It sounds like you're scanning from a photo. How are you making the determination that the grains are too small to be relevant? How large is the Kenny Baker mask in the photo - 1 square inch? If you're scanning it at high resolution to where photo grain doesn't matter (ah, there's another thing: the grain of the photo paper, not just the film) then why does it look like you enlarged a 72 dot-per-inch image in your 4-image comparison?
So let's take the Baker vs Corbis shot. You have 3 yellow rectangles. T he upper left rectangle has one spec that appears to line up. You have a lower rectangle with to red lines, but the Corbis details are barely visible. Then the upper right rectangle with the larger thing - the Baker shot shows two bright dots but the Corbis shows something roughly like the greek symbol "Pi". I'd like to be able to agree with you, as I'd love to say beyond doubt that the Kenny mask is THE screenused ANH and not just a copy.
But saying "... a scanner picked up exactly the same artifacts as an HD screen capture in both something as large as the grill and as small as those details I point out on the cheek, or the detail on the neck for that matter. " Well, the neck is a larger detail and easier to match up, but the cheek comparison isn't a slam slam dunk yet for the following reasons:
1. The Corbis shot's cheek surface is far garner than the Baker shot. Have you eliminated the false positives? There are white specks on the Corbis that do not appear on the Baker s hot.
2. Why does the Corbis shot cheek corner look cut off and the Baker shot looks sharp?
3. If a few white specs seem to line up, how do you explain all the white specks that don't.
In other words - especially #3 - you have to eliminate the false positives before stating something so confidently as "proof".
Also, the production of ESB should have been after Chronicles, correct? We know that during the Chinese Theater and the Chronicles photoshoot, the mask was repainted.
That would mean the cheek surface would have been altered.
"If anything compression just smooths out the details.... "
Sorry, it doesn't. I wish it were that simple, but unfortunately that's not accurate. If you're fortunate enough to have a TV that has artifact-removing technology by adding a gaussian blur, then more power to you!
I think you'd find the following article quite informative in this matter. It shows that MPEG-2 compression does not necessarily smoothe out the details , but can add noise.
High Definition Blog » HDTV Artifacts
On the topic of noise, have you familiarized yourself with the various scanning modes of your bed scanner? Some software may add dithering (noise) to improve some images. My page scanner is a LED-based flatbed. It's a $100-120 cheapie but it has different modes that each yield different results, and an advanced mode that allows a combination of settings until you feel you have the best image possible. However, going from optical to digital still relies on software to interpret data and to present an image. Each time I do a scan and zoom in, the noise is different.
"The Corbis photos were taken with Tri-X pan black and white film which even though it is ASA 400 isn't going to be an issue with flash. When you get to ASA 1600 or more in film then it could be an issue. It is no different than the size or density of pixels on CCD chips. "
I've done B&W photography 20 years ago and did the whole dark room and developing my own prints thing. If you're scanning at 72 dots per inch, you may not notice much. If you scan at 300 dots per inch, if you're saying you won't see the grain, then here is a great example:
http://elliottmccrory.com/weblog/wp-content/uploads/2010/01/aug_1993_047_handsonly1.jpg
Here the person used a Canon F1. You can see the grain of the ASA 400 film. Lots of white dots too.
Thomas, I really want to, but I don't see unmistakable coincidence just yet. You've made a good effort (so has Carsten). I encourage you to keep looking.