Sure, here is the same blend surface that “RefitTrim” fails to convert properly:
Test RefitTrim.3dm (870.1 KB)
thanks, I see what you mean, RefitTrim could indeed slide points here. I’ll ask our dev what can be done.
2 Likes
@Rhino_Bulgaria Looking at this more closely though: If the points are sliding over their control polygon, it ends up in a larger surface deviation than when doing a refit and then match. See attached
Test RefitTrim_inspection.3dm (976.1 KB)
My test with “MoveUVN” showed 0,01 degrees deviation to the left (shorter) edge after sliding of the control points manually along the U direction (I actually used “Along tangent”, because I forgot to switch to “Along control polygon”). This is far better than the 0,52 degrees achieved by “RefitTrim”.
After further adjustment of those control points along the normal direction, I achieved 0,00 degrees.
Test RefitTrim with MoveUVN.3dm (99.5 KB)
I just opened your model, but I fail to see any bad deviation which is worse than what the “RefitTrim” does automatically.
1 Like
did you compare all the layers in that file I sent?
Yes, I did. All the surfaces were far better than what “RefitTrim” could achieve.
this is confusing btw. One of the surfaces in that file is a direct result of RefitTrim
All of the other surfaces are better than the surface made by “RefitTrim”. Only the latter shows unacceptable deviation on either end.
What does that mean? 
it shows the deviations between the original (trimmed) and the modified surfaces, so the way I read it is that RefitTrim followed by MatchSrf is pretty much identical to your manual method.
Note that in this case, your surface is twisted, and that makes that in order to keep the surface as close as possible to the original, it has to make a compromise in edge continuity.
It seems like there is a necessary trade-off - either you maintain the tangent direction or you minimze the deviation at the edge - where those are not compatible, as here, you can’t win, something has to give. Right?
The RefitTrim edge comes closest, by far, to the original edge in Gijs’ file.
I could certainly see the utility in getting as close as possible by sliding on the tangents only, but that is not, as far as I know, something currently considered by RefitTrim- I’ll ask the developer.
-Pascal
2 Likes
Not sure what does it mean that my surface is twisted, as I only used “MoveUVN” to slide 2 of the top 6 control points. I also used Osnap to snap the first control point to the end of the lower target surface. Before doing so, I extracted the top edge of the original trimmed blend surface, in order to use it as a guide for the manual control point editing. Then I untrimmed the blend surface and ran “MoveUVN”. It’s a pretty standard workflow.
if you extract the Control Polygon, you can see that in between 3rd and 4th, the surface makes a twist, and this makes that you can’t simply slide and keep the surface deviation 0
Well, as I mentioned who’ve, the manual sliding resulted in a deviation of about 0,01 degrees. Not prefect, but still at least 50 times more accurate than the 0,54 mm deviation produced by “RefitTrim”.
Hi Bobi - the refitTrim results are .04 units off of the original edge , that is what RefitTrim is shooting for, a close match to the trimmed edge if possible; as far as I know it does not condsider the tangent direction.
Sliding points on the tangents keeps the tangent direction at the cost of meeting the original edge. That may be fine, but not what RefitTrim does… you cannot have both in this case because the surface tangent control polylgons are not parallel.
-Pascal
Hi @pascal , in my example the manually modified surface via “MoveUVN” along the U direction was capable of keep the edges at 0,0000000 mm deviation (basically no deviation at all) from the two adjacent surfaces, whereas the tangency got a very small hit at just 0,01 mm that was easy to fix with the N slider in “MoveUVN”. On a more complex, curved example the manual adjustment would be slower compared to the combination of “RefitTrim” + “Match surface”. However, the 3d model I provided is a good example of how “RefitTrim” could benefit from some improvements so that it will also take respect to both, distance and tangency.
I’m not saying you have not done a good job - it looks like the adjusted surface is only about 50% farther off the original edge than what RefitTrim gets on its own. I am struggling though, to understand how this case constitutes a ‘RefitTrim failure’.
-Pascal
It’s the 0,54 degrees deviation of tangency that I consider a failure. In my opinion, “RefitTrim” must take the tangency into consideration while converting the original trimmed surface to fit inside its trimmed border.
Also, I think that “RefitTrim” will hugely benefit from having its own floating panel with some interactive options for the user to try and decide which one is better, including integrated tangency analysis (or maybe automatic opening of the “Edge analysis” tool.
2 Likes
Yes, this is also true for all advanced surfacing tools.
2 Likes