Given your expertise, do you think the final result (ISP pic) is more or less likely to be close to the original video image that the still was created from. People have wondered why no one (supposedly) has identified this person and I'm curious just how true --considering the amount of distortion, compression, etc-that image is to the original if that makes sense.
My opinion is the images released by ISP are similar to the original video image data, except for the following differences:
- some amount of digital zoom (no new information, but larger image) has been done
- there appear to be some aliasing artifacts present which might be due to a print and re-scanning process
Performing a Fast Fourier Transform (FFT) shows the spatial frequencies where most of the image content is and can used to establish what the resolution of the original images from the video were.
Best I can tell, is the ISP image that appears rotated, and was released on Feb 15th was resized 3x larger than it's original resolution in the video. The 2nd image ISP released on Feb 15th appears to be 2x larger in size than it's original resolution in the video, based on looking at each image's FFT result.
Left hand side is RGB image from ISP, right hand side is FFT result of the Green band only.
If you understand FFT's, feel free to critique my analysis, I am open minded about it.
If you resize the images down by a factor of 3x and 2x respectively, you would likely have images that were very close to the original crops from the video that LE retrieved. They would look like this:
ISP image #1 downsampled 3X to native video resolution:
ISP image #2 downsampled 2X to native video resolution:
Sorry for the long post, but that is what you get from me when I start talking image stuff