David Rudiak's Latest on the Ramey Memo

(Blogger’s Note: David Rudiak has supplied his analysis of what Kaleb wrote about his work on the Ramey Memo. I thought that David’s analysis would be of interest to those who visit here regularly. I also believe that it seems we have not progressed beyond where we were which is to say that there is no consensus. Following is what David wrote in response to Kaleb.)

1. Why the extremely low resolution (only about 11 pixels wide/character when our best scans are at 10 times that)?  There may be reasons to use reduced resolution because of limitations in software, time, or computer power (or increased resolution doesn't make it more readable), but resolution should definitely not be dropped _this_ much. Any final Ramey memo should be at least 3 or 4 times more resolution than these last results by Kaleb.

2.  As a corrollary of #1, such low resolution will _necessarily_ introduce artifacts by eliminating higher detail.  Thus, e.g., an "S" in the word "DISC" got blurred/smeared into the letter "O" (in Kaleb's most smeared version) and we are informed by him that it is 
definitely an "O" and the word definitely isn't "DISC".  Really?  Well what is it then?  Besides the fact that just about everybody else who has ever studied the memo now agrees the word is "DISC", is Kaleb aware of the historical context of this message?  1) How the word "disc" was a brand new use of the word to describe the strange flying objects being reported the previous 2 weeks, 2) As a result was frequently placed in "scare quotes" to indicate the new, unusual usage, and  3) Roswell base and Gen. Ramey both putting out press releases and _public_ statements actually using the word "disc" to describe what was found, transported to Fort Worth, and displayed in Ramey's office. Thus, no reason why it shouldn't also appear in an internal message about Roswell. I've attached an example newspaper front page with articles using DISC to describe Roswell, sometimes in quotes, sometimes not.

There is no question in my mind and most other readers of this message that the word is indeed "DISC" (in quotes). The proper way to try to interpret this message is to bring ALL information to bear on it, including historical context and linguistic analysis, and not just twiddle computer processing dials to the point that all sorts of artifacts can be introduced and muddy the waters.  If Kaleb wants to continue to claim that the word isn't "DISC" but "??O?", then he needs to provide a four-letter word there with the "O" that makes grammatical, semantic, and historical sense and isn't just a bunch of gibberish.

(Incidentally I've scientifically tested this word and many other words against the original teletype font using a cross-correlation tests, which creates a degree of match between the original letters and the degraded one.  In the case of "DISC", both the "I" and "S" tested as #1 probability letters, whereas "O" came out around #7 in the 3rd letter position.  So, it isn't just my say-so or others opinions that agree that the letter is really "S".

To illustrate the degradation of the image and how this has created an artifactual "O" in place of an "S", I've created a graphic with comments (2018 DISC image comparisons.jpg) showing two of Kalebs images plus an old one of mine off of one of my original 8x10" prints, and one of our 2015 ScanPro 3000 scans.  Notice the huge difference in image sizes and letter quality.  Going to such low resolutions is going backwards, not forwards.  Kaleb's alternate image which he says retains some of the fine grain is better, but still much poorer than images I was obtaining off of blow-up prints almost 20 years ago. (see graphic again).

3. What's the purpose of re-rotating the message when our best scans on the ScanPro 3000 had the message about as horizontal as overall possible?  Rotating the text away from horizontal only makes the message harder to read, not easier.

4.  Ditto all the other crap all around the message.  Remove it.  We don't need to see Gen. Ramey's thumb twice, rotated and unrotated, nor some mystery protractor dial. What's the point of leaving this stuff there?

5.  Why is there no correction for perspective, which squashes down the letters typically by about 50% in the vertical direction?  NOT doing this can change the appearance of letters into something else?  E.g., with the letters compressed, a capital "V" in normal proportion can be squashed down and appear as a small "r" to some people.  I'm also wondering if Kaleb is aware that the message is all-caps teletype font, and there will be no small "r"s or other small letters in the message, and it certainly isn't a mix of cap and small letters.

6.  As an example, I've taken Kaleb's latest rendition which he says retains some of the fine grain (and IMHO is thus more readable than the first examples where the letters are more smeared) and done the following:  1.  Rotated about 30 degrees back to the horizontal position; 2.  Cropped away all the unnecessary and distracting junk around the message;  3) Stretched the message 50% in the vertical direction to make the letters closer to the true proportions of teletype font; 4) Done a simple lightening function.  See second attachment.  Just very basic, simple stuff that took about a minute to do.  That's the basic format I want to see this message at in the end, not rotated obliquely, not squashed down, not with a lot of other unnecessary things there.  And also at much higher resolution.

7.  I've also attached one of my lighter version, full resolution attempts (from our ~10000 pixels wide scans) at flattening and straightening out the message), which (ideally) corrects for perspective distortion and letter distortion, makes all letters equally sized and spaced (as is true for impact printer, non-proportional font, which is what Teletype font is), straightens out all the lines and edges of the paper, and gets rid of the folds.  I used the morphing program Abrosoft 
Fantamorph 5, which is capable of handling such large files.(I did this work about a year and half ago and haven't been able to get back to it because of many pressing personal matters.  I thought everyone here was aware of it.)

As in ALL image processing, there are again artifacts introduced by this, particularly smearing of very distorted letters where there is a lot of perspective distortion, such as the top of the page above the top fold, the center of the page in the middle fold, and at the left where Ramey's thumb is warping the left margins of the paper (and shadow is making it very difficult to see exactly where the letters actually are). Also where the letters and words aren't clearly visible, it takes some educated guesswork as to where they are, which affects the warp model.  But overall, this I consider to be a much better rendition of the memo. Once I get a warp model, I consider good enough, it can be applied to all the various lightness levels of a particular sequence of Scanpro scans to try to extract more information.

E.g., a lighter scan might be used to try to make out anything in the shadows, whereas darker scans help bring out the more visible letters in the middle of the memo.  And you guys seem to have methods of mixing the various light and dark scans to try to suppress film grain.  This I think would be better done on the flattened, straightened images.  But don't overdo it. Making the message clearer to read is the primary goal, not getting rid of a maximum amount of grain (which can very easily be done just by cropping it out between lines and between words). Remove too much grain and letter detail is also going to get compromised.It is clear Kaleb is very dedicated and put in a lot of hard work into this, much more so than the average person.  But so far, I don't see the results getting better. Indeed, let's get back to basics, which should include higher resolutions, emphasis on letter clarity and retaining necessary detail, straightening things out, and keeping them properly proportioned.

It might also help me if Kaleb's methodology was provided.  Currently I don't have a clue how he is approaching this.