Smooth pinch-zooming and panning on Windows Phone 8

I know you're talking about 8 and I'll post a link to an article related to 7, but it was very useful when playing around with Windows Phone so here it goes:

https://www.wintellect.com/building-touch-interfaces-for-windows-phones-part-3/

I don't imagine that much has changed since then...

I wanted to get this right from a mathematical point of view. The result is something similar in correctness to Telerik's PanAndZoomImage. If you aren't interested, jump straight to this gist (It works with WP7.1+). You'll need to reference System.Windows.Interactivity and the Windows Phone toolkit.

Usage:

<Image Source="http://i.imgur.com/ZbKlRzK.jpg">
    <i:Interaction.Behaviors>
        <phoneApp1:PanAndZoomBehavior MaxZoom="10" />
    </i:Interaction.Behaviors>
</Image>

Math

Panning and zooming uses 2 out of 4 transformations of CompositeTransform, namely Translation and Scaling. The key point is understanding how to compose two of those scale+translate transforms. I'll use haskellish notation, because it is less of burden to type and read. Our 'primitives' are

1. scale s = scale around (s.x,s.y) with factor s.x in x direction and s.y in y direction
2. translate t = offset all points by t.x in x direction and t.y in y direction

CompositeTransform scales around a center point, which is expressed like

scaleAround c s = translate c . scale s . translate -c

The following rules hold (do the math if you don't believe me, all operators are componentwise):

1. translate a . translate b = translate (a+b)
2. scale a . scale b = scale (a*b)
3. translate t . scale s = scale s . translate (t/s)

A CompositeTransform is like

transform s c t = translate t . scaleAround c s
                = translate (t+c) . scale s . translate -c

When composing two of those transforms, we have to move around primitives until we get to such a form above. Let a and b be those two CompositeTransforms. So we get:

transform' = b . a
           = translate bt . scaleAround bc bs . translate at . scaleAround ac as
           = translate bt . translate bc . scale bs . translate -bc . translate at . translate ac . scale as . translate -ac
           = translate (bt+bc) . scale bs . translate (ac+at-bc) . scale as . translate -ac
           = translate (bt+bc) . translate (ac+at-bc)*bs . scale bs . scale as . translate -ac
           = translate (bt+bc+(ac+at-bc)*bs) . scale (as*bs) . translate -ac
           = translate (bt+bc-ac+(ac+at-bc)*bs) . scaleAround ac (as*bs)
           = translate (bt+at*bs+(bs-1)*(ac-bs)) . scaleAround ac (as*bs)

This is just because I was frustrated with the amount of profound documentation on why certain people do certain things.

For the actual composition code, looko here

I know it's a late answer but here's another sample project which could help to resolve this problem http://code.msdn.microsoft.com/wpapps/Smooth-flick-and-zoom-with-7760c7f7