Fusion
Community
What happens is this;
- so far so good;
then this;
It doesn't give me any clues as to why it can't be created. More typically, the error message will be that the loft is self-intersecting when logically it shouldn't.
Thanks for the tip about hiding the sketch, that's handy and already improves the surface around that joint.
What happens is this;
- so far so good;
then this;
It doesn't give me any clues as to why it can't be created. More typically, the error message will be that the loft is self-intersecting when logically it shouldn't.
Thanks for the tip about hiding the sketch, that's handy and already improves the surface around that joint.
Too many profiles with too many spline points.
Please share your design. Export it as a .f3d file and attach it to the next post.
Too many profiles with too many spline points.
Please share your design. Export it as a .f3d file and attach it to the next post.
Is this the same boat as in this thread ?
The usual advice is to start with fully defined sketches, meaning that they are fully dimensioned and constrained.
I might not go that far, but except of coincident constraints that "glue" the arc and line ends together, your sketches at the minimum tangent constraints.
7 Profiles is still too much to get a truly smooth surface. If you use the Inspect-Isocurve analysis and toggle on the curvature combs you'll see what I mean.
Is this the same boat as in this thread ?
The usual advice is to start with fully defined sketches, meaning that they are fully dimensioned and constrained.
I might not go that far, but except of coincident constraints that "glue" the arc and line ends together, your sketches at the minimum tangent constraints.
7 Profiles is still too much to get a truly smooth surface. If you use the Inspect-Isocurve analysis and toggle on the curvature combs you'll see what I mean.
It's not the same boat but it's the same project, and the file at the point I attached it was 'roughing out', to check that the 'whole moulding' method of profile projection was giving me a satisfactory result. The pic below is the outcome of the original discussion. I'll be more than happy to be corrected Peter but I don't think my lazy constraining is causing the problem. This is probably a very bad habit to have got into but, when I have a lot of splines, arcs, tangential relationships and so on, I tend to just 'fix' the whole sketch when I'm happy with it, and the sketches are more constrained this afternoon than this morning. One time, and one time only, I manually constrained an entire spline and would never want to do that again.
I fully accept that the more profiles there are, the less likely it is that the surface will be 'fair', and less is better. I guess I could just use some of the profiles as rail guides. But what about this strange behaviour with respect to lofting perpendicular profiles? I don't understand why it will do the loft when it only has to join the two mating profiles, but not when I include the next profile along. I'm sure I've had instances where it just does it, as well as those where it won't, and I'd love to know what makes the difference.
It's not the same boat but it's the same project, and the file at the point I attached it was 'roughing out', to check that the 'whole moulding' method of profile projection was giving me a satisfactory result. The pic below is the outcome of the original discussion. I'll be more than happy to be corrected Peter but I don't think my lazy constraining is causing the problem. This is probably a very bad habit to have got into but, when I have a lot of splines, arcs, tangential relationships and so on, I tend to just 'fix' the whole sketch when I'm happy with it, and the sketches are more constrained this afternoon than this morning. One time, and one time only, I manually constrained an entire spline and would never want to do that again.
I fully accept that the more profiles there are, the less likely it is that the surface will be 'fair', and less is better. I guess I could just use some of the profiles as rail guides. But what about this strange behaviour with respect to lofting perpendicular profiles? I don't understand why it will do the loft when it only has to join the two mating profiles, but not when I include the next profile along. I'm sure I've had instances where it just does it, as well as those where it won't, and I'd love to know what makes the difference.
The sketches in the model I have attached are not fully constrained, or dimensioned either . They simply have the "right" amount of constraints to be able to create a loft that also does not break when inputs are changed.
I used control point splines for this design.
I consider this design so far a good starting point.
This includes only three loft profiles. A (sketch)point on the front, one in the middle and one at the end.
You could now add two loft profiles, one on each side of the center if more control over the shape is needed.
This will degrade surface quality, but if done with care it should definitely be OK for a model.
The sketches in the model I have attached are not fully constrained, or dimensioned either . They simply have the "right" amount of constraints to be able to create a loft that also does not break when inputs are changed.
I used control point splines for this design.
I consider this design so far a good starting point.
This includes only three loft profiles. A (sketch)point on the front, one in the middle and one at the end.
You could now add two loft profiles, one on each side of the center if more control over the shape is needed.
This will degrade surface quality, but if done with care it should definitely be OK for a model.
Peter has explained his result, I have a single loft from your data.
One minor change due to the loft rules, i.e. All profile articles must be in the same sketch.
Your intended rails are my profiles. Your sections sketches do not help provide a nice boat,
Pleased the section sketches did connect to your "Rails" well done.
Purple lines in this sketch (bow) were projected for use as a profile, however this curve comb shows why things are not smooth. Less is cleaner.
Edit my loft for the settings
Might help...
Peter has explained his result, I have a single loft from your data.
One minor change due to the loft rules, i.e. All profile articles must be in the same sketch.
Your intended rails are my profiles. Your sections sketches do not help provide a nice boat,
Pleased the section sketches did connect to your "Rails" well done.
Purple lines in this sketch (bow) were projected for use as a profile, however this curve comb shows why things are not smooth. Less is cleaner.
Edit my loft for the settings
Might help...
Accepted solution
No, actually I have not. I've only posted a design with a timeline, but that isn't self-explanatory.
For example I only use single-span control point splines, but I have not explained why.
A fit-point spline in Fusion 360 is a 5 degree multi-span non-uniform rational B-spline (NURBS) , a very complex mathematical object that consists of internal curve segments (spans) and definitions that keep the transitions between these spans smooth. See for example the section on B-Splines here.
A single span control point spline, on the other hand is mathematically equivalent to a Bezier spline and is inherently smooth. In Fusion 360, for large gentle curves it is also much easier to control than the fiddly tangent handles. That is particularly the case when you want to create 3D curves.
As it pertains to surface based workflows, that is the reason I prefer to work mostly with 3 and 5 degree single-span curves. I use fit point splines mostly when In need to create a blend curve between two existing edges.
So that's curves. How about surfaces ?
Fusion 360 can import higher than 3-degree NURBS surfaces, but it can not crate them. Whatever curve you use in Fusion 360 for creating a surface, the surface will be a 3-degree surface.
For the bottom profile I needed to create a curve between two (straight) lines. I used a 5-degree curve with 6 control points. For a single span control point spline, the number of control point sis always one higher than the degree.
To achieve a smooth transition between the lines and the curve I applied a continuity (G2) constraints.
There are no smooth surfaces without smooth curves and transitions! I also go through great lengths to keep any clutter out of sketches. For example I often use the vertical/horizontal constraints to eliminate construction lines. That is purely personal preference, however.
When creating the top sketch, to ensure the rail start and end coincide exactly with the 1st sketch I projected 2 points as shown into the top sketch and used those as start and end points for the rail
The top rail is not a single span spline as I was not able to represent the shape well enough with only 6 control points. It has 7 control points.
Then I surface-extruded the top profile from sketch one and trimmed it with the top rail. The resulting edge is very smooth:
In theory one could use an intersection curve, but the resulting curve has visible artifacts:
Then I created an offset plane and placed it somewhat in the center of the object. I then intersect-projected the upper surface and lower helper surface into the profile sketch to have precise points to constrain the 3-degree profile curve to.
Then I lofted, mirrored, stitched and shelled the object. I was surprised shelling worked as flawlessly as it did.
I am surprised, because lofting into a point is often a problem. Lofting creates NURBS surfaces. The shape of a NURBS surface is controlled by a 4-sided grid of control vertices (CVs) and associated weights. Lofting into a point will cause the control points of one side to collapse into that point. The weights will approach infinity and the curvature of the surface will rapidly degrade, often causing problems with thickening, offsetting and shelling.
We got lucky this time.
No, actually I have not. I've only posted a design with a timeline, but that isn't self-explanatory.
For example I only use single-span control point splines, but I have not explained why.
A fit-point spline in Fusion 360 is a 5 degree multi-span non-uniform rational B-spline (NURBS) , a very complex mathematical object that consists of internal curve segments (spans) and definitions that keep the transitions between these spans smooth. See for example the section on B-Splines here.
A single span control point spline, on the other hand is mathematically equivalent to a Bezier spline and is inherently smooth. In Fusion 360, for large gentle curves it is also much easier to control than the fiddly tangent handles. That is particularly the case when you want to create 3D curves.
As it pertains to surface based workflows, that is the reason I prefer to work mostly with 3 and 5 degree single-span curves. I use fit point splines mostly when In need to create a blend curve between two existing edges.
So that's curves. How about surfaces ?
Fusion 360 can import higher than 3-degree NURBS surfaces, but it can not crate them. Whatever curve you use in Fusion 360 for creating a surface, the surface will be a 3-degree surface.
For the bottom profile I needed to create a curve between two (straight) lines. I used a 5-degree curve with 6 control points. For a single span control point spline, the number of control point sis always one higher than the degree.
To achieve a smooth transition between the lines and the curve I applied a continuity (G2) constraints.
There are no smooth surfaces without smooth curves and transitions! I also go through great lengths to keep any clutter out of sketches. For example I often use the vertical/horizontal constraints to eliminate construction lines. That is purely personal preference, however.
When creating the top sketch, to ensure the rail start and end coincide exactly with the 1st sketch I projected 2 points as shown into the top sketch and used those as start and end points for the rail
The top rail is not a single span spline as I was not able to represent the shape well enough with only 6 control points. It has 7 control points.
Then I surface-extruded the top profile from sketch one and trimmed it with the top rail. The resulting edge is very smooth:
In theory one could use an intersection curve, but the resulting curve has visible artifacts:
Then I created an offset plane and placed it somewhat in the center of the object. I then intersect-projected the upper surface and lower helper surface into the profile sketch to have precise points to constrain the 3-degree profile curve to.
Then I lofted, mirrored, stitched and shelled the object. I was surprised shelling worked as flawlessly as it did.
I am surprised, because lofting into a point is often a problem. Lofting creates NURBS surfaces. The shape of a NURBS surface is controlled by a 4-sided grid of control vertices (CVs) and associated weights. Lofting into a point will cause the control points of one side to collapse into that point. The weights will approach infinity and the curvature of the surface will rapidly degrade, often causing problems with thickening, offsetting and shelling.
We got lucky this time.
Edit: Peter, ignore 90% of what I've written below; I've had the time to go through your file properly now and, while I haven't yet fully digested how I can apply this as I never knew you could loft from a point, I suspect you have opened the door for me to a whole new level of lofting. FYI there have been a good few debates on the ship modelling forums about creating hulls and F360 tends to get dismissed as hopeless for that purpose: but I've never seen anyone do it this way, nor get such fair lines using any other software.
Original:
Well, you've both given me plenty to chew on, some insights into NURBS and an idea of when and why I'd use control point splines rather than fit point splines. Peter, I don't want to take up your time but can you point me to something that explains the difference between 3-degree and 5-degree surfaces.
I think you've probably also explained why, sometimes, I struggle with perpendicular lofting - I think it's most likely that it involves lofting to points and sometimes you win, sometimes not. So I won't waste time trying to fix something I can't, but will instead see how both of your answers change my workflow.
Can I throw a spanner into the works - these boats are clinker built, which means the hulls are made up of (usually about 10) overlapping planks, or strakes to be nautical. Screen snip below. To complicate things a little further, they don't overlap at either end, but blend together. This kind of problem/issue becomes even more present on age-of-sail ship hulls (as opposed to little boats) as they tend to have lots of fore-to-aft features like decorative rails, wales (thickened sections of the hull) and complex bulwarks. I don't see a way to apply your 'top-to-bottom' lofting method to produce these shapes though I can see this process enabling the creation of some very accurate sections and rails to work to, via splitting, which in theory at least should make the curves of prow-to-stern lofts much more fair. And that's a major part of the challenge with hull-modelling, so still a valuable lesson.
Edit: Peter, ignore 90% of what I've written below; I've had the time to go through your file properly now and, while I haven't yet fully digested how I can apply this as I never knew you could loft from a point, I suspect you have opened the door for me to a whole new level of lofting. FYI there have been a good few debates on the ship modelling forums about creating hulls and F360 tends to get dismissed as hopeless for that purpose: but I've never seen anyone do it this way, nor get such fair lines using any other software.
Original:
Well, you've both given me plenty to chew on, some insights into NURBS and an idea of when and why I'd use control point splines rather than fit point splines. Peter, I don't want to take up your time but can you point me to something that explains the difference between 3-degree and 5-degree surfaces.
I think you've probably also explained why, sometimes, I struggle with perpendicular lofting - I think it's most likely that it involves lofting to points and sometimes you win, sometimes not. So I won't waste time trying to fix something I can't, but will instead see how both of your answers change my workflow.
Can I throw a spanner into the works - these boats are clinker built, which means the hulls are made up of (usually about 10) overlapping planks, or strakes to be nautical. Screen snip below. To complicate things a little further, they don't overlap at either end, but blend together. This kind of problem/issue becomes even more present on age-of-sail ship hulls (as opposed to little boats) as they tend to have lots of fore-to-aft features like decorative rails, wales (thickened sections of the hull) and complex bulwarks. I don't see a way to apply your 'top-to-bottom' lofting method to produce these shapes though I can see this process enabling the creation of some very accurate sections and rails to work to, via splitting, which in theory at least should make the curves of prow-to-stern lofts much more fair. And that's a major part of the challenge with hull-modelling, so still a valuable lesson.
Forums Links
Can't find what you're looking for? Ask the community or share your knowledge.
Post to forums
