CLI tool to download pixiv image and ugoira without loss of quality.
OS: Linux,Windows10
Version: 0.7.6
12/2023: usable
- Install
- Run
- Notes
- Detail of lossless
- Play ugoira on smartphone?
- Art filename
- Clipboard pickup + yt-dlp
- How to make lossless ugoira(for developer)
git clone https://github.com/smilingrumia/pixiv-extraction
sudo apt update
sudo apt install curl p7zip-full ffmpeg python3-tk python3-brotli
Open mp4fpsmod -> Release -> DL the latest(here will be 0.26)
# some deps that need
sudo apt install autoconf libtool
tar xf mp4fpsmod-0.26.tar.gz
cd mp4fpsmod-0.26
./bootstrap.sh
./configure
make
# move compiled binary to pixiv-extraction/
mv ./mp4fpsmod <path>/pixiv-extraction/
We need your cookie(login-credential) to get the art.
I will explain how to get cookies and place in pixiv-extraction folder using firefox.
Step 1
Run firefox and login to Pixiv.
Open an art, address bar will look like: https://www.pixiv.net/en/artworks/12345678
F12(open devtool) -> Network tab -> F5/Refresh the Pixiv page
A list in Network tab will be displayed.
May be on the top,
select: Domain(www.pixiv.net) File(12345678) Type(html)
On the right window, goto Headers-> Request Headers
Turn ON Raw headers
Right Click -> select all -> Right Click -> copy
Open pixiv-extraction/httpHeader/pixiv_artpage with text-editor and paste your header
This Header must contain Cookie
Save.
Step 2
Next in Network tab
select one with: Type(json)
And like Step 1, copy&pasete the header to pixiv-extraction/httpHeader/pixiv_artlist
This Header must contain Cookie and x-user-id
Save.
Close devtool.
Step 3
Open some art(picture)
Right click 1 image -> Open link in new window
Address bar will look like: https://i.pximg.net/img-original/img/2016/01/02/03/04/05/12345678_p0.png
F12(open devtool) -> Network tab -> F5/Refresh the Pixiv page
On Network tab, select: Domain(i.pixiv.net) File(12345678_p0.png)
and copy&paste the header to pixiv-extraction/httpHeader/pixiv_art
if contain follow, EXCLUDE THAT LINE!
If-Modified-Since: <something>
Range: <something>
If-Range: <something>
This Header DON’T contain Cookie or x-user-id
Save.
mpv shoud play ugoira(mjpeg) smoothly.
install the latest mpv.
configure
edit mpv.conf
Linux: nano ~/.config/mpv/mpv.conf
loop
idle=yes
force-window
# volume more than 100% brakes sound quality
volume-max=100
# don't show notifications when change volume, seek ,etc
osd-level=0
# play on original size
video-unscaled=yes
# this may help on "viewing big ugoira -> press 'n' to go to next, but mpv still fixed big"
#no-border
Edit input.conf
Linux: nano ~/.config/mpv/input.conf
UP add volume 5
DOWN add volume -5
n playlist-next
p playlist-prev
mpv basic commands
n next video (this works when Drug&Drop multiple .mp4 on mpv)
p previous video (this works when Drug&Drop multiple .mp4 on mpv)
UP volume up
DOWN volume down
RIGHT 5sec next
LEFT 5sec back
Alt+RIGHT right rotate
Alt+LEFT left rotate
f full screen
q quit
Normal mode
./extraction.py URL1 URL2 ...
this will look like
./extraction.py https://www.pixiv.net/en/artworks/12345678 https://www.pixiv.net/en/artworks/23456789 ...
Clipboard mode
./extraction.py -c
On firefox Right Click -> L on the art thumbnail should copy the URL.
copied URL shuld look like: https://www.pixiv.net/en/artworks/12345678
do this on all art that you want.
Ctrl+C on the terminal, type y and Enter to start download.
To see version, just Run ./extraction.py
Where is saved?
Images: save_images/
Ugoira: save_ugoira/
Images
Original image are downloaded.
Ugoira
Technically ugoira is sort of jpeg and frame-rate information.
each one image, have X millisecond to wait, this mean that ugoira is VFR(variable frame rate)
To be lossless as image-quality and frame-rate, pixiv-extraction save ugoira as VFR mjpeg(conventionally as .mp4)
is not gif,apng,webm or lossy mp4.
And can revert to original ugoira images with:
ffmpeg -i ugoira.mp4 -vcodec copy %06d.jpg
Install mpv player
Open mpv -> Settings -> Advanced -> Edit mpv.conf -> type loop -> SAVE
copy ugoira to smartphone and play.
By default, the art will be saved like art-title_001.jpg, and if already exist, will be art-title(art-id)_001.jpg
if want to save like art-id_001.jpg, change the follow:
extraction.py
SAVE_FORMAT = 0
to
SAVE_FORMAT = 1
Liked the clipboard url pickup?
a bonus script ./opt/clipget.py can be used with yt-dlp.
./clipget.py
(copy what you want)
Ctrl+C
‘dllist’ will be created in the same directory as clipget.py
cat ./dllist | xargs yt-dlp
Format: VFR mjpeg
How to:
- Download ugoira.zip and ugoira_meta(the frame rate information is here)
- make mjpeg video: ffmpeg -i ugoira/%06d.jpg -vcodec copy ugoira_pre.mp4
note: ffmpeg may skip the 1st frame when play.
make video with this:
000001.jpg(1000ms) 000002.jpg(1000ms) 000003.jpg(1000ms)
play like:
000002.jpg -> wait 1000ms -> 000003.jpg -> wait 1000ms
solution of pixiv-extraction is to make a copy of 1st frame:
000001.jpg(1ms) 000002.jpg(1000ms) 000003.jpg(1000ms) 000004.jpg(1000ms)
000001.jpg and 000002.jpg is identical.
- make timecode.txt from ugoira_meta
- VFR the video: mp4fpsmod -o ugoira.mp4 -x -t timecode.txt ugoira_pre.mp4
- Play ugoira with mpv