?
I'd be curious to see what the Fourier spectrum looks like. Any chance you'd be willing to share a full frame?
posted by rlk at 4:58 AM on October 4, 2014
There's no solution that's both easy and universal. There's a possibility that you could recover a clean still by aligning and averaging multiple adjacent frames, but that would only work if the two grids were in motion relative to one another. But I infer you were shooting a screen from a stable tripod?
I'd be curious to see what the Fourier spectrum looks like. Any chance you'd be willing to share a full frame?
posted by rlk at 4:58 AM on October 4, 2014
Response by poster: Thanks for the help so far. I've given up on fixing this image anytime soon, since it seems to be more trouble than it's worth, and I can make do with the original image. But if anyone is interested (for sport or academic curiousity) I can PM them the still image.
posted by theyexpectresults at 7:47 AM on October 4, 2014
posted by theyexpectresults at 7:47 AM on October 4, 2014
This thread is closed to new comments.
If you have to work with the recorded video, you will have to retouch the image after denoising to get reasonable colour.
If you can post here or message me the full image you need to extract, I'll see what can work.
posted by Gyan at 1:46 AM on October 4, 2014