Matching tangency to less than 0.111 degrees

@Gijs , do you have any idea why GEC can’t display any value lower than 0,1 when I try to improve the tangency in this example? (the control point on the right side responsible for the G1 tangency)?
No matter how I move the control point along the normal direction (up or down), the value is never below 0,1 degrees and the graph is always very tall at the very end.

Global edge continuity - can’t lower the G1 of 0,1 degree.3dm (817.4 KB)

@Rhino_Bulgaria the 0.111 degrees is the minimum angle possible with the current point layout. This is caused by the tangents between the end point and the previous point matched G0 to the target.
Through Rhino MCP/Claude I was able to improve the result, see attached document. It had to move the first and second row of control points marginally to get an almost perfect G1 match

Here you see the points it moved and the amounts (at most 0.035 units moved)


matching-improved.3dm (1.2 MB)

The automatic result of “Match surface” was much worse, this is why I improved it manually via the Gumball. However, the side control points were matched to the green curve which is perfectly matched to the target surface at G2 and 0,0000000 degrees deviation. No idea why the end row of control points coinciding with the green curve was not able to fall under 0,111 degrees.

Anyway, I will examine the file you uploaded and further investigate the issue.

Yes the U direction is matched perfectly, it’s the V direction that affects the match not reaching less than 0.111

It’s unfortunate that Rhino’s “Match surface” actually makes the deviation worse rather than to improve it. The Rhino MCP/Claude result is much better.

I managed to slightly improve it, but it’s just a very minor 0,001 degree improvement, going down from your 0,016 degrees to 0,015 degrees.

Since the lowest possible setting for “Drag strength” was limited to 0,001 mm, at some point I was forced to use “MoveUVN” with a much greater precision of 0,00001 mm and 0,000001 mm.

matching-improved Bobi.3dm (1.5 MB)

How about 0.008 degrees:


matching-improved2.3dm (596.8 KB)

Your example shows the urgent need to re-write the entire code of “Match surface” while Rhino 9 is still in BETA, because the current implementation is extremely flawed. If you move any control point along the matched edge in your example, you will notice that the whole control polygon gets extremely distorted. Or, if you apply “Match surface”, both G0 and G1 will become way worse.
“Match surface” was neglected for more than two decades. It even can’t keep the control points in their position in your example (and in many other cases). It still can’t match the FLOW of the target surface when using G2. It can’t match the 3rd row of control points only along the normal direction. And so on… You should attach your 3dm file to the YT devoted for improving the “Match surface” tool as a reminder how bad it performs at the moment.

That is exactly how the control points of a single span NURBS surface work.
Every control point affects the entire surface except for edges.
Control points along an edges of the network also affect the entire length of the corresponding edge of the surface except for the corners.
A control points at a corner also affect the corresponding corner and adjacent edges.
If you want control points to affect only part of the surface and edges then you need to use a multi-span surface.
This is how NURBS math works. It is not affected in any way by MatchSrf.

Have you ever tried to move any the matched control points by just 0,00000001 mm and see what happens next? The History is doing crazy things.
The “Match surface” tool performs equally bad in this example, making the G0 and G1 much worse instead of improving it.

Lets see if a tangency below 0,1 degrees could be achieved with a lesser amount of control points.

0,005 degrees with less control points achieved by the upcoming “Blend studio” plug-in made by a very talented programmer (not me):

matching-improved comparison.3dm (1.6 MB)

Tell us more about this plug-in.

Any ideas when this is due to be released?

Hi everyone,

BlendStudio will be released when I feel comfortable with it, and when it works exactly the way Bobi and I want it to. When it’s ready, I’ll open a separate topic here on the forum with examples and videos. Until then, Bobi’s videos show where it is right now.
A big thank to Bobi for all the testing, ideas and honest feedback, and for sharing the videos from his tests.
Thanks for all the interest!

Sorry, I didn’t want to put any pressure on you.

The plugin looks extremely interesting, just what is missing to make Rhino better at quality surfacing.

Keep up the good work!

Hi Waiduci
Take your time—quality comes first! BlendStudio looks very promising, and advanced manual control over surface blending is exactly what Rhino has been missing for years. Huge thanks to you and Bobi for your hard work in addressing this long-standing gap. Looking forward to the official release!

Here I play with some of the settings of the “Blend studio” plug-in regarding the customization of the UI. The colours, the size of control points, the visibility of the individual elements, and display precision for G0, G1 and G2 could be changed separately at any given moment depending on your needs.

There was about one minute of freezing between the 6:22 minute and 7:25 minute, because the plug-in tried to find the best possible re-trimming of both target edges to minimize the G0, G1 and G2 deviation. The calculation takes significantly shorter time when only one of the target edges is trimmed. That option could be set manually for an instant result, and from my testing so far it performs best when the number is set between 0,02 mm and 0,1 mm deviation from the original target trimmed edge (or curve projected on an untrimmed surface). This is why the plug-in shows two different values for the calculated G0, G1 and G2 (before and after re-trimming).

Some more testing of the “Blend studio” plug-in. There is a new setting for the precision of the automatic calculation of the best possible deviation from the target curve on surface used to re-trim the target surface. Priority is given to G0 first to find the least gap, then moving to finding a solution for G1 and G2. The calculations take significantly less time now.

The blend surface is built directly onto the target surfaces, so the latter don’t need to be trimmed preliminary.



@Gijs , at the 2:41 minute of the last video I noticed that something weird is happening with the Zebra analysis. Even though the menu says “Stripes” (set to black and red colours, despite that I never use red) and “Static” is not selected, the Zebra analysis visible in the viewport shows the Rainbow mode with multiple colours and obviously it’s static. Then I move the “Frequency” slider and the rainbow stripes change accordingly. I can repeat that visual bug every time when I re-open the same 3dm file. I attach it here just in case:

Loft plus Match surface.3dm (7.4 MB)

Hello everyone,

Just a note to add that there will also be version of blendstudiocrv, for more precise controll, it will be released together with blendstudio, just quik show on link:

The post above mine is flagged for some reason. Was it because it contains a link to a video demonstration of the “Blend studio curve” plug-in, and the user account is new? The video is quite interesting to watch.

Since the vertical true arc extrusion is degree 2, it has weighted control points in the middle. “Blend surface” was unable to ignore the weighted nature of the lower target edge, hence almost all of its control points became weighted. While changing the control point weight helps to improve the G0 and G1 of the blend surface, it generates weighted surfaces that cause issues with the adjacent surfaces later.