Publisher image / filesize / workflow question(s)

[internetuser]

I'm trying to figure out the best way to get a lot of images into a publisher file for Amazon KDP. Getting 550 8.25x11 images into a 550 page document and getting under the 650mb file limit is tricky, and I'm not sure exactly how Publisher is working with my images. They are 2550x3375 jpegs (the math I've arrived at when adding a 0.125" bleed at 300dpi), and are ~550mb in total after exporting all 550 of them from DxO. When running them through Publisher, the final PDF size winds up being anywhere from 5GB down, depending on various settings, none of which make a lot of sense to me at the moment. 

I guess my first actionable question is this: is there some combination of export/settings in Publisher where I can simply export the jpegs from DxO PhotoLab precisely how I want them, and for the resulting Publisher PDF file size to remain whatever the sum total of the jpegs is? (Or as close as possible, accounting for some increase in size for PDF metadata, etc...) I'm really just trying to figure out a workflow where I can do most of the editing/filesize work outside of publisher, in terms of image-tweaking, and to cut out Publisher's jpeg compression. In short, I don't know where the file bloat is coming from, and I don't understand when or how 500mb of jpegs becomes 5GB of PDF. 

I hope this is making sense. I've been trying to get to the bottom of this for five days and my head is a bit scrambled.

Thanks in advance for any help or experience you might be able to bring to this.

Edited March 13, 2023 by internetuser

[Dan C]

Hi @internetuser,

Welcome to the Affinity Forums

15 hours ago, internetuser said:

is there some combination of export/settings in Publisher where I can simply export the jpegs from DxO PhotoLab precisely how I want them, and for the resulting Publisher PDF file size to remain whatever the sum total of the jpegs is?

Unfortunately this isn't really how Publisher works in terms of composing documents and exporting to PDF, the 'sum total' of your image files is not directly translated into your Affinity document, as the app has to read, interpret and then export your images - it cannot simply 'passthrough' the file container from the image you have exported from DxO, meaning the size of the Affinity Document / PDF cannot be directly calculated based on the size of the input images.

15 hours ago, internetuser said:

Getting 550 8.25x11 images into a 550 page document and getting under the 650mb file limit is tricky, and I'm not sure exactly how Publisher is working with my images. They are 2550x3375 jpegs (the math I've arrived at when adding a 0.125" bleed at 300dpi) (...) and to cut out Publisher's jpeg compression

Your images at 2550x3375 means each image requires 8,606,250 pixels or 8.6MP. 550 images at this size equates to 4,733,437,500 pixels, or 4,733MP.

Exporting these images to PDF, without JPEG compression, would require multiple gigabytes of space. Although not entirely accurate, the following is an estimate based on these pixel sizes -

15 hours ago, internetuser said:

In short, I don't know where the file bloat is coming from, and I don't understand when or how 500mb of jpegs becomes 5GB of PDF. 

Can you please provide a screenshot of your PDF export settings from within Affinity?

Is your Affinity Document at 300dpi, as well as the images - or is there a difference between the image DPI or colour space and the Affinity document DPI/colour space?

[internetuser]

3 hours ago, Dan C said:

Your images at 2550x3375 means each image requires 8,606,250 pixels or 8.6MP. 550 images at this size equates to 4,733,437,500 pixels, or 4,733MP.

Good morning, Dan. Thank you for the welcome, and thank you for this thoughtful and considered reply. I can now see that my problem was a total lack of a correct conceptual framework for what a PDF is, and in thinking of it as a image container instead of as an object that has to be rendered, so you have precisely answered the question of "where the bloat comes from" and put to rest the question of whether exporting without JPEG is optional. The document is 300dpi, and the images are being exported from DxO in sRBG and 300DPI. I can hardly expect you to teach me the entirety of how images and publishing works, so I will run some further tests (at a scale smaller than 550 pages) to see if I can't get to the bottom of this. The "workflow" involves a bunch of moving and experimental parts (Stable Diffusion -> ESRGAN -> PNG-to-JPEG conversion -> DxO -> Publisher) and my incredible ignorance of publishing/image fundamentals probably exceeds your paygrade.  Thank you again. 

[Dan C]

No problem at all, more than happy to help

46 minutes ago, internetuser said:

The document is 300dpi, and the images are being exported from DxO in sRBG and 300DPI

Thanks for confirming that for me, that sounds like the best practice as the less an image has to be 'converted' when exporting, the smaller the exported file size should be (in theory!), though one easy way to reduce a PDF size is to reduce the Raster DPI value when exporting.

When exporting your PDF document, I'd recommend trying the '(digital - small size)' preset as this will down sample the images in the PDF, hopefully producing a smaller PDF file.
In this dialog there is a 'JPEG Compression' slider, the lover this value the less space the images will take up in the PDF file, but equally the lower the value the less perceived quality will be retained in your image, so you may want to experiment with multiple values until you're happy with the compromise between quality & file size.

You can also see an estimated file size in the Export dialog, meaning you shouldn't need to physically export the PDF each time you change these values to understand the size the PDF file will be once created.

I hope this helps!

[internetuser]

2 minutes ago, Dan C said:

You can also see an estimated file size in the Export dialog, meaning you shouldn't need to physically export the PDF each time you change these values to understand the size the PDF file will be once created.

I think Amazon KDP is going to kick back any PDFs that aren't at 300DPI (their algorithms are radically inflexible) and I don't mind the challenge of trying to minmax the compression to get in at almost exactly 650mb. As for the file-size preview, I have a reasonably beefy machine (13700k, 128MB ram, 3090ti) and the amount of time it takes for Publisher to calculate a file size is... considerable. I'm going to spend the morning trying to figure out the relationship between DxO PhotoLab5 exports and Publisher exports (exporting from DxO with varying degrees of JPEG compression and feeding the images into Publisher and then exporting with varying degrees of JPEG compression, etc.) Initial tests with pixel peeping at 600% suggest there's way less image degradation happening than I had assumed there would be. Luckily, the whole project is mostly just a ridiculous thought-experiment time-sink and no clients will be harmed in the process. Thanks again.