VIKUS Viewer

VIKUS Viewer is a web-based visualization system that arranges thousands of cultural artifacts on a dynamic canvas and supports the exploration of thematic and temporal patterns of large collections, while providing rapid access to high-resolution imagery.

Documentation

This repo contains the HTML, CSS and JS of the VIKUS Viewer software. To get started you will have to clone this repo and run a webserver. We recommend nginx for production, but any web server will work too. There is no server-side logic necessary. To minimize the loading time your should make use of GZIP compression on JS and CSV files. Also enable HTTP/2, since Multiplexing will heavily help with loading all those image assets.

IIIF

VIKUS Viewer can be used with IIIF Collections through the vikus-IIIF-generator.

Metadata

To use the VIKUS Viewer for a custom image collection, you need to prepare metadata files that describe the collection and objects, and configure the visualization. To get started, you first need to create a /data folder which will contain all metadata and image files. Have a look at the metadata generated for the Van Gogh collection (Van Gogh Museum) as a reference for the following descriptions.

config.json

This is the configuration file that defines the project name, data URLs, columns, styles, and what is shown in the detail sidebar of your collection. The URLs don't have to be absolute, but it can be handy if your assets are not hosted on the same server.

data.csv

The data.csv holds all the metadata information for each object in the collection. The following fields are mandatory: id , keywords, year.

id is is linked to the name of the corresponding image. (id: 123 -> 123.jpg)
keywords comma seperated list of keywords for the tags on the top
year can be a number or a string, will be sorted ascending
_fields these are custom metadata fields (note the prefixed underscore)

All of the columns are beeing sticked together to enable the freetext search.

timeline.csv

The timeline.csv holds the information for the timeline displayed underneath the years.

year can be a number or a string, is linked to the year field in data.csv
title the headline of the blurb
text first detail text when zoomed in a little bit
extra additional text when zoomed to the maximum

info.md

This is the information displayed on the left side when opening the visualization. You can put in any kind of Markdown.

Images

Apart from the metadata, you need to preprocess the image files, i.e., to generate sprites and textures for the different zoom levels. Please see the vikus-viewer-script for the details. After running the script you can place the resulting folders into /data or any other location. Make sure that the texture URLs in the config.json point to these folders.

Similarity (t-SNE/UMAP)

As an alternative to the temporal view, you can create a t-SNE/UMAP layout based on image similarity. The script in vikus-viewer-script creates a tsne.csv which can be put next to the data.csv in the /data folder. Image similarity is calculated via the imagenet activation logit and then run through t-SNE. An additional spacing step ensures no overlaying images in the distribution. You can also create the layout through a Jupyter Notebook in python using CLIP.

Layouts

You can add layouts or remove the lime layout in the loader.layout section of the config.json. Add a custom layout in this format: {"title": "UMAP", "url": "umap.csv", "scale": 0.8 }. The scale parameter is optional and can we tweaked depending on your layout and number of images.

Credits

VIKUS Viewer was designed and developed by Christopher Pietsch. The VIKUS Viewer software is based on the visualization code behind Past Visions, a collaborative effort by Katrin Glinka, Christopher Pietsch, and Marian Dörk carried out at the University of Applied Sciences Potsdam in the context of the Urban Complexity Lab during the research project VIKUS (2014-2017). Related Paper: Past Visions and Reconciling Views. The T-SNE view has been implemented for the Sphaera project with funding from Chronoi-REM

cpietsch/vikus-viewer