cs.technion.ac.il/~weber/Pub … coords.mov
reynantem.blogspot.com/2010/11/g … -tool.html
An idea I’ve had, which I suspect would do the same thing, would be to improve the curve warp layer so that instead of a start point and end point, it can take a list of vertices… So you have one list of vertices that defines an area to be deformed, and another list that the image is deformed to. (There would need to be a way of hiding and showing the different sets of ducks, or there will be trouble selecting them)
一样吗？
Very nice the Gimp tool, but how does it work? does it do triangulations and maps the triangulation later?
We have a Warp Tool that does something similar but with limited source points (4 in a square). The curve warp works different because the source is a single segment and it the entire 2D plane is mapped to a deformed bline with arbitrary blinepoints. In Gimp example the source is a polygon and the destiny is a polygon. Our main difference is that the source possibly can be animiated too and the destiny of course also.
Current Warp layer has the problem of the perspective because the distortion can be done inside and also outside the source rectangle. The polygon to polygon deformation seems to be useful only for internal points so that problem doesn’t exists.
The real problem is to decide the mapping of the internal points, because it has lots of different approachings on what is the “coordinates” of the inner point based on the rounding control points…
For three contour points (a triangle) the map coordinates are simple but how to proceed with four or more points? Adding triangle control areas at the same time the control points are added?
-G
Hello genete
Michael Muré posted explanations on the french Blender forum, the Blenderclan.
There are:
wave.google.com/wave/waveref/goo … BvsoJ9FmFH
and
docs.google.com/View?id=dczx8zht_13cjj838cq
and his blog in French:
pellelatarte.fr/category/gsoc/
I don’t understand it , but maybe it can help you 
Great link mad0! (I don’t understand a word in quick read too!.. Maybe a closer look…)
-G
I was attempting to use the Skeleton Deformation layer on a raster image in the same way that the Skeleton layer is used on vector graphics.As I understand it the Skeleton Deformation is not able to perform a deformation on only a specific area of an iamge. To solve this could a leaf be taken from Animata’s book?
To deform an image a user creates a cage around an image, triangulates it and then, after creating a skeleton, selects which bone will influence which points in the cage.
Some screenshots to illustrate
Yep, animata seems to offer some cool things … but never test it …
Code source (cpp) : code.google.com/p/animata/
Animata : real-time animation software (how to build on linux and more … ) web3.lu/animata/
Definitely have a try! It’s a bit slow under Linux but it’s fast enough to test out the bone and cage transform feature that I mentioned
From what I understand the Skeleton Deform uses a sort of cage transform internally, but the number and position of control points are hardcoded.
If that could be opened up for users to add more then I think it’d be a step in the right direction