I could do so, if I could obtain the exact algorithm used by the ELA web site. Photoshop has a generic filter-building filter that is precisely designed so that a person can custom-craft something just like this. I will experiment and get back to you - after all the point here is to *reliably* detect fakery!lux wrote:I agree with this observation.Sukiari wrote:
Often times I see you and a few others post these ELA 'analyses' with no further comment, as if they prove anything. ANY JPG which is compressed to a moderately high level (such as you will find for distribution on the web) will show pretty colorful sparkles unless it is exceptionally low contrast, or it has muted color transitions.
I have seen ELAs posted here that purport to 'prove' fakery (and massive fakery is out there) without further comment, as if some pretty purple and pink and blue sparkles prove anything.
I wish you would. That would be huge.My challenge is to ascertain the algorithims used in ELA, use the Photoshop custom filter to build your own ELA, and take some good candidates for fakery from NASA's site in the original high-rez and high-bit depth, and ELA it.
I'm not saying that the ELA is always totally useless, but if an image is compressed, it will tend to show some of the pretty colors in the ELA, and as compression rate increases so do the pretty colors. Do not take my criticism of ELA as some sort of assertion that fakery is not rampant - I know it is!
*EDIT: Just so you know, I edited my previous posts to more accurately reflect my views, and I tagged my edits with a couple of asterisks so you can tell what I edited. I do not think it changes my viewpoint, but clarifies my thoughts.*