Compressing PDF documents to oblivion

[Wrichik Basu]

I am applying to several universities for Masters. The universities are asking me to compress all the required documents into a single 5 MB file.

I have to submit my 10th and 12th standard marksheets and gradesheets, along with their reverse side. These documents are around 5 MB each because they are digitally signed and downloaded from the official Govt. website. I distilled these using the Adobe Acrobat Print to PDF (I have the paid version of Adobe), and the total size decreased to 12 MB. Add to these the scans of my UG semester transcripts, which are 1.3 MB total. Upon that, a photo ID card, so a total of around 15 MB.

After combining all these documents into a single file, I tried compressing using the "Optimize" feature in Adobe. With the default settings, the size decreased by just 100 KB. If I force further compression by downscaling all images to 100 ppi, the scanned documents become illegible. I also tried to compress each file separately, but without success.

I tried a few online services as well, but in vain.

I contacted the admissions team regarding the issue. The reply was pretty straightforward: "You say you can't do it, but a lot of students are doing it. We are confident you can do it if you try hard enough. If you still can't, please don't apply."

Any assistance is appreciated.

 

[Baluncore]

Provide a link to documents that are available on the web.
You may have to OCR some of the documents.

 

[Borg]

Are you using ZIP tools like 7-Zip or WinZip?

 

[Wrichik Basu]

Provide a link to documents that are available on the web.
You may have to OCR some of the documents.

No links accepted. Even if they did, the documents have to be downloaded by logging in and are inaccessible to the public.

All documents have already been OCR'ed.

Are you using ZIP tools like 7-Zip or WinZip?

Zip files not accepted. Only PDF, max 5 MB.

 

[Borg]

If I force further compression by downscaling all images to 100 ppi, the scanned documents become illegible.

What types of images are these? If they are .bmp images that were generated from a Microsoft copy command, they could be 2 MB each (it drives me crazy when people send me massive emails like that). Changing the resolution won't affect the size of those very much because they have a bunch of OLE code associated with them. You might try capturing a screenshot of the images and saving them as jpeg or png files which would be much smaller. Just be sure to delete the old MS code.

 

[jedishrfu]

I think the best you can do is reduce the image resolution of each image to be readable but small. Replace scanned text with ocred text. Use jpgs where you can make them more lossy.

https://www.adobe.com/acrobat/hub/how-to/how-to-email-large-pdf-files

You may have to give up sending some docs to reduce the size. Also there may one or more images that causing the most issues that you could focus on.

 

[Wrichik Basu]

What types of images are these?

No idea. I downloaded the files as PDF, did not create them. I verified the signature on the documents, and the distilled them. Regarding downscaling images, I am talking about this option in Adobe: Optimize → Advanced Optimization.

 

[russ_watters]

All documents have already been OCR'ed.

I don't think that's true. I think when you print to pdf you rasterize the entire document, including the text. This makes a massive difference in file size. Like, an oder of magnitude or more. You need the text stored as text. The easy way to check is if the text in the docs is selectable.

Some of what you scanned, too - is it available in a digital to digital pdf from the original source? OCR is limited and graphics on original docs are often vector as well (so, much smaller).

Zip files not accepted. Only PDF, max 5 MB.

Zip files don't help here anyway - they do very little on pictures because they have to be lossless.

 

[Wrichik Basu]

I don't think that's true. When you print to pdf you rasterize the entire document, including the text. This makes a massive difference in file size. Like, an oder of magnitude or more. You need the text stored as text. The easy way to check is if the text in the docs is selectable.

The text is selectable. I re-did the OCR after distilling.

 

[hutchphd]

Can you indicate roughly how many pages of each kind of material the total package represents? PDF is usually pretty good,,,,something seems amiss. I would also look at which Docs are least compressed by the advanced compression routines and concentrate on those first.
Of course on a more humorous note it would seem prudent to compress the life out of any parts of your record you would like to de-emphasize. Maybe they will only look at the really good parts if the rest is nearly illegible!!

 

[Wrichik Basu]

Did it. Finally.

Converted each downloaded PDF to PNG with low quality setting. Converted those PNG files using an online service (https://png2pdf.com) and then combined those PDF files in Adobe. The final file size is 4116 KB.

And now, the university website is down.

Anyway, a pretty bad method, but the documents are still legible.