Tired of KOReader splitting your webtoon panels? I made a fix.

[QQRush]

Hi everyone!
I read a lot of vertical webtoons on KOReader(on e-ink device), but there was one thing that always bothered me: normal scrolling doesn't care where a panel actually ends.
This often results in characters, dialogue bubbles, or entire scenes being awkwardly split between two screen views.
So I started working on a plugin to solve this problem: Smart Webtoon Scroll.
Instead of simply moving down by a fixed amount, the plugin analyzes the webtoon and tries to find natural visual boundaries, adjusting each scroll so panels and dialogue remain together whenever possible.
The video shows the exact same webtoon running side by side:



LEFT — Standard KOReader
RIGHT — Smart Webtoon Scroll
Some of the features currently included:
Panel-aware smart scrolling
Detection of both white and black separators
Avoids splitting panels/dialogue when possible
Special handling for very tall panels
Classic, Continuous and AUTO modes
Continuation across images/files
Adaptive white/black side background
Page preloading for smoother navigation
Configurable preload amount
Keeping the plugin lightweight has also been one of the main goals. KOReader runs on many e-ink devices with limited CPU and memory, so throughout development we tried to keep the code and image analysis as simple and efficient as possible, avoiding unnecessary processing and features that could make reading feel slower.
The goal is not to turn KOReader into a completely different reader, but to keep the normal KOReader experience while making vertical webtoons/manhwa more comfortable to read — without unnecessarily weighing KOReader down.
I also want to be transparent about the development process: I developed this plugin with a lot of help from ChatGPT, especially for coding, debugging, analyzing edge cases and turning my ideas and testing feedback into working Lua code.
There was a lot of back-and-forth testing involved: change something, load it on KOReader, find a webtoon that breaks it, figure out why, change it again... 😄
So while the idea, testing and direction came from my own experience using KOReader, AI assistance played an important role in building and refining the plugin.
It's open source and I'm still improving it. Feedback is very welcome — and if you have a particularly weird webtoon that manages to break the detection algorithm, I'd genuinely like to see it!
GitHub:
QQRush / smart-webtoon-scroll.koplugin
Hope it's useful to other KOReader/webtoon readers!

[Jellby]

Is it only for "webtoons", or would it also work for regular comic books? Say a vertical comic book page, with the reader in landscape orientation, and fit-to-width... A page would take a couple of scroll downs. Would the plugin try to avoid splitting panels here?

[QQRush]

Hi, it's designed specifically for webtoons, using it for comic books is not the best. But wanting to add a comics mode is not complicated.

[Jellby]

Just for reference, I already mentioned it here: https://www.mobileread.com/forums/sh...2&postcount=11

[QQRush]

Originally Posted by Jellby

Just for reference, I already mentioned it here: https://www.mobileread.com/forums/sh...2&postcount=11

Hi Jellby!

Your comment and the two example images you shared actually inspired a new project! 😄

I've been working on **Smart Comics Scroll**, a new KOReader plugin partly based on my previous **Smart Webtoon Scroll**, but designed specifically for traditional comics and manga in landscape mode.

Instead of navigating panel by panel, the plugin displays comics at full width and tries to detect the spaces between rows of panels, so page turns happen at more natural positions without unnecessarily splitting panels.

It also includes automatic whitespace trimming and a feature that can slightly scale down the content when this allows a complete panel row to fit on screen.

**GitHub:**
https://github.com/QQRush/smart-comics-scroll.koplugin

**Video demonstration:**
https://youtu.be/EC0JDg7Stds

I've also credited your original comment and illustrations in the README, since they were the inspiration for this project.

It's still an early beta, but it's already working quite well in my tests.

I'd really love to hear what you think, especially since your suggestion started the whole thing!

Thanks again for the inspiration! 😊