Daily top illustration with images and json format meta data on Pixiv.
All data are from Pixiv. We use the tool of Pixivly/get_pixivly to get all these data, you can crawl the data by yourself if needed.
We organize the data by year. Every year is a folder and has image and json. All pictures are under image with id in its file name. Meta data information are all under json.
|-- 2015
| |-- image
| | |-- pixiv_xxxxxx.jpg
| | |-- pixiv_xxxxxx.jpg
| | |-- ...
| |-- json
| | |-- 20150101_01.json
| | |-- 20150101_02.json
| | |-- 20150102_01.json
| | |-- 20150102_02.json
| | |-- ...
20150101_01.json stands for top 1-50 of date 20150101 and 20150101_02.json stands for top 51-100.
There's something you should notice, since some links/images are invalid, so there might have no picture under image with some ids. What's more, we only provide the small version of the illustrations with average size is 240x480.
We provide the data through Baidu Driver, and we will update the data every year.
Since I am a Big Anime Fan and Pixiv is a great community where people share their best works, I think it's a good idea to do something like Data Mining or Machine Learning stuff with top illustrations provided by Pixiv.
Welcome to do something interesting :)