asked on

Screen Capture to PDF - Inverse Order

Long meetings online. Many documents, some not shared, or surprised to see online / not announced with meeting notes.

Need to capture the pages as they are shown on the screen, then easily / automatically reassemble them as a PDF.

Key thing would be the inverse order nature of this task - such that while page 1 is saved first and, maybe, page 15 is saved last - and is thus "newer" - the reassembled file would need to come together with the oldest image first. In reverse order, that is.

Is there anything similar to this out there that Experts use in this situation?

Sincerely,

OT

David Favor

Can this be done? Yes.

Is is practical... maybe... many not...

Here are the steps to do this... at least how I'd do this if I had to...

1) Whatever tool was being used for the meeting, size the tool so the entire window is exactly 1080p (1920x1080) or 2160p or 4K/8K.

2) After meeting, export the file in the exact same format as file was recorded.

3) Steps #1 + #2 are required, if you expect to read any small text without edge distortion.

Generally I record in 1080p.

4) During export you can get away with setting an 1 FPS (frames/second) as your export frame rate, which only works for meetings where documents are actually fully visible for 1 second.

5) Now use ffmpeg to break the final video into 1x image/frame, likely using PNG images for this... maybe JPEG... you'll have to experiment.

6) Now you have your entire meeting as a set of individual frames, then you can also write code which checks the visual quality of images finding which images are of highest quality.

7) At this point you'll use some sort of selection mechanism, like file preview in Windows or Mac where a human scans images to determine when an image changes.

Note: This can also be done by image diffing, using a tool like ImageMagick, where you consider say... a 75% visual match to be the same frame... and this code step will be complex to automate. For infrequent meetings, use human eyeball scanning. For many meetings all day, every day... best write this code.

8) Now take the images you've selected in #7, then stitch them together in reverse timestamp order using Poppler tools.

9) 1x additional step you might take, which will likely be highly useful + eat up massive CPU cycles... will be to run tesseract-ocr on each image, creating a related human readable text component.

10) If you do #9, you'll have an additional PDF stitching step, where you stitch together all the text components, into 1x text component/page of your PDF, then inject the text component first, then inject all your page images in reverse order.

Note: Adding the text components is highly useful, as this allows PDFs to be text searchable, so if you have 10,000s meetings where you require finding a single document, searching is trivial... compared to eyeball review of 10,000s of meetings trying to find 1x PDF page of interest.

David Favor

All this is easy to conceptionalize.

And code is straight forward to write.

Likely this will require significant budget + time to get working, then maintain over time.

oaktrees

ASKER

Hi David,

Outstanding! Yep - that's what I was thinking of! Was hoping there was something on the market like this already? Looks like the only way to get it done is to build it?

Sincerely,

OT

Joe Winograd

Hi OT,

Need to capture the pages as they are shown on the screen...

When I first read that, my thought was that you want to hit a key or key-combo (a user-defined hotkey, a function key, the PrintScreen key, whatever) when you see something on the screen of interest (one of the "Many documents"). After capturing such pages (by manually hitting the key/key-combo each time), you then want to put them together in chronological order, i.e., first captured to last captured, into a single PDF. Is that an acceptable method or do you want to stick with the idea that David posted, which is to capture the entire session with video recording software and then extract/process the frames of the video afterwards? I'm supposing the latter, since you said, "Outstanding! Yep - that's what I was thinking of!", but wanted to check in with you on it. Regards, Joe

oaktrees

ASKER

Dr. J!

GREEEEETINGS, Good Sir!
You wrote:

the PrintScreen key, whatever) when you see something on the screen of interest (one of the "Many documents"). After capturing such pages (by manually hitting the key/key-combo each time), you then want to put them together in chronological order, i.e., first captured to last captured, into a single PDF.

That would work perfectly, too as you're right - usually it's more of a page, wait-a-while, next page situation.

Sincerely,

OT

Joe Winograd

Do you participate in the meetings via an app/program or a browser? In either case, which one? And what OS?

SOLUTION