I stumbled on a way to extract maps from older APs and make them look nicer using a touch of machine learning (but without having to install a ton of things on my machine). I showed a few fellow GMs, and rather than explain how I did it to each one, I wrote up a blog post.
(Spoiler warning: I used book four from Council of Thieves as my example, so if you really don't want to see a map from that book, consider yourself warned).
https://katieplaysgames.wordpress.com/2022/07/04/adventure-path-maps-vtt-an d-minimizing-heartburn/
It's not perfect, but it does make maps that lead to fewer headaches from players who have likely zoomed way in.
If you open the PDF in Adobe Reader you should be able to left-click to select the image (might be trial and error to find a place that you can click to select it), and then right-click to copy image. Then open Paint and paste the image in, save the image, and you should be good to go.
Very occasionally I'll find a PDF that won't let me select the image in which case I use "Some PDF Image Extractr" (version 1.5) to pull the images from the PDF. (I've don't recall having to do this with a Paizo book though).
If you open the PDF in Adobe Reader you should be able to left-click to select the image (might be trial and error to find a place that you can click to select it), and then right-click to copy image. Then open Paint and paste the image in, save the image, and you should be good to go.
Very occasionally I'll find a PDF that won't let me select the image in which case I use "Some PDF Image Extractr" (version 1.5) to pull the images from the PDF. (I've don't recall having to do this with a Paizo book though).
You're missing the up-scaling part.
That's not just resizing, where every pixel becomes four (or more) gigantic pixels.
There's a decent range of software out that uses machine-learning and pattern libraries to actually create realistic data. You can't point it at a photo of illegible text and expect to be able to read it because the detail it creates isn't literally correct, but for purposes of appearing likely, it's great.
Imagine a photo of some trees. "AI" up-scaling will create more leaf detail. The leaves won't be exactly where they really were when the photo was taken, but they will be where leaves probably would be. Thus to the human eye there's more detail.
Well, in the case of AP maps, it improves quality dramatically.
I haven't read the linked blog because... I don't do blogs and I have my own software pipeline, but if you're doing digital gaming and don't know about this stuff, it's probably worth a read.
I guess I should've been more clear.... kcunning, instead of extracting the map image the way you do in the blog post you can start off with a better quality map if you open the PDF in Adobe Reader you should be able to left-click to select the image (might be trial and error to find a place that you can click to select it), and then right-click to copy image. Then open Paint and paste the image in, save the image, and you should be good to go.
Very occasionally I'll find a PDF that won't let me select the image in which case I use "Some PDF Image Extractr" (version 1.5) to pull the images from the PDF. (I've don't recall having to do this with a Paizo book though).
If you extract the map image using either of these methods you'll end up with what appears to be a better image than even the upscaled one you show in your blog post.
| 1 person marked this as a favorite. |
I believe the meme response would be "why not both"?
My workflow is to extract the image from the source, much like you describe, only using full Acrobat, with a bounce through Photoshop because annoying things often happen with the transparency channels in the original PDFs. That's all lossless. Then I upscale.