Originally Posted by Joe Bloggs
What's the best/easiest way to find the true resolution of an image?
eg. if you have a 1920x1080 image, is there a simple way of determining whether that image was really up-scaled from a lower resolution or filtered in a way that reduces resolution or whether it really contains has the full resolution?
I know downscale-upscale is one method.
Suspect best and easiest would have to factor in expenses, too. Haven't performed any effective resolution testing, but just going by what's been posted on AVS previously...
Stacey mentioned in some '03 posts
using a Hewlett Packard spectrum analyzer to measure master tapes of several films and finding 800--1300 lines was typical effective resolution. (Assume as I outlined earlier
, with Joe Kane's comments, that measurement refers to equivalent lines per picture width, not height.) Hardware spectrum analyzers are costly. Presumably a master tape, say an older HD-D5 recording at ~270 Mbps compression, has minimal noise and edge processing, at least compared to later stages, and is 1080/24p (format resolution).
Without access to master tapes, perhaps the various PQ grades of Blu-rays would reveal differences in edge processing and noise (see the grading threads in the Blu-ray software section) as well as their typical maximum effective horizontal resolutions.
A less costly spectrum analysis on single detailed frames might be possible using dr1394's technique
with free/low-cost PC shareware. His two jpegs show a frequency-vs-contrast plot of a stadium crowd scene. "Basically, it's an FFT of the luma values from a 4:2:2 YCbCr [uncompressed] image," he wrote. So, presumably your PC needs the RAM to handle an uncompressed 24p frame converted to 4:2:2 YCbCr--if that's feasible from Blu-rays.
Easiest of all might be convincing an organization such as the United Nation's ITU to undertake measurements similar to their early 2000s international study of resolutions on film (static), then as projected on theater screens. Here's consultant Matt Cowan's 2002 pdf paper
on the results, using the ITU's data, and here's another pdf paper
detailing the ITU's measurement methods. -- John