Fillet surface vs Blend surface difference

By the way I coded such a functionality by myself. Its basically a extrapolation of each controlpoint row.
Sure you cannot trim like this if there are not 4 edges guaranted, and you cannot represent any edge-> esp. with kinks. But being a good modeller prevents you from having any kink or weird surface patch layout in your model.

no,

It’s not “possible” in the sense that you can trim surfaces with any arbitrary curve, precisely, without changing the shape of the underlying surface.

@TomTom It appears that the “true trim” in the illustration you provided above is not to the trim curve. Is the resulting “true trim” to an isocurve? If so the trimmed surface can be rebuilt exactly with the trimmed, isocurve edge as a boundary. Perhaps you post a file with the geometry in addition to the screen shot.

Surfaces which are trimmed with a trim curve which is not an isocurve, and then rebuilt to the new edge with the original number of control points will usually not be exactly the same shape as the original trimmed surface before rebuilding. As I mentioned above there are exceptions depending on the shape of the trim curve - an isocurve for the trim curve is one example. However the difference between the rebuilt surface and the original surface may be insignificant for the particular use of the surface. If the difference is not insignificant then it can be made smaller by rebuilding with more control points. With enough control points the rebuilt surface can be as close as needed to the original surface.

Well,

I thing there is a misunderstanding here.
First of all real trim shouldn’t replace masking trim. Both are important features, but rhino
sadly only provides one of them. Real Trim was also part of VSR Tools only called SplitSurface.
Obviously it is not possible to extrapolate controlpoints in a sense that you can fit any boundary, true.
Its also true that a real trim slightly changes the shape, depending on overall impact of the change.
The more a trimboundary differs from the surface edges, the more deviation will be. Usually deviation is
under 0.001 mm, which is quite acceptable. Now why is that? The trick is to extrapolate rows of the controlpoints
until a boundary is best matched. Sure in most cases you won’t match the full new boundary, but that’s not important for blend, who cares if the trim curve sligthy differs to the new aligned edge.
It is also true, that this kind of blend works in any cases, but in those cases when its going to work, the blend will be much better with a far less controlpoints.
I’ll attach the fileexample.3dm (220.8 KB)

How do define