I looked at it with PDFDebugger... There's a background image that is 4MG compressed which is used by both pages, likely a background.

Then I looked at it with NOTEPAD++ and searched for /Length. This was possible because it didn't have compressed object streams.

There is a second large image with 5 MB. Then I ran a regular expression and got this:

Zeile  10698: <</BitsPerComponent 8/ColorSpace 9 0 R/Filter/FlateDecode/Height 3234/Intent/RelativeColorimetric/*Length 4784102*/Metadata 87 0 R/Name/X/Subtype/Image/Type/XObject/Width 2522>>stream     Zeile  49536: <</BitsPerComponent 8/ColorSpace/DeviceGray/DecodeParms<</BitsPerComponent 1/Colors 1/Columns 2447>>/Filter/FlateDecode/Height 3161/Intent/RelativeColorimetric/*Length 5345666*/Name/X/Subtype/Image/Type/XObject/Width 2447>>stream     Zeile 103697: <</BitsPerComponent 8/ColorSpace/DeviceGray/DecodeParms<</BitsPerComponent 1/Colors 1/Columns 2448>>/Filter/FlateDecode/Height 3161/Intent/RelativeColorimetric/*Length 5509296*/Name/X/Subtype/Image/Type/XObject/Width 2448>>stream

So you have 3 large images in total. I didn't bother researching where they are, your PDF is very nested. I suspect that these are more backgrounds, which contain these "dirty" lines.

Tilman

On 20.01.2025 16:56, Aaron Mulder wrote:
OK this is a long shot but... have a look at this PDF:

https://media.dndbeyond.com/compendium-images/phb/downloads/DnD_2024_Character-Sheet.pdf

It's 16 MB.  Eyeballing the thing, it doesn't seem like there's that much
complexity in there, though it does have a lot of background images or
textures.

Is there any way to inspect it for "what part of this is so huge?" and
possibly cut some things out to craft a version more like 1 MB?  You know,
if there are a few 4 MB images embedded I could just edit it to cut them
out, or whatever -- some loss of fanciness is OK to me.

I'm going to be creating a bunch of digital D&D character records with that
sheet and it seems like an epic waste of storage and bandwidth 😂

I looked at it in the PDF Debugger and couldn't find a way to identify all
the elements on the page, much less by "largest first", though I may have
missed something.

Thanks,
       Aaron

Reply via email to