Currently, they are all independently scaled to the timeline height based on the visible data. So, if a page has 1MB of images, its scale is [0, 2] and it always renders as 50% * height. Another category could have range [0, 100MB] and 50% * height would represent 50MB. So it currently isn't possible to visually compare different category timelines. This seems like it strictly reduces the usefulness of the split out timelines. I think they should retain the relative sizes like in the stacked line graph, which seems to be scaled to [0, max(sum(categories)]. We probably instead want the scale to be [0, max(categories)] so that the max-value category reaches 100% * height at its max value.
<rdar://problem/26589754>
Hmm, the intent was that the individual timeline graphs emphasize the change in an individual category in the selected time range. If one category dominates the others, then scaling the others (which may have changes of 50%-100%) would be unnoticeable. Maybe that isn't too bad, given the one category is so much later, but it makes debugging the non-largest category more difficult.
To elaborate. A stacked line chart is great for showing the combined data over time. However, it is very difficult to see the changes to an individual category. That is what the individual timelines in the timeline view is meant to show: the relative changes for an individual category. If you were to redraw the individual categories at the same scale as the overview's stacked line chart, then, to me, it would seem rather pointless to draw it again.
Well, we could add a toggle to switch between the two scales.