Hi Michael,

I tried the 2 pass approach using curve instead of surfaces and though I had to figure out how to do it, it was a bit tricky , I managed to create the button the way I wanted and position it on the target surface.

This idea of creating a kind of 2.5D object (basically a flat object (2D) with a little thickness) bending (flow) it in 1 or 2 of direction and kind of applying it where I want is an operation I would use quite often in my work.

In the joined file you'll see in front row the setup I've created, from left to right, the button, the base lines, the target curves and surface and "crossair" where the final flowed button will be positioned.

The second row shows the setup for the first flow, you'll noticed that I rotated (top view) the button in the center of the target curve.

The third row shows the setup for the second flow, you'll noticed I've rotated the button 90 degree in front view before using flow.

The fourth row shows the button rotated back to its proper orientation.

The fifth row shows the button properly "oriented" on the target surface.

I suppose all this can be automated in the future and I hope I'm not the only one that would use such a series of operations. Maybe you have already something in mind to facilitate this kind of 2 curves bending.

Even if this specific application never happens I believe it's much easier to do then having to create the button in place directly on the surface. Imagine a much more complex shape like something out of Zsurf for example.

This flow command is simply awsome, I'm sure many user will find many use for it. Excellent work!

Thanks,
Felix

Hi Felix, that's a great result!

Yes, using flow mapping from a base curve to a target curve can make it a bit easier to apply a bend only in one direction at a time and minimize distortion.

> Maybe you have already something in mind to facilitate
> this kind of 2 curves bending.

Well, just using the flow command twice in sequence seems like it does the job for this case quite well already.

Do you mean having some special option just for the convenience of a couple of less clicks?

It's not generally very good from a UI design standpoint to stick in a whole ton of custom options for doing specific things that can already be done pretty easily with just a couple of actions in sequence - that tends to lead to UI bloat making the UI get overly filled up with specialized stuff and a UI that is totally chock full of stuff is more difficult to use.

See this previous post for some more explanation on this UI problem:
http://moi3d.com/forum/index.php?webtag=MOI&msg=4442.19

The philosophy of "more is better" does not really apply very well to UI design, at least in the case of trying to keep things easy to use which is a big focus for MoI.


So if you are able to get a good result for this case by using the technique of using flow twice in sequence, then that's what I would consider to be a successful setup already.

What you're talking about there should amount to only a few seconds to apply the flow command a second time, right? Shaving a few seconds off of a construction technique like this is not really a significant improvement. Compared to all the other setup that you're already doing it's not really a significant stumbling block and all together the overall technique is saving a whole lot of time compared to constructing the warped object directly without any use of flow...


But you could probably use surface to surface flow to expedite this if you made a target surface that was not pinched together - that would be something like take these 2 crossed curves that you have here:




You can use the Sweep command to build a curved surface from those that is more uniformly shaped and not pinched near the top or bottom, that should help to minimize distortion in surface-to-surface flow.

Select one of those curves for the sweep profile, then run Construct > Sweep and select the other curve as the rail, that will build a curved but not pinched surface that looks like this:



This kind of surface that does not have pinched together areas in it will reduce distortion when using it as a flow target.

Basically if you want to do a 2 directional flow, that is what the surface to surface flow will do already, just try to make the surface have a more uniform semi-rectangular layout to it and don't pinch parts of the surface to be narrower or wider than other parts.

- Michael

Hi, I dont know if anyone asked this before but I think an optional color-coded-normals could be useful, cause when I export surfaces to modo sometimes they're seeing the wrong side, just a suggestion.

Hi Ed - in Modo if you set your material to be double sided it shouldn't really make any difference which particular side is the positive normal direction side.

Is there some reason why you don't just set it up like that so that it doesn't matter one way or the other?

- Michael

Hi, there is a reason: some translucent materials and fur materials doesn't work with the double sided option, but if there is too much work needed to make it possible... maybe it is not that important.

Hi ed, well I do want to have some mechanism for seeing and altering the normal direction while inside of MoI.

It just has not been a real priority yet since for the most part I have figured that if your rendering application has some features that are sensitive to the normal direction then they would also probably have tools in there to do the flipping inside that application.

It kind of turns out that is maybe not a correct assumption though since it's not uncommon for there to be bugs in rendering applications with handling imported vertex normals while doing the flipping. They just have to flip the vertex normals as well as the face normals, it's not particularly high technology stuff so you'd think that it wouldn't be a problem but, ....

Do you often put fur on renderings of objects that you've created in MoI? Most of the time I'd think that fur would go on character type models and not really on the kind of stuff that is typically modeled with MoI.

- Michael

Hi Michael,

I knew somehow this one seem to easy. I tried sweep as you suggested but it is almost as complicated because in order for the button to fit on the original target surface, I have to create each curve for the sweep exactly the same length of the side of the base plane, this prevents any scaling (still have to flip normal). Lastly for some reason the button get rotated about the normal of the 2 curve used for the sweep where they intersect (see joined image, red line). Just in case, I've rotated the button, the seam was vertical where the curve shows at the top of the surface.

> Do you mean having some special option just for the convenience of a couple of less clicks?

No, I was thinking more of something like a custom script and I think I can create it myself someday (and share it).

I have a passion for furniture making and I'd like to add ornaments or carving like features to my work and not just glue on type of ornaments. This new flow command is just great for that and yes many time the target surface will be flat but there will also be quite a lot curved surface as well (legs, corner post, etc.) and somewhat less compound curve surface as in this case.

The simple button I've used here is just a proof of concept kind of thing I used to learn how to use this new flow command. Eventually, I intend to also use highmap displaced surface via Zsurf probably and use flow to make it conform to the surface I want.

Moi moi moi, I'm getting adicted...
Felix

Hi Felix,

> Lastly for some reason the button get rotated about the normal
> of the 2 curve used for the sweep where they intersect (see joined
> image, red line). Just in case, I've rotated

Can you post the 3DM model file for this case where you're getting unexpected rotation?

It could be that you're seeing the result of uneven parameterization of the sweep surface. One thing that is good about the curve to curve flow is that it is based on distance traveled along curves.

The surface to surface flow is not based on 3D distance but rather parameter space distance, and it can be possible for parameter space to be compressed or stretched in relation to 3D space. When that is the case it will result in compression or stretching of the surface-to-surface flow as well.

Right now I think that you can get that kind of uneven parameterization if you just draw a control point curve with fairly uneven spacing between the points and use that for a part of the sweep. You can use the Rebuild command for now to reconstruct a curve with uneven parameterization in it, and then using the result of the rebuild for constructing the sweep may give you a better result.

But that is one nice thing about the curve to curve one as compared to the surface-to-surface one - that the curve-to-curve one is arc length based and is not sensitive to parameterization like the surface to surface one.

- Michael

Hi Michael,

here is the file you requested.

I would also think that using curves instead of surface present advantage in cases like mine. Specially when the curves are simple arcs as in this case. But for say freeform curves, it might become quite difficult to "bend" and position the flowed object correctly, I'll see when I get there, for now lets forget about this.

Thanks,
Felix

Hi Felix, I'm not really sure if I'm following the rotation problem - there are quite a bunch of objects in that last file you posted and I'm not entirely sure which ones were used to make that particular result that had the improper rotation.

One thing that I did notice though is that in your original object, the seam of the circular part is at a different rotation than the nearest point of the star, like this:



Did something happen like you got a 90 degree rotation in the result of the Flow because the surfaces happened to have a different UV axis direction placement, and then you rotated the base object a bit to compensate? If you wanted the seam of the outside circle tube part to align with the star, that would need to be aligned in the base model.

Maybe I'm just not understanding what the rotation problem is though...

- Michael

Hi Michael
I have copied the ringcircle script in and there is an error when running it
cheers
shayne

Look here Shanyno:

http://moi3d.com/forum/index.php?webtag=MOI&msg=4441.9

Copy that new ringcircle htm file over the old one in the commands directory.

Well I often use fur as grass in my architectural renders. Aside from that, I remember once I applied a glass texture in a window and it looked black in the render, then I flipped the polygons and nothing happened, it still looked black, but as I said before and as you think, it is not a priority. Some rare times the shell command cant handle some surfaces and I leave it as surfaces and if there are a lot of surfaces its hard to predict which are which side, but it is not the day to day situation.

Hi Michael,

sorry again, I didn't realise that the rotation was actually 90 degree, I was just looking at the star and it wasn't pointing up and I jump to the conclusion that it rotated by some strange angle.

In the file (cleaned up) when looking at the seam, it starts at 0 degree and ends up at 90 degree. This is caused by the UVs orientation like you say.

Thanks,
Felix

Hi Felix - I think this will be a lot easier when you will be able to have more control over how the surfaces get aligned to one another, it should help to avoid having things getting rotated by 90 degrees unexpectedly like can happen now.

- Michael

Hi Michael,

will this kind of control be controllable itself. For example, when I used the 2 steps option with curves I had no choice but to choose and "nearest" end when what I really wanted is to have the result right in the middle? I understand that with surfaces, choosing a "corner" would likely just orient or align the target UVs with the base UVs.

Its fairly easy with arcs (curves) to match length of the base and target curve or to rotate the flowed object to where you want it but with a more "freeform" type curve it can become difficult, to reposition the flowed object. Maybe setting up history would help with this, I haven't tried but a simple target point to mark the desired center of the flowed object position on the target curve could be helpfull. But again this implies added UI complexity. There are so many thing to consider to keep things as simple an effective as possible and yet flexible enough to be used in many situation with a minimum of pre and/or post number steps.

Regards,
Felix

Hi Felix,

> will this kind of control be controllable itself.

It will be focused on controlling alignment of the UV directions between each surface - but the transformation effect will still be to map from one whole surface to the other whole surface. Like you were mentioning trying to do something more than that would bring about quite a lot of complexity.


> when what I really wanted is to have the result right in the middle?

Well, in "stretch" mode for the curve to curve flow you just position your object in the middle of the base line and that will then map to the middle region of your target curve as well.

For the non-stretch mode it will map things by distance traveled along the curve, so to get a lot of control over the placement you could do something like create a line that had a length equal to the length of the target curve and then centering your objects to transform on that line would give you the effect that you need.

You can set up a script to measure the length of a curve here:
http://kyticka.webzdarma.cz/3d/moi/#CurveLength

In the future I expect to make that easier to get at without a script being necessary, on some kind of extended properties dialog but for now that script is a way you can determine it.

- Michael

Hi Michael,

"For the non-stretch mode it will map things..." that's a good if not excellent idea, since it's much easier to create a strait line of a specific length then a curve. Better yet, I'll get exactly the result I want with minimal effort.

Thanks,
Felix

Not intending to necromance but this is EXACTLY the kind of tool I am looking for.

Great one...

a step to more organic models...