Deejay chiama Italia!
Exactly 
I have an example here where Xnurbs completely fails and FillSrf produces a very weird result, while the CAD I use at work produces a heavy but still usable surface:
xnurbs-fail.3dm (304.1 KB)
@eobet weāre still in the very early phase and trying to learn how the parameters work. Esp. the stretching is something that still needs a lot of investigation. While in general, stretch should be left alone, I just got pretty nice results adding a teensy bit
Fillsrf_Success.3dm (269.9 KB)
:
I really hope there will be a proper dialog for this tool in the future, and not just command line options. I saw those and my eyes just glazed over, which is why I didnāt try them. 
Also, for the final tool, the U/V degrees and spans will be hidden I guess, because I donāt see that having any bearing at all on a tool like this (stretching and bending would be nice to have as sliders, though).
Could you add Edit support as well?
You know I only know of two tools ever in Rhino to support Editā¦ Blendcrv/blendsrf and Xnurbsā¦ (imo, every tool in Rhino with history support should support editingā¦ itās not really history without it.)
Disagree. I like/need explicit control of the degree and number of spans, but I may use tools such as this somewhat differently than @eobet.
ā¦what itās intended for.
This isnāt for doing Class-A on a four sided patch. This is a last resort tool for filling irregular shapes (openings with odd number of sides) which places its UVs at weird angle and trims back the result, right? UV is not important here, only tolerance is! Because if you canāt join it, what use is it?
I know Gijs showed an example where the UV line up with the surrounding surfaces (and Xnurbs also has such an option), but thatās the exception for this tool because the surface it creates will in 99% of cases be so heavy and weirdly matched up to its surroundings that itās impractical/impossible to edit (especially given Rhinoās tools for editing UV heavy surfaces are soo poor).
Yes this is in the works
Note that in FillSrf there is no explicit control (yet). The spans are the initial spans.
Mm, well my point was that the tool has spans and tolerance, but if tolerance isnāt met the tool has failedā¦ so why not hide the spans options (at least from the GUI when that happens) because the least amount of spans should yield the best option, shouldnāt it? And if not, the user could perhaps add a guide curve or point instead?
Iām just trying to suggest optimizing the tool for cognitive load on the user (cue someone saying something about āpremature optimizationā).
Btw, I like the new color scheme on the icons! 
Especially that new purple on the meshā¦ where have I seen that before? 
Well in general, yes that is correct, the tool will only add spans if necessary. This is done in the refinement steps. The initial spans however allow you to add spans more symmetrically. Itās very comparable to the initial grid quads in meshing. Sometimes it helps to give the tool a few more starting spans.
Itās too early to say if these can be added automatically, but this is certainly an interesting point that I will discuss with Menno. The goal is to eventually find a good balance between control and simplicity.
Good to hear! I like it too. The new schemes (and icon rework) is @marika_almgren 's effort
@menno amazing work
@Gijs so finally i will get a vanilla Rhino solution for the 70ās modelling challange / rowenta ek 55 ??
the semester starts here next week and I will ask my students to find some nice, additional test challenges. maybe we should have a separate topic āWip - fillSrf challangesā ? - so this topic keeps the focus on āfeatureā
kind regards -tom
I checked it out the other day and FillSrf gives a very similar result.
@eobet Can you maybe share the āvery weird resultā and the settings used to obtain it? Iām interested to learn about any peculiarities.
Iāve been trying this with a few different settings:
Default settings
Default settings with Domain=Projected
Domain=Molded with stricter tolerances: position 0.001; Angle 0.5
Yes, this is definitely what weāre working towards. As weāre getting more experience with this new way of making patches, weāre learning what does and does not work with default settings.
So, ultimately I hope we can do away with, or at least hide initially, most of the options that are now under Advanced. Current thinkng is you get to change tolerances and use a slider that influences the shape somewhat (the slider will be a heuristic that sets stretching, bending and rocBending).
@eobet I was able to use Xnurbs after adding a new surface in the area where the three surfaces meet. One problem was that, according to the analysis, the two surfaces forming an āarcā were in G0. Anyway, I tried using the Xnurbs result, but the surface it generated is still terrible.
Round1
Rhino 1 vs Xnurbs 0
Fillsrf_Success_xnurbs test.3dm (628.1 KB)