[Plugin][$] Curviloft 2.0a - 31 Mar 24 (Loft & Skinning)

fredo6

@thomthom said:

:thumb:
Very nice Fredo.
I'll be looking into the revised definition of a Quad this week. I like your suggestion.

I will release an update with

Border: Soft + Smooth
Diagonals: Soft + Smooth + Hidden

@thomthom said:

(Btw, I see that the preview still display black meshes - you know that SU8 now makes use of view.drawingcolor, yes?)

I know, but for Curviloft, I'll keep Black anyway (due to the visual matching with the various colors of the lines).

Fredo

fredo6

NEW RELEASE: Curviloft 1.2b - 26 Sep 11

IMPORTANT: you need to install LibFredo6 4.2 (or higher). With this version of LibFredo6, the ZLoader__Curviloft.rb is no longer necessary.

This release includes more options for the generation of pseudo-quads for usage by the new QuadFaceTools by Thomthom.

Options are available in the Default Parameters dialog box
Curviloft Quads Param.png

By default, pseudo-quads are generated with:

Borders: Soft + Smooth
Diagonals: Soft + Smooth + Hidden

This allows to have a nice-looking surface, while being compliant with all functionality of Thomthom quads tool (which should be upgraded soon)

See main post for download

Fredo

thomthom

FYI to users of QuadFace 0.3 and earlier, you must set the border edges to not be soft.

So for QuadFace 0.3 and earlier:
Diagonal: Soft + Smooth (Hidden may be set - not required)
Border: Not Soft (Smooth and Hidden can be set, but not a requirement)

QuadFace 0.4
Diagonal: Soft + Smooth + Hidden (All required)
Border: Not diagonal. (Any combo can be used as long as not all the checkboxes are ticked.)

Dave R

Thank you Fredo.

@thomthom said:

QuadFace 0.4

Thomas, I see links to 0.3 but not 0.4. I must be looking in the wrong place?

Carrozza

Thanks for this great plugin! ^__^

thomthom

@dave r said:

Thank you Fredo.

@thomthom said:

QuadFace 0.4

Thomas, I see links to 0.3 but not 0.4. I must be looking in the wrong place?

Yes, you must look on the harddisk of my private computer for version 0.4.
( There's a teaser video in the first post: http://forums.sketchucation.com/viewtopic.php?f=323&t=39442 - Toob: http://www.youtube.com/watch?v=NY-l93C-fPw )

Ben Ritter

Fredo, thank you again.

derei

One question, this version doesn't use anymore ZLoader__Curviloft.rb file? I downloaded it and I found just the directory inside.

Thank you.

fredo6

@unknownuser said:

One question, this version doesn't use anymore ZLoader__Curviloft.rb file? I downloaded it and I found just the directory inside.

Thank you.

With LibFredo6 v4.0 and above, the ZLoader__xxx files are no longer used.
You can purge them by using the menu Windows > LibFredo6 Settings... > Purge Obsolete Files

Fredo

derei

thanks, I did it!

ivreich

Hi Fredo

I'm not sure if this question has been answered before, but is there a way to ensure all segments in a loft have the same number of columns of quads and that all the junction curves meet up? I have played around with all the various lofting options with no success.

For example, in this image below all three profile curves have 30 segments but the resulting mesh has disjointed junction curves, making it hard to use thomthom's loop/ring selection tools properly. Also, some of the quads seem to be merged into a single long surface?

Cheers,

Joel

All 3 profiles curves have 30 segments but junction curves do not match up

fredo6

Joel,

Sections are computed idependently for the vertex matching, so if contours are very different, you may end up with the situation you describe.

So far, the only possibility is to use the Global F-spline options. It draws a mesh between the first and last contours trying to be as close as posible to the intermediate contours (but not matching their vertex).
This method is the first one of the group of two icons at the right of the Junction methods section in the palette.

Thanks to tell me if this option is useful (I included it just in case....)

Fredo

ivreich

@unknownuser said:

So far, the only possibility is to use the Global F-spline options. It draws a mesh between the first and last contours trying to be as close as posible to the intermediate contours (but not matching their vertex).
This method is the first one of the group of two icons at the right of the Junction methods section in the palette.

Thanks to tell me if this option is useful (I included it just in case....)

Fredo

Hi Fredo

Thanks for your reply. yes i did try the global F-spline options and these are the only options which work. They are definitely very useful; thanks for including them!

What I don't understand however is that the 3 profile curves shown in my example are very similar: they are actually three copies of the same uniform b-spline with slight adjustments to each. They have exactly same number of segments and the segments are all aligned approximately above each other, So I'm not sure why curviloft creates extra intermediate junction curves in some places?

Is it possible for Curviloft to gauge when all the profile curves have the same number of segments and do direct vertex matching without the extra junction curves? The reason being that the most logical way/neatest way of doing a loft is to use Bezierspline to give each of the profile curves the same number of segments before lofting.

Also, any idea about the second issue? the one where the quads are all merged into a single long face? Sorry if I'm a bit long winded, but I'm just very excited about the integration between Curviloft and Quad face tools...

Cheers,

Joel

fredo6

Could you post the model so that I have a closer look

Thanks

Fredo

jgb

@ivreich said:

They have exactly same number of segments and the segments are all aligned approximately above each other, So I'm not sure why curviloft creates extra intermediate junction curves in some places?
Joel

I too have noticed (and remarked about a long time ago) CL's penchant for adding multiple vertexes where none should be logical.

CL does a fantastic job of lofting, but often adds a lot of unnecessary lines that I spend a lot of time simplifying by simply deleting the redundant lines and triangulating the quad face. It is rarely a logical midpoint on a straight-line split, they seem to me to be almost random in generation, both position and number.

I haven't tried the new version yet, and I won't have time for a while, as I only get less than an hour on the PC a day due to a temp. 12 hour a day job.

ivreich

@unknownuser said:

Could you post the model so that I have a closer look

Thanks

Fredo

Hi Fredo

I've attached a model showing the problems I described. I've showed the exact steps i took and colour coded/labelled the problematic areas.

Thanks so much for looking into this!

Cheers,

Joel

Model Screenshot

CL Test Model

pilou

Have you same result when you scale the B Splines *100 ?

fredo6

@ivreich said:

I've attached a model showing the problems I described. I've showed the exact steps i took and colour coded/labelled the problematic areas.

Joel,

I had a look and there is definitely a problem in the vertex matching. Actually, even if the number of vertices are the same, thereare some which are very close and unequally spaced. This create issues with the current matching algorithm which still needs to be improved in case there is no 'sharp' edges to be used as a reference start point.
Curviloft Matching vertices.png

In addition, the problem of vertex matching makes the Global F-Spline method not really work. But I knew I had problem on this one.

I am afraid that this cannot be fixed instantly. What I suggest is that you play with the vertex matchning by dragging the junction curves and eliminating the auto-pairs.

Fredo

ivreich

@unknownuser said:

I am afraid that this cannot be fixed instantly. What I suggest is that you play with the vertex matchning by dragging the junction curves and eliminating the auto-pairs.

Fredo

Thanks for looking into it anyway Fredo. I guess this issue of vertex matching becomes very important now that there is a lot of buzz going on around thomthom's quad face tools which require a regular set-out of quad faces in order to properly utilise features such as loop and ring select etc.

My dream is that one day this kind of quad-based system will lead to a free-form paneliser tool for SU. I posted a topic on it here:

http://forums.sketchucation.com/viewtopic.php?f=323&t=39992

It has huge architectural implications for architects like me who use SU heavily as a design tool. Not many people seem to be interested in it though...which is rather strange.

Cheers,

Joel

konara

thanks freddo ...!!unbeleivable..