You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've noticed that in RF3 and also in RF4, the handles jump after adjusting them. Now that I can see the shape of the curve, it looks to be because the transform of the geometry and the transform of the curve do not quite line up.
This results in the ability to mangle strips by adjusting them too much.
(Since the current behavior is similar to v3, this issue is not a priority for v4.0)
It makes sense, because the created geometry follows the original stroke, and then the handles do their best to fit that stroke after the fact, so the two will never match up exactly. A smooth curve with only two control points will pretty much never match an arbitrary stroke.
To improve this, what if the curve that the handles use was generated the same way the geometry is placed? IE there is a point at every edge and face center. Then it would always match up 1:1.
The challenge then would be how the handles work. How could we control all those points in an artist-friendly way?
First, there would always be control points with handles on the first and last quads, and on any intersections just like in RF3. Then, we could sample the curve at each point and draw a control point with no handles wherever there is a significant difference in the tangent angle as compared to the handles on the ends. Then, the transform of each point along the curve is interpolated between the visible control points.
It would look something like this:
That way, not only would we have much more control over the strip, adjusting it should be completely stable and reversible.
The text was updated successfully, but these errors were encountered:
I've noticed that in RF3 and also in RF4, the handles jump after adjusting them. Now that I can see the shape of the curve, it looks to be because the transform of the geometry and the transform of the curve do not quite line up.
This results in the ability to mangle strips by adjusting them too much.
(Since the current behavior is similar to v3, this issue is not a priority for v4.0)
It makes sense, because the created geometry follows the original stroke, and then the handles do their best to fit that stroke after the fact, so the two will never match up exactly. A smooth curve with only two control points will pretty much never match an arbitrary stroke.
To improve this, what if the curve that the handles use was generated the same way the geometry is placed? IE there is a point at every edge and face center. Then it would always match up 1:1.
The challenge then would be how the handles work. How could we control all those points in an artist-friendly way?
First, there would always be control points with handles on the first and last quads, and on any intersections just like in RF3. Then, we could sample the curve at each point and draw a control point with no handles wherever there is a significant difference in the tangent angle as compared to the handles on the ends. Then, the transform of each point along the curve is interpolated between the visible control points.
It would look something like this:
That way, not only would we have much more control over the strip, adjusting it should be completely stable and reversible.
The text was updated successfully, but these errors were encountered: