KC Viz has graph zoom and node/label size zoom. We have considered label size changes before, and the Dudas et al 2014 paper discusses this within their framework (http://semantic-web-journal.net/system/files/swj755.pdf).
I am not entirely convinced of the need for layout zoom in Biomixer, but node zoom might be desirable.
Dudas et al refer to node/text zoom as graphical, which is unfortunate; it always makes me think of changing the viewport/layout distances.
One possibility is to not quite implement graphical zoom in the way it is usually done. Normally, that sort of zoom does transformation on the underlying coordinates, without being bounded by the viewport. I am not personally a big fan of viewports that allow the graph to get out of them. I would like to consider a graphical zoom that is actually layout parameter zoom (also see #452). That is, when zoom is activated, the layout's parameters are changed to accomplish coordinate changes, rather than transforming node coordinates. For force layouts this could be unwise. It also restricts users, perhaps needlessly.
I have theoretical and practical problems with zooming beyond the viewport. Theoretically, I believe that, depending on the purpose of a graph visualization, if you need to zoom out, you have too many nodes in the graph. Practically, there are a lot of sub-features and associated features required to support spatial zoom to make the user interaction comfortable and complete.
KC Viz has graph zoom and node/label size zoom. We have considered label size changes before, and the Dudas et al 2014 paper discusses this within their framework (http://semantic-web-journal.net/system/files/swj755.pdf).
I am not entirely convinced of the need for layout zoom in Biomixer, but node zoom might be desirable.
Dudas et al refer to node/text zoom as graphical, which is unfortunate; it always makes me think of changing the viewport/layout distances.
One possibility is to not quite implement graphical zoom in the way it is usually done. Normally, that sort of zoom does transformation on the underlying coordinates, without being bounded by the viewport. I am not personally a big fan of viewports that allow the graph to get out of them. I would like to consider a graphical zoom that is actually layout parameter zoom (also see #452). That is, when zoom is activated, the layout's parameters are changed to accomplish coordinate changes, rather than transforming node coordinates. For force layouts this could be unwise. It also restricts users, perhaps needlessly.
I have theoretical and practical problems with zooming beyond the viewport. Theoretically, I believe that, depending on the purpose of a graph visualization, if you need to zoom out, you have too many nodes in the graph. Practically, there are a lot of sub-features and associated features required to support spatial zoom to make the user interaction comfortable and complete.