Hi Petr,

> Aha... I didn't realize that contrary to background image the
> object is getting squished as you move a cursor towards the
> opposite frame corner.

Well, it is different than the background image if you track along a horizontal line while going over, that will do the 1D non-proportional scale in this case which images don't do.

But really I meant any kind of scaling at all. It would be kind of frustrating if you could get a mirrored and arbitrarily scaled object by dragging over there, but not get a mirrored object of the same size, actually though that seems to be the standard behavior in 2D illustration programs that do allow dragging to the opposite side.


> I think I am not completely clear about this part.
> Do you mean the width of the bounding box of frame (distance
> between grips) or object's bounding box or distance between
> grip and world Y axis?

I mean width of the object's bounding box. In the example I showed the edge of the bounding box happened to be on the world Y axis, so that's all I meant by referring to the Y axis specifically there.


> Will a position of the pivot point (both scale and rotate pivot)
> impact on mirroring itself?

Not for the one that I showed in that previous picture. I would plan to have that available in the same spot and behavior regardless of pivot mode/placement.

However, if you have scale-from-center turned on, I was thinking about having other snap points in the other frame corners so that you could do an exact "in place" mirror (around the center point) by targeting those points.


> Will we be able to identify/change the location of a mirror axis?

Well, it will always be the 2 bounding box edges opposite of the corner that you choose. So the choice of which corner you drag will let you pick between those edges, but there won't be any way to set it to some arbitrary line elsewhere, for that you'll still need to use the regular Mirror command.


> Your illustration includes some point, does the mirror axis (parallel
> to y axis) pass through that point at the end of a curve or is there
> just hard-coded Y-world mirror axis?

The mirror axis will always be the edge of the bounding box, but of course when you have an endpoint of a curve sticking out to one side, then that end will be on the bounding box edge.


> Another question. Is a rotate icon always on the top side of the frame?
> In other words, when I rotate the object using the editing frame, deselect
> it and re-select the same object, does the original frame or new one with
> the rotate icon on the top appear?

If you grab it and rotate it, it will then spin around to some other location, but if you unselect and then reselect, then the whole frame will get reset on the new selection and the rotate grip will be back at the top of the axis aligned frame again.

In order to keep the rotation between selections, I would have to maintain some additional persistent state associated with objects... It could be possible to do that, but it brings up a bunch of issues.

It seems to be pretty common for 2D illustration programs to reset the frame like this, like for example Xara does it after every rotation even if you don't change the selection. Adobe Illustrator seems to try to maintain the frame orientation if you have things like a bunch of rectangles all at the same angle selected, but it too will reset it to world-aligned after every rotation if you have like 2 rectangles at different angles selected at once.

- Michael