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

TIG

mac1

I've just made EscapeArtist's example.skp into v7 - try downloading it again from the same post... http://forums.sketchucation.com/viewtopic.php?p=309006#p309006

EscapeArtist

@tig said:

mac1

I've just made EscapeArtist's example.skp into v7 - try downloading it again from the same post... http://forums.sketchucation.com/viewtopic.php?p=309006#p309006

Thanks.

mac1

TIg
Thanks.
As a point of clarification. I am not implying there is any thing wrong with the plugin. I found cases where I have made simple models that show this character and think it is caused by my sloppy approach of applying some of the constraints one must follow when modeling especially when using the follow me tool. I have a feeling if the suggestions made by Fredo are followed this my not show up or its the hand springs I am going through to get the model.
Thanks again

mac1

Escape Artist
Some observations ref your model for what they are worth:
1)I was expecting to see the bulk head curve length for what I thought was the keel line to each side to be equal but in most cases there is almost a 2 to 1 difference. Not boat designer, is this normal??
2) At bulk head number 2, counting from stern as #1, this is an obvious kink in the "keel" curve fit just before #2. This IMHO maybe causing some problems. Better tangent here my help??

Still seeing the self intersecting faces in the SU model but only 1( 4 to 5 range before) after the deletion of duplicte vertices which are now 1714 vs 6983 so my hand springs to get from SU8 to SU7 causing some problems.
You may want to run ThomThom,s clean up tool to see if it helps any. That can be done with in SU. It will not give same stats as quoted above. I used TIG's extrude by rails and it gives the same ripple in the surface at the area mentioned above as Fredo's tool does and reason for looking there.
I will be doing more checks and will let you know if I find anything more of interest
Good Luck
UpDate Date 3 Apr

1) Yes I noted shortly after post the model is only portion of Hull;
2) Self intersecting faces traced to dae and /or kmz conversion I do to get into mesh tool. Kmz gives same result but if dae does not have the option of triangulate set then it does also. With it set see only very few and these can be deleted and corrected in SU
3) Attempting to better fit the chine using a cubic spline interpolation did not give any significant changes in surface result

jgb

While playing with Curviloft (CL), drawing a practice non-scaled air duct for one of my airplane models, I noticed something about how CL handles some defining elements.

My model starts with a tilted ellipse, travels straight back then up over a spar forming a "D" shape (flat side of "D" down), then transitions down to a circle at the engine intake.

I can't show a picture as my old GPU produces solid black export JPG's with SU-V8 due to SU-V8 needing OpenGL 1.5 and my GPU can only handle OpenGL 1.3, so it has to be words only.

The shape was formed with an ellipse, then the "D" shape and a circle, with a long curve depicting how the shape will transform, essentially along the centerlines of the 3 defining shapes. My first tries came out pretty good, taking about 5 seconds to plot the surface. Compare that with about a week's work trying to get that shape by manually tweaking a mesh, and never getting it right.

Anyway, the transition from ellipse to "D" was too flat. I played with the transition curve a bit but was not satisfied. So, I needed to add an intermediate shape on the transition to more circularize the duct. That would be a circle with the bottom slightly flattened, somewhat egg shaped.

As usual, I created the shape on its 1/2 side, then copied it, moved left, flipped and rejoined the 1/2 copy to its original to form the symmetrical duct. Keep in mind, neither 1/2 shapes were true curves, but were just joined line segments. This became important!

I put CL to work and 5 seconds later got a twisted mess between the ellipse to the new shape to the "D" shape. The back duct formed properly.

Nothing I could do prevented that. If I took the intermediate (egg) shape out of the CL definition, the duct formed properly. Put it in and it twisted.

Upon close inspection I noticed that most of the twist started from the right side of the ellipse, went to the left side of the "egg" then back to the right side of the "D". But some parts of that duct formed correctly.

So, after some experimentation, I used the WELD.rb on the copied 1/2 side (without facing) and LO the duct formed correctly.

Some more experimenting confirmed this. CL will form a surface on drawn curves even if they are not "proper curves". But if you copy and flip a non-proper curve (set of line segments) CL will twist the shape as if the copied curve retained parts its pre-flip orientation.

So, if you are getting twists in your surfaces, and any parts of your defining shapes are copied and flipped, make sure you WELD the copied shapes to form a full proper curve.

thomthom

@jgb said:

I can't show a picture as my old GPU produces solid black export JPG's with SU-V8 due to SU-V8 needing OpenGL 1.5 and my GPU can only handle OpenGL 1.3, so it has to be words only.

"Print Scrn" ?

jgb

@thomthom said:

@jgb said:

I can't show a picture as my old GPU produces solid black export JPG's with SU-V8 due to SU-V8 needing OpenGL 1.5 and my GPU can only handle OpenGL 1.3, so it has to be words only.

"Print Scrn" ?

Good Idea, good thing I thought of it.

I'll try tomorrow am early. I have a bit of time, and I will attempt to recreate the problem/solution and Print Scrn it.

mac1

@jgb said:

@thomthom said:

@jgb said:

I can't show a picture as my old GPU produces solid black export JPG's with SU-V8 due to SU-V8 needing OpenGL 1.5 and my GPU can only handle OpenGL 1.3, so it has to be words only.

"Print Scrn" ?

Good Idea, good thing I thought of it.

I'll try tomorrow am early. I have a bit of time, and I will attempt to recreate the problem/solution and Print Scrn it.

It is always better to post your model vs jpg, lots of info lost in jpg. You will get better results with model so post in 3d ware house

fredo6

@jgb: as suggested, it would be a good idea to post the skp model so that I can have a look. In principle, Curviloft is not sensitive to Sketchup curves as it works on edges. The only benefit of curves is for the selection process.

Also, when you have twisting, you may try to untwist, by clicking on the section. You get a popupu window and then you can change the value of the twist angles.

Finally, which tool do you use: loft by Spline or Loft Along?

Fredo

jgb

As I only have about 1 1/2 hrs to get on my PC in the mornings for the last and next few weeks, I did not have time to fully recreate and document the drawing to post here. Not even time to do a screen print this morning.

Part of the problem was replicating the initial conditions that caused the twist, then the fix. Trying to setup the scenario in process order, sometimes the fix would work, sometimes not, even with copies of the real fix.

So I will get more time on Sat. to get this working. Please bear with me.

Fredo; I have not gone up the learning curve enough to play with twist angles, as well as other niceties in the app. As to the tools used, I created all my curves by hand, not using Splines. As this was a quick and dirty practice model, I just used circles and stretched them to get close to a shape I wanted. In a few, I needed to explode the curve and manually reshape it.

Also, I did not use the built-in selection of curves by number. I found it a bit confusing (No, I did not read your very excellent doc fully, yet), I always try things intuitively, before resorting to reading.

But here is how I did the curve selection, discovered by accident. I simply selected all the control curves and the disconnected path, hit the "Create Loft junctions following a given path" and the curve formed. I am simply amazed Curviloft can discern the correct path.

fredo6

@jgb said:

Also, I did not use the built-in selection of curves by number. I found it a bit confusing (No, I did not read your very excellent doc fully, yet), I always try things intuitively, before resorting to reading.

No shame! I did not write a documentation (This is why the plugin is still in Beta)
The best way is to have a look at the video and explore.

Fredo

jgb

OK, finally got time to show you what happened.

If you open the model, you will see 3 sets of curves. The 2 sets on the right are copies of the first set on the left. Make visible the layers, 1 at a time as I explain what I did and what happened.

Twisted Curve Demo

Layer 1. The main curves of the duct. The intake ellipse transitions up over a wing spar then down to the circular engine intake. It is not to scale.
Layer 2. The duct curve to follow. In 2 parts, forward and aft. The aft curve formed OK when it was separated from the forward part, so I left them as separate pieces.
Layer 2.1 The 4 curves I selected to form the duct. I did not define which was the path, just clicked on "Create loft junction following a given path". Like I said, I am amazed Curviloft could discern the path.
Layer 3. These are the lines I wanted the duct top and bottom to follow. They were never selected for Curviloft and are an approximate shape only. I used them to create the intermediate curve to eliminate the flat underside seen in...
Layer 4. The resultant first workable duct. The transition was way too flat. The duct needs to have a near constant cross section area throughout.
Layer 5. Moving to the middle set, I created a 1/2 D section by making an arc, then exploding it so to adjust the shape by eye. Then I copied the lines, moved them left and flipped them. Next I moved the copy right to meet with the original set. I've done this hundreds of times before.
Layer 6. Selecting the 4 main curves and the new transition curve, produced this twisted duct.
You will see that the lines seem to go from the right side of the ellipse to both the left and right sides of the transition, then back to the right of the over spar curve.
Layer 7. When I welded the transition curve into a true curve....
Layer 8. The duct formed correctly.
Layer 9. The aft duct formed OK.

Note a few things. When creating this demo, I had difficulty reproducing the twisted duct. The version you see here and may not be able to recreate, was from the original trials on the left set.

Also, a correct duct may not be reproducible. I had trouble reproducing it from the copied set of curves.

So that's it for now.

mac1

See if this is close to what you want. I left portion for your exercise
I put all your primitive geo back to layer 0, created faces for each of the profiles, intersected the center line ref with each of those and used Freedo's tool loft junctions. It looks like you need some more profiles aft
http://sketchup.google.com/3dwarehouse/details?mid=7da88c5f8c6cdc29427d16763a0ff49

jgb

Thanks guys, and for the PM "solutions" as well. I will look at them later this week, as I have no time just now.

Actually I'm not looking for a solution. I have it in the rightmost copy, and it is good enough for my learning purposes. I will not use that particular duct in any model, as it was not to scale, nor to fit the actual airplane at this time. The "real" one will be shorter and the curves a bit more involved to avoid aircraft structure. It is simply a learning exercise for Curviloft using a shape I had difficulty creating using a mesh and tweaking it.

The essence of my post was that it seems Curviloft tried to follow the original DIRECTION of the flipped copy of a curve in segments causing a twist in the surface, and was for Fredo to analyze. Once I welded the flipped curve segments, my duct formed correctly.

rock1

wow,what a plugin fredo!

I should have joined this forum earlier. I never knew about it. What a cool collection of plugins is present here!!

Rich O Brien

Hi Fredo,

Can you look at this file?

CurviLoft_Helix.skp

Using 'Skinning of Shapes' but slight errors appear?

jeff hammond

@unknownuser said:

[attachment=2:o25qv8mt]CurviLoft_Helix.skp[/attachment:o25qv8mt]

Using 'Skinning of Shapes' but slight errors appear?

you have to switch the properties..

for instance, try the same thing with tig's extrude edges by rails.. use the helix lines as the rails and the little shorties as profiles (well, make them curves first).. you'll see a similar though actually worse error.. then, do it again but choose the helix as profiles and the little lines as rails.. all is better.

to do that in curviloft, click in a black area of the preview skin to bring up this dialog:

[attachment=1:o25qv8mt]001.jpg[/attachment:o25qv8mt]

then use the top left button to re-orient
[attachment=0:o25qv8mt]002.jpg[/attachment:o25qv8mt]

Rich O Brien

Thanks Jeff.

Strangely I did use TIG EebyR it it worked perfectly. Though I used one shawty as both 'Profile' and 'Melding Profile'. Didn't try it with separate profiles.

Anyway thanks for the fix

tibet

hi ,
i have deleted curviloft icons by accident.
could you please tell me how to restore them in sketchup 8 pro
thanks

TIG

@tibet said:

hi ,
i have deleted curviloft icons by accident.
could you please tell me how to restore them in sketchup 8 pro
thanks
Go to the first post in this thread and download the files and install them as instructed. The icons will be restored and some other scripts etc will be overwritten - but that doesn't matter...
However, if you don't mean you've actually thrown away the files but you mean you've 'closed the toolbar' or 'deactivated the extension' then simply reactivate the extension through Window > Preferences > Extensions and/or switch on the toolbar through View > Toolbars...