reCBZ - CBZ repacker
CLI utility for repacking comic book archives (.cbz). Can be used to greatly reduce disk usage, upscale, and optimize comics & manga for reading on mobile devices and e-Readers. Can also be used to convert image formats in bulk (see other uses).
Purpose
I own a fairly large library of manga, and it takes quite a bit of space on disk. This isn't really a problem most of the time, but it limits what I can put on my Kobo e-Reader (which has "only" 32GB of storage). I prefer to keep the original files intact on calibre on my computer, but use this tool to optimize the .cbz files in bulk so they use less space on my Kobo, by resizing the pages to a slightly lower resolution, and saving them as a different format in black and white, which can cut the size of a high quality .cbz by half.
For example, by repacking with --auto/WebP, this can cut the size of the first volume of Chainsaw Man from 180MB to just under 96MB, without affecting image quality. Over the 11 published volumes, that amounts to over 1GB saved (which is quite a lot when you consider most e-Readers still have only 4GB)! And that's just by changing the format, the size can be further reduced by another 50MB by downscaling to 120% display resolution, while still maintaining optimal visual clarity on a 6" 300PPI screen — effectively tripling the amount of manga that can be stored on your device.
Note that due to how lossy images formats like JPEG/WebP work, compressing and overwriting the same file many times over will eventually lead to image degradation that is noticeable to the naked eye, so by default this program creates an optimized copy while preserving the original, although lossless formats are also available. As a general rule, you can be more aggressive with compression on black and white images.
Install
python -m pip install reCBZ
or build from source:
git clone https://github.com/avalonv/reCBZ
python -m pip install -e reCBZ
Usage
recbz [options] files
Accepts a valid .cbz or .zip file, or a collection of files. With no arguments passed, it will repack the file(s) with slightly higher compression while retaining the original format.
The output file(s) will always be saved as filename [reCBZ].extension, unless --overwrite is specified.
Examples:
Click to expand
Convert 'Blame! Master Edition v06.cbz' to various formats and ask which one to repack with:
recbz --assist 'Blame! Master Edition v06.cbz'
Convert two volumes to lossless WebP at twice the Kindle resolution:
recbz --fmt webpll --size 2250x3000 'Our Dreams at Dusk v01.cbz' 'Our Dreams at Dusk v02.cbz'
To repack all books in the current directory (e.g. a series), use a '*' to match .cbz files:
recbz ./*.cbz
- On Windows, slashes '/' should be replaced with backslashes '\'
Automatically convert and repack all books on the 'Blame!' folder:
recbz --auto ./'Blame!'/*.cbz
Rescale all books on the "Saga" folder to 1440p 3:4, convert pages to grayscale and save as high quality JPEG:
recbz --size 1440x1920 -bw --quality 90 --fmt jpeg ./Saga/*.cbz
Configuration
General options (no effect on file size):
Click to expand
--nowrite -nw
- Dry run. The repacked archive isn't saved at the end, making other options completely safe.
--compare -c
- Does a dry run with a small sample of images, converting them to available formats using current settings, then displays a disk usage summary for each.
--assist -a
- Same as --compare, except it then asks you which format to use for a real run.
--auto -A
- Same as --compare, except it automatically picks the best/smallest format for a real run.
- Most of the time this will be a .webp. If you wish to exclude this format, you can add --nowebp.
--overwrite -O
- Overwrite the original archive. Specifically, it will be converted to a valid .cbz structure, meaning that non-image files will be discarded, and the folder structure will be flattened, any images sharing a name will be lost. Make sure you understand what this means before using this.
--recursive -R (TODO/Unimplemented) see #examples
- Search all subfolders in the current path for .cbz or .zip files to convert.
- Exercise caution when using with --overwrite, may lead to loss of data.
--verbose -v
- More progress messages. Can be repeated (-vv) for debug output.
--silent -s
- No progress messages.
--processes 1 - 32
default: all available CPUs (100% utilization)
- Max number of processes to spawn. This will only improve performance if your CPU has cores to spare (it's not magic!). Lower this to 2 or 4 if you're experiencing high memory usage.
--sequential
- Disable multiprocessing altogether. Use this only if you're still experiencing memory issues, or for debugging.
--zipext .cbz or .zip
default: .cbz
- Extension for the new archive, signals to the OS which mimetype to open files with (they're the same internally).
--zipcompress 0 - 9
default: 0
- Compression strength for the archive (after images have been converted). The default (0) is strongly recommended, setting it to higher values is nearly always counterproductive, it will barely affect archive size (if at all) as the images are already compressed, but will significantly increase the time it takes to open it.
Image options (affect file size):
Click to expand
--fmt format
default: same as source
- Format to convert images to. One of: jpeg, png, webp or webpll — png and webpll are lossless. Try -c to get an idea of how they compare, this will vary depending on the source format. Omitting this option will preserve the original format.
--quality 0 - 95
default: 85
- Image compression quality for lossy formats, will have a large impact on file size. Smaller values produce smaller files at the cost of visual quality. This option only applies to lossy formats
-
Notes:
- Low values degrade image quality less in WebP than they do in JPEG. Similarly, grayscale images are less affected by this setting that color ones, so you can lower it a bit more when using --fmt webp or --grayscale.
- Values higher than 95 will usually increase file size without actually improving quality.
--size WidthxHeight
default: don't rescale
- Rescale images to the specified resolution, using Lanczos interpolation. Does its best to detect and preserve landscape images.
- Add --noupscale to disable upscaling, so images can only be downscaled (as long as they're greater than value).
- Add --nodownscale to disable downscaling, so images can only be upscaled (as long as they're less than value).
- 1440x1920 (3:4) is suitable for most 6"/7" e-Reader screens. For smaller devices, setting this to 150% of your screen's resolution is usually the best compromise between quality and file size, still allowing you to zoom-in to read the lore critical thoughts of that moe character.
-
Note: this isn't magic. Please don't upscale a low quality source to upload to manga sites and claim yours is higher quality, because it isn't, and it will annoy people.
--grayscale -bw
- Convert images to grayscale. Useful for e-Paper screens, reducing file size by another 10% to 20%. Provides no benefit to comics which only have a few coloured pages (manga).
Other uses
Although this was explicitly created with manga and comics in mind, it can be used for bulk rescaling and conversion of images in general (it's pretty fast at that thanks to parallel processing), you just need to pack them into a ZIP archive first. There are some important caveats: non-image files will be automatically discarded, and the folder structure will be flattened (every image will be written to the same folder), meaning that files which share a name will be lost. Be very careful when using --overwrite.
Note about WebP
Generally speaking, the WebP format tends to compress images more efficiently than both JPEG and PNG, allowing both lossy and lossless methods. This leads to a few noticeable quirks when converting from lossy to lossless and vice versa, which are covered here, but overall, if you're confident your reading software supports it, this is probably the best option for saving disk space.
It isn't perfect however: WebP adoption outside of web browsers has been glacial, and it is not universally supported yet, meaning it might not open on older devices and the vast majority of e-Readers (Kindle/Kobo) — although Koreader allows you to get around this limitation.
TL;DR If you're repacking content for the purpose of sharing with others on the web, it is strongly advised to avoid this format, as many devices still aren't incapable of displaying them.
Why not support .cbr and .cb7 archives?
Both RAR (.cbr) and 7z (.cb7) are non-standard compression formats. Undoubtedly they have helped many people compress files on PC, but they are not pre-installed on most operating systems, and thus cannot be opened on most mobile devices and e-Readers without tinkering. Additionally, WinRAR is a proprietary program which limits official access to Windows, as the name suggests, which makes it annoying for future users that plan to read in other devices, and cannot be bundled with free software.
Also, the compression algorithm used to pack images into a comic book archive has a negligible effect on the finished archive size, as the images are already compressed, so even if these programs can achieve higher compression ratios than zlib/zip in theory, they offer little to no advantage for image content.
TL;DR If distributing manga or comics over the high seas PLEASE stick to the standard .cbz format, as it's guaranteed to work on nearly every device. RAR is bad. Stop using RAR for this.
WIP/TODO If you need to convert .cbr or .cb7 files to .cbz, use this script.
