Hadn't thought of that, but, in my case it won't work. I have many of the same component strung out along the length and breadth of the aircraft. These would be skin panels, frames, passenger seats, washrooms, etc, as well as structural parts and equipment. So moving the components origin to the SU origin is unworkable. It would also make the bounding boxes so large that it would be a visual nightmare trying to select individual components with all the overlapping bounding boxes. It would also require me to change the comps axis position every time I moved the comp.

Also, changing the comps local axes point won't help either, especially when the comp is rotated a bit in final placement. Regardless of rotation, I will still need to measure every components location of its most forward point from Datum Zero and then measure its CofG offset from that point as well.

If your CofGravity.rb could measure the distance from SU origin in X-Y-Z (Green-Blue-Red) to the comps computed CofG then that would eliminate any measuring and would give me a more accurate moment offset.

CATIA does this moment offset (distance from Datum Zero to CofG of part) automatically and generates the moments table as the parts are defined (as solids with material defined). The table is dynamic, so if parts are moved, the CofG of the airplane is updated in real time, but then CATIA is purpose designed for aerospace and costs $35,000++ a seat. A bit too rich for my blood.

Tig

CofG has become a MUST Have for me lately. I am starting to map out the CofG for a few hundred groups and components on 2 major aviation models. Grouping helps but I have a serious need for an improvement to CofG.

Some background......
To compute the CofG for an airplane, I need to locate each of the components CofG in relation to a common point of origin, called the "Datum Zero" line, which is almost always the nose tip, but could be offset fore or aft of that. Multiply the distance from that origin to the CofG for that component gives a moment arm value. Sum all the moments values and divide by the total weight gives the airplanes center of gravity, which has to lie very close to the center of lift.

Then calculating the variables of fuel, payload, etc and the control surfaces ability to vary the center of lift gives me the operating envelope, or an indication of what needs to be moved to put the CofG and CofLift within the operating envelope.

Right now I have to measure the offset of the CofG from the components most forward point, and the distance of that forward point from Datum Zero. Those values, and the components weight go into an Excel spreadsheet. A significant problem is when my components bounding box is rotated away from normal, making the finding of the most forward point difficult.

Grouping helps but I have a serious need for an improvement to CofG.

Would it be possible to map the components CofG offset (X-Y-Z) from the SU axis origin point, as an option?

I can then place the models Datum Zero on the SU Axis Origin point, and it would vastly simplify my task. I would only need to enter the offset and weight into my Excel sheet, and not have to measure anything.

Thanks.

Here's a tip to cover that situation. When you draw guidelines parallel to an edge that are supposed to lay "on face" start your guideline on the edge, move the cursor up/down a bit on the edge and then pull towards the direction you want. Nothing new here.

But, instead of leaving the cursor on face where you want the line to be, move the cursor to some alternate perpendicular (or so) edge on the same face. This forces the guide line to stay on face. At that point you can enter the offset distance in the VCB or move along that edge to the desired offset location.

To test if 2 perpendicular (or crossing non parallel) guidelines are actually intersecting, place the cursor on one GL and move it to the intersection. You should see a little red X. That indicates GL intersection.

If all you see is a colored dot, the lines do not intersect. The dot means the point at the cursor lines up axially with both GL's.

This works with GL's and edges as well.

@thomthom said:

Problem is that you cannot poll keys within the SU API.

OK, I stand corrected.

Key polling is a fundamental function. Can't believe those who designed the API left that one out.

@dave r said:

@jgb said:

Tig

A suggestion if I may.

...

JPP needs a cancel or quit button.

Are you suggesting this should be done by TIG?

Sorry, No slight intended.
I sometimes get confused which Ruby Magician did what.

Also; In my experience (dated for sure) Key Down Polling inside an outer or mid loop should not really affect performance, unless you are forced to use a nested call.

@dave r said:

You'll have to unsoften that edge to separate the rectangle from the rest of the surface. You can do that with the native Line tool or the Line tool in ToS and just connect the ends of the adjacent edges of the rectangle. Or you can display Hidden Geometry, select that edge and unsoften it.

Faster; easier ..... Shift+Ctrl Erase (tool) Just click on the soft "missing" line segment.

Tig

A suggestion if I may.

Sometimes JPP can take "forever" to complete, even seem like it's hung. Other times it zips along in a sprightly fashion. Seems no connection to complexity, I've had a 30 face push take about 3 minutes and a 150 face push take about 40 seconds.

However on more than one long push a very few seconds into the push I realized I was in feet and not inches, so the result would have been scrapped in any event. I normally push/pull the pre-surface well beyond the dimension I want and key in the actual dimension, then hit enter twice, so I don't get a visual clue to my error. Other times I just don't want to wait what will be a very long wait.

JPP needs a cancel or quit button.

Cotty & Trion

The reason I would split the wing into 2+ parts is to preserve the chord section along the inner span section.

Doing it in 1 go (per side) gives a nice looking result, but the chord tends to be flattening towards the outer section and wing tip.

That is NOT a fault of Curviloft; it's just the way it works going from a "shape" to a point. Intermediate proportional "shapes" result in a better overall curve.

This may not be relevant to a model for "show" but could be critical if accuracy is important. Such a case would be if drawing internal wing structure was part of the model, rather than just an outward appearing skin or mold line.

@box said:

@jgb said:

You gave up 3 episodes of ST Voyager to do this

No no not at all, one eye on model one eye on seven of nine.

@jgb said:

Seriously, what pluggin is that?

Tig's extrusion tools can be found here.

http://sketchucation.com/forums/viewtopic.php?t=25362

Two things wrong with your excuse......

1. You can't do 3D with only one eye.

2. Only one eye on her???? What is wrong with you

Trion

I concur with Fredo.

What you need to do is take a copy of the chord at the root and scale it to (at least) 1 other rib location near the tip, but just before the tip curve starts.

Then CL skin individually the inner top and outer (tip) top airfoil, then do the bottom in 2 goes as well. Only explode the surfaces after all curves are satisfactory.

Should take all of 5 minutes.

If you have a problem, come back here with your SKP file and I will do a detailed explanation.

You gave up 3 episodes of ST Voyager to do this

Now that is dedication to SU.

Seriously, what pluggin is that?

@box said:

You got me curious about this so I had a bang at it. I don't have Curviloft installed on this computer so this was done with Tigs extrude tools and Bezier as the only plug ins.

Pretty much the way it was done B4 Curviloft!!

And how long did it take you to do this?? (nice work though )

CL can reduce the time spent on creating complex surfaces by a factor of 10 or (way)more.

Plus, it is so easy to scrap a bad surface, tweak the perimeter and hit go to fix something.

Way back B4 CL, I spent hours getting a fairing (almost) right. A design tweak forced a redraw of that fairing, that took almost as long to redo as original. A second design tweak also forced a scrap of that fairing and 2 others, BUT with CL, all 3 were fixed in less than 10 minutes, and were smoother fairings as well.

Jeff

You twigged a realization. It may not be just big surfaces, but as you say, having more than 4 sides. Airplane fairings and shapes usually have more than 4 defining edge contours. And they tend to be big, so, I assumed it was the complex size, rather than the 4+ sides. In fact when I think about the last "big one" I did, I had to reduce it to 2 big quad and 1 tri-sided perimeters, so what you said holds true, and is something I need to consider next "big" shape.

As for patching, I still use CL to fix formed ridges and excess lines that CL may produce. I judiciously delete the offending sections from the overall surface, forming quad sided perimeters and CL the hole. It almost always produces a better shape, just as your example shows.

I also spend quite some effort in first reducing the perimeter segment count, getting rid of tiny or divided straight lines with a single line. That too, makes for a cleaner surface.

I do this a lot, because the vast majority of fairing perimeters are formed by intersecting one or more curved defining surface(s) through the aircraft skin. Intersect produces a lot of tiny fragments and gaps when the intersection lies very close to a vertex or a nearly parallel line.

@unknownuser said:

250 installed witout any problem

And STILL on SU-V6..... Go figure

You can zoom INTO the object and see the problems defined in red or orange. It really does not matter which you tackle first. Put your cursor on one item and deselect the group/comp. Then select again and edit. The cursor is on the problem. 99% all you have to do is delete it.

You can also change to XRAY view (Wireframe will not show problem faces) In XRAY you can move the cursor right onto the problem from outside the object, then edit it.

Sometimes you may have a circle that seems to be well outside the object. SI found a line fragment within the groups bounding box but totally outside the object. Pan and zoom such that the circle is seen against BLANK background, ie open sky or ground. Place the cursor in the circle, then deselect and edit again. Make a sweeping right to left selection to catch the offender and delete.

Note: You cannot edit the object while SI is active. You cannot activate SI while in edit mode.

There is also a companion pluggin, Solid Solver that "fixes" these kinds of errors, BUT, it can also destroy (undoable, almost always) an object that has a lot of problems.

I once had an object that started with about 600 lines and ended up with about 30 when Solver finally got done, so it is a good idea to SAVE before trying to Solve a complex mess.

I also have had objects that were deemed by Solver as "Can NEVER be made solid" fixed with just a few missing lines added.

@unknownuser said:

I provided some elements of cleansing in Curvizard::Clean up. Part of it is integrated into Curviloft as well (you may get a message).

Fredo

I really need to spend some time to learn some new tricks. Curvizard is in my update SU list, like a bunch of other stuff.

You have 2 choices.

1. CURVILOFT the hole using the skinning tool. Select the perimeter contour of the hole and click the green checkmark. Gives you a group of the best curved surface it can find.

Besides, if you are drawing airplanes, Curviloft is indispensable, especially for fairings, wingtips, canopies, nacelles, fuselages, wings and empanage. In fact; the whole airplane.

2. Using the wheel well door skin group, place it in its proper closed position. Edit the door by deleting all the well perimeter lines, but NOT the lines defining the shape of the curve. Best to make all lines hard for this, or you may lose some curve points.

Then edit the wing skin and remove all well perimeter lines as you did the door. Again, make any curve defining lines hard first.

Explode the door skin group. Join vertices to form the missing surface triangles. It is far easier to delete "broken" lines and replacing them, rather than trying to join a very tiny gap. That part of the surface should be identical to the original. Then make all the lines soft and smooth.

Shift+CTRL+Erase to make them hard lines again.

SU's Intersect is problematic. It almost works.

It does however skip lines, will not intersect a very short face, causes surface flatness issues and tosses in unwanted hidden lines. There seems nothing you can do to prevent any of that.

Keep in mind, Intersect does NOT split the object, as you want to do. It only draws a common dividing line where the 2 objects intersect.

That being said, there are a few things you can do to correct or ease the correction.

First, It is best to draw your helicopter in a half-hull, as a component, then copy and flip its side. Move the copy on axis to join the 2 halves. Make the "dividing" line hidden or at least soft so it won't show later. That will eliminate the intersection operation. Any gaps along the dividing line are easy to spot and fix.

But, say you need to split something along a plane or a shape. (no pun intended) First make that dividing plane/surface a group, so it does not interfere with the objects geometry. Move it into position.

Edit the object and INTERSECT WITH MODEL. It is faster and better to pre-select only the faces that will actually be intersected by the plane. This prevents other intersecting objects to be intersected as well, adding to the cleanup job.

Now you can delete the intersecting face group, but best to wait until finishing fixing stuff.

There are a few things you can do to split the object, either delete the 1/2 you don't want and/or group/comp the part you do want. You can also group/comp the other 1/2.

Now comes the tedious part. Edit the cleaved portion. Hide the rest of model.

Closely examine along the perimeter of the split, looking for line fragments extending or face gaps along the line.

For extended lines, where it has no faces on either side, unhide the model and draw a line from the next vertex to where the line should end, then to the opposite vertex, using the plane group as a reference. 2 faces should form. Delete the line fragment.

Almost always, Intersect will miss only a single segment between 2 points on the split perimeter. Just join them endpoint to endpoint and the missing face should form.

If not, do a very small select lasso on each point to ensure there is not a tiny invisible line fragment, preventing the face. Delete those fragments, and redraw the line. This may also cause other adjacent faces to disappear, and you need to fix them too. Like I said, tedious work.

In many cases, that line may be very tiny, so it may be better to move one vertex (endpoint move) onto the other, eliminating the need for the line, but delete any fragments first.

Surface "unflattening" and extra hidden lines are related. When Intersect does its job, it sometimes cannot put an intersect line in place where an object line is very close to it. Upon cleanup, SU interprets the original objects line as an intersect line, and that causes a surface to unflatten, with hidden lines added to take care of the folding. We are talking less than .001 inch here.

If you need to zoom in very close and get near field clipping, then zoom out and turn off Perspective view. Now you can zoom in as close as you want.

The best way to fix this is to move the endpoints of the line onto the surface of the plane, forcing the face to reform and the hidden lines to hopefully go away. You need to use the MOVE+ALT to move the endpoints, and an axis (red, green or blue) direction. If the plane is not axis oriented, you may have to move the endpoint 3 times (each axis) to get it properly placed on FACE of the plane face.

There are also Pluggins that may help move endpoints onto a flat intersecting face.

I do this sort of thing a fair bit, and without Curviloft, I'd still be trying to finish the first one.

But here are a few tips on what I've learned so far.

When CL creates a bad surface, most usually it is because A) too many vertices, B) Doing too big a surface in 1 go and C) Not enough curve guidance.

A) Cut back on the vertices. Instead of the usual 24 segment circles, use 12. And by all means, try to keep the segment count on opposite sides of a contour the same, and evenly spaced. Not usually possible. When there are big differences, CL will create a very high poly count mesh.
It tries to connect opposite points in parallel curves to the sides of the contours. That will create a myriad of extra lines and new vertices on the contour, and this is very difficult to clean up if you want to reduce the line count.

B) Do big surfaces in 2 or more sections, using common contour lines. Do not explode the CL groups till you are done and satisfied. This makes it easier to redo or modify a section, and you can be sure adjacent sections are formed from the same contour lines.
After CL does its thing and you explode it, the contour would almost always have many more segments than before, making the next surface that much more complex. Also, making several sections prevents any complex surface from influencing the next, like at the intersection of the windshield to the body sides and nose of the Huey.

C) If your contour makes significantly big changes, and B) didn't help much, then you need to add additional contour control lines between the original contours, and do them as separate sections.
I will first make the surface, that does not form the way I want. Then, I will draw a new contour line midway along a curve that is what I want, using the bad curve as a guide. Then I delete the bad CL surface and CL the section as 2 sections. This will almost always give you the desired end result. When happy, explode everything.

I cannot live without Curviloft, in spite of it's (very) few warts.