Does anyone here know a self hostable service (or a demo of someone else hosting it) which can split common .cbr filetype to a .cbz or other filetype to split the panels?
So for example the 1 image I put with this post, would be 8 seperate images (pages).
That would ideally somehow make the comic books readable on a (somewhat old) black and white travel-Kobo…
You must log in or register to comment.
This isn’t a thing because there are many comics that don’t adhere to “frames”. They overlap with others, use the whole page, etc.
But beyond this, decompress your CBR/cbz files and use imagemagick to find frames and isolate them.
This is a thing. It just doesn’t work very reliably for the reasons you mentioned.
It does work kind of ok on some comics like the one OP shared and generally pretty well on manga since that tends to be more consistent with the boxes.
I don’t know of anything open source or self hosted that does this though.
This sort of thing ain’t my bag baby.
isnt this more of an image vision task than a aimple utility task?
I’d be amazed if each of those panels coordinates were annotated somewhere
the coordinates aren’t there i think, but there are github projects out there that “detect” the panels and suggest split based on that. For most of the panels of most of the comics, that would be more than enough to do a clean split. I just can’t find a real relatively easily deployable service that incorporates it.
https://maxhalford.github.io/blog/comic-book-panel-segmentation/
Removed by mod
