JPEG – The Gory Details – Part 3 – A Xerox PDF

I obtained a Xerox Work Station PDF and many markers add up.

Update: Found another one with JPEG background and text monochrome bitmaps

Such as  the “YCbCr” tag is in the JPEG Comment…

However the Quantization tables are different for Cb and Cr so we have not found the complete answer but we have reasonable certainty that the document may have been generated on a Xerox Work Station

In addition, the document contains 1 JPEG background and several monochrome bitmaps.

I am feeling quite certain that we have found the workflow.

Strangely enough the following producer is identified Mac OS X 10.7.2 Quartz PDFContext

But I also found a PDF produced by Xerox Work Center which shows the Producer observed in the White House Document

Creator Tool Xerox WorkCentre 7232
Metadata Date 2011:04:04 16:02:23+02:00
Producer Mac OS X 10.6.7 Quartz PDFContext

Okay the clipping layer in the Xerox document is

0 0 612 792 re W n 

When saving with preview, it adds a border. 

12 12 588 768 re W

Obj 4

obj 4 0
 Type: 
 Referencing: 5 0 R
 Contains stream

  <<
    /Length 5 0 R
    /Filter /FlateDecode
  >>

 'q Q q 
0 0 612 792 re W n 
q 0 794.88 -614.4 0 613.2 -1.44 cm /Im1 Do Q 
0.9961 sc 
q 0 667.44 -510 0 522.24 94.56 cm /Im2 Do Q 
q 0 194.4 -251.52 0 595.92 94.56 cm /Im3 Do Q 
/Cs1 cs 0.1647 0.1882 0.1725 sc 
q 0 376.8 -521.28 0 527.76 374.88 cm /Im4 Do Q 
/Cs2 cs 0.9961 sc 
q 0 147.6 -171.36 0 540.72 348 cm /Im5 Do Q 
/Cs1 cs 0.1765 0.2275 0.2 sc 
q 0 186.72 -172.08 0 197.76 215.52 cm /Im6 Do Q 
/Cs2 cs 0.9961 sc 
q 0 177.36 -131.52 0 303.12 563.04 cm /Im7 Do Q 
q 0 49.68 -22.32 0 193.92 559.2 cm /Im8 Do Q 
q 0 20.16 -15.6 0 519.36 714.72 cm /Im9 Do Q 
q 0 49.92 -77.28 0 462 540 cm /Im10 Do Q 
/Cs1 cs 0.7412 0.5176 0.3412 sc 
q 0 54.48 -4.32 0 243.12 447.84 cm /Im11 Do Q 
/Cs2 cs 0.9961 sc 
q 0 23.76 -26.16 0 587.52 33.12 cm /Im12 Do Q 
/Cs1 cs 0.1765 0.2196 0.1961 sc 
q 0 51.12 -45.36 0 59.52 522.72 cm /Im13 Do Q 
0.1804 0.2431 0.2039 sc 
q 0 9.6 -8.64 0 374.16 372.96 cm /Im14 Do Q 
0.1686 0.2314 0.1922 sc 
q 0 9.12 -8.88 0 426.24 294.24 cm /Im15 Do Q Q'

23 thoughts on “JPEG – The Gory Details – Part 3 – A Xerox PDF

  1. I believe that folks working with the CCP like Garrett Papit have claimed that compression will not produce more than one 1 bit mask. You should send you findings to Mike Zullo.🙂

  2. What are do the references to the clipping mask and the white border look like in the Whitehouse pdf?

  3. Gorefan:

    q 18 14.40002 576 763.2 re W n

    There seems to be a floating point error, but basically this line says “Starting at (18, 14.4) draw a rectangle of width 576 and height 763.2, which will be a clipping mask.” Note that a clipping mask means that nothing outside of the mask gets displayed or printed – it does not change the page size. This gives us a border that is .25 inches on the sides and .2 inches top and bottom.

  4. Yep. If I look at the xerox scan, I see 0 0 xxx xxx re W n, the clipping mask is the size of the image, but if I print to PDF, a clipping mask is added such as 10 10 xxx xxx re W n

    That reminds me to decode Obj 4 which is basically a clipping mask, followed by images, rotated, scaled and translated, and some color settings for the monochrome bitmaps.

    Again this shows strong evidence against a Illustrator-like approach

  5. The Whitehouse pdf has both a white border and a clipping mask that are different sizes.

  6. The lines that have sc at the end are defining the color of the monochrome layers. If it’s a true grey (that is, the R, G, and B components are the same) a single number gets passed to the DeviceGrey colorspace, otherwise three values get passed to a DeviceRGB colorspace. The lines ending in cm /Im# Do Q are for adding the images. The six numbers before the cm are an array (it’s actually a 3×3 array, but the right column is always the same, so not specified). This can perform any or all of the following actions with the image: translate, rotate, scale, and skew. I’ll let NBC describe this in more detail, I think it’s what he is doing with his Part 3 of the LFBC Gory Details.

  7. The white border is actually contained in the JPEG image. (You can verify this by clicking on the image NBC provided). This was done by the scanner’s Edge Erase. It is not “in” the pdf itself.

  8. I’ll let NBC describe this in more detail, I think it’s what he is doing with his Part 3 of the LFBC Gory Details.

    Yep

  9. The white border is actually contained in the JPEG image. (You can verify this by clicking on the image NBC provided). This was done by the scanner’s Edge Erase. It is not “in” the pdf itself.

    10-4

  10. The original certified copy is 8.5×11 and the printed lfbc pdf with the white border is 8.5×11.

    Does edge erase actually erase the image or is the image scaled down to fit within the printable space of the printer (white border)?

  11. Edge erase, erases does not scale. It’s meant to remove punch hole marks and the kind.

  12. Compare the edges of the Guthrie photo with the edges of the lfbc pdf. Looks to me like they are the same.

  13. Count the repeats of the basketweave pattern. The Guthrie photo has more.

  14. “Count the repeats of the basketweave pattern. The Guthrie photo has more.”

    Did you release the clipping mask? Because I thought they counted the same.

  15. Yep, after the mask was released. I counted from several spots just to make sure I wasn’t getting some weird artifact from the picture.

  16. For example, right side of the rotated JPEG (if you look at the JPEG in the LFBC Gory Details Part 2 thread, the top side) the basketweave pattern looks like this, starting from the edge of the lines:

    =||=||=||
    ||=||=||=

    On the Guthrie photo, you get this:

    =||=||=||:
    ||=||=||=|

    There’s about a quarter of an additional repeat (or half a half-repeat, if that’s easier to visualize).

  17. Yeah, I see what you are saying. But isn’t that “quarter of an additional repeat” still smaller than the white border surrounding the pdf?

  18. Nope, according to my calipers it is almost exactly the amount that is missing.

  19. The JPEG is ~23.75 repeats wide, the Guthrie photo is ~24.25 repeats wide. That gives ~0.25 difference for the border.

  20. Producer does match. This does seem like a possibility. But why is the Creator Tool showing Xerox instead of Mac Preview, as in the WH PDF?

  21. Producer does match. This does seem like a possibility. But why is the Creator Tool showing Xerox instead of Mac Preview, as in the WH PDF?

    Good question and I addressed this somewhere🙂

    Imagine a pdf arrives in your Mac inbox. You have various ways of dealing with it

    1. Save attachment – Producer and Creator metadata remain Xerox Workcentre
    2. Open in Preview and save – Producer is replaced by Quartz, creator remains Xerox
    3. Open in preview and print to PDF – All metadata now matches

    Careful experimentation… It helps to have a Mac as well.

Comments are closed.