Hi Brian,

> ? Number of different flow results = (number of Base surface edges) * (number of Target surface edges) * X ?
> ? # = 4 * 4 = 16 ?

There are a bunch of duplicates in that listing, I think it's 4 different possible rotations with each rotational position being possibly diagonally mirrored or not, so 8 different possibilities.



> rather than trial and error to get the desired final result?

That's actually why it is controlled by the pick location, so that you can control it directly with your pick rather than going into some dropdown and using trial and error going through 8 different possible choices.

I've been thinking about adding a mirror (basically exchange u and v directions) as another option so you could correct the case of when you have clicked right in the corner of each surface which then does not control the mirroring very well. When you pick you don't want to pick directly on the corner itself, more like on one edge somewhat near the end of the edge but not right directly at the end since when you're on the corner you're actually on 2 edges at the same time.

So like one of these 8 red spots here:




- Michael

Thank you Michael for your explanation.


For my own education, I tried 8 flows with non-Rigid, from the center planar rectangle with the same selection point, to 8 catenary curved surfaces, with the
selection point in an oval. Four came out as "rotations." The other four came out as "rotated mirror reflectons."
Holding a printout up to a mirror shows this as well.

Similarly I tried 8 flows with Rigid selected, and the same selection points. Similarly 4 were "rotations."
The other 4 were "non-mirror," with letters shifted "out of alignment."
EDIT: the shifted letters are corrected by Flip surface normals, so things are more complicated....

Hi Brian, yeah that's right, in addition to the rotation and u/v mirroring the mapping of surface normal to surface normal can also come into play. That one does not have any effect in several cases though like if you are flowing curves that are only right on the surface, or objects that stick halfway out on both sides of the surface. But if stuff pokes out just to one side then if the default normals does not do what you want the "Flip surface normal" checkbox can be used to correct that one.

Hmmm, now that I read your post more carefully I see that you're getting slightly different behavior with the "rigid" option enabled, that happens with your text because the rigid option takes the bounding box around each individual object and basically only flows the center point of that box from the base to the target, then does a move + rotate of the object to align the base position with the target position. With your text, the bounding box of each object is a little different, that's why you will see those results. The rigid option would have more uniform behavior when you have some array of a bunch of duplicated objects and you're trying to put that pattern onto a shape without any distortion.

- Michael

Thank you. :-)