juxta
juxta copied to clipboard
Generates large collages of images using OpenSeadragon
As now all modern browsers offer WebP support, it’d be great if that tile format would be supported. Similar for AVIF, although support is here a bit less mature.
When performing tSNE sorting, add a step where it is possible to do things such as converting to greyscale before analysing, but without using the converted images for the collage.
The [libvips](https://libvips.github.io/libvips/) tool is much faster than ImageMagic or GraphicsMagic for creating tiles form images. It has `dzsave` which generates all levels of a DeepZoom pyramid structure, but at multiple...
The base `juxta.sh`-script takes a `.dat`-file with images (and optionally metadata) as input. This should be more flexible, so input can be * Files with image paths (or URLs) `images.dat`...
Re-locating an image after forus has changed can be a challenge for at collection og 10,000+ images. One way to help could be to divide the overall collage into areas...
When the user moves a mouse to press one of the navigational buttons, the infobox can easily cover the buttons. Maybe this can be solved by raising the z-order of...
The verbatim image name `missing` signals that an image is missing on purpose. This is not handled well by `RAW_MODE=automax` that attempts to run `identify missing` which fails and shews...
@jesperlauridsen suggests to add bookmarking capabilities (change the URL when panning & zooming).
Creating a collage of millions of tweeted images takes days. Re-rendering the collage to honour deleted tweets could be done without re-downloading the images, but it would still take a...
Observations of `juxta` suggests that as much as 1/3 of all processing power used is by the kernel. This is probably due to the large amount of calls to `convert`...