Work in several views¶

A view is one scatter plot plus its 3D viewer and its own control panel. Views live as tabs, and a tab can be torn off into its own window. They are how you compare two things without losing your place in either.

Opening and switching¶

The tab bar hides itself when there is only one view, so a single-view session looks exactly as it did before.

Each view gets its own accent colour and a coloured dot in its tab. That colour is repeated in the window title and in the title bar of every widget the view owns — which is the only practical way to tell which of three open connectivity tables belongs to which view.

Putting a selection in its own view¶

This is the main reason views exist.

1. Select a group of points in the scatter plot.
2. Selection → Open in New View (Shift+Cmd+Ctrl+N).

The new view contains only those neurons, titled selection (247). Everything else is gone — the embedding is redrawn for the subset, so a cluster that was a smear at full scale becomes structure you can actually see.

Inside a selection view, Backspace (Selection → Remove from View) drops the selected neurons out of it. Work down a cluster by removing what you have dealt with until the view is empty.

The original view is untouched, so you can always go back.

Tearing off a window¶

Drag a tab vertically out of the tab bar and it becomes its own window. Dragging horizontally reorders tabs instead.

Two windows side by side is the natural way to compare a connectivity embedding with a morphology embedding of the same neurons — one view each, both visible.

Window → Move View to New Window does the same thing from the menu, and Window → Merge Window into Parent puts a detached window's views back where they came from.

Shared state¶

Views of the same project are not fully independent. What is shared:

• Annotation state. Neurons you have pushed annotations for are marked in every view, so you can see what has been dealt with regardless of which view you did it in.
• The project itself — the meta table, the data sources, the credentials.

What is not shared: the selection, the embedding, the clustering, the colouring, the scope filters, the control panel state. Those are per-view, which is what makes views useful.

Keeping it manageable¶

Every view can own a connectivity table, a distance heatmap, a feature comparison and a meta explorer. Four views means potentially sixteen windows.

Window → Hide Widgets of Inactive Views deals with this: with it ticked, the widgets belonging to background views are hidden and come back when you switch to that view. The widgets keep their state — they are hidden, not closed.

View → Reset Layout restores the default stacked 50/50 arrangement of the scatter and viewer panes if you have dragged the splitter somewhere unhelpful.

Turning off the 3D viewer¶

When you are working with large selections, the 3D viewer is the expensive part — every selected neuron is a mesh to fetch and render.

View → Synchronize Viewer (ticked by default) is what couples the scatter selection to the viewer. Untick it and selection becomes instant, at the cost of not seeing morphology. The viewer keeps whatever it last loaded.

For a session that is purely about the scatter plot, the layout button in the top-right of the scatter pane (Show only scatter) gives the plot the whole window.

That is the last how-to. If you want the model behind all of this rather than the steps, the concepts pages are next; if you want the full inventory of controls, that is the reference.