Control Point curve has no start tangent

Hi
V5.
Control Point curve is the default one when clicking the curve toolicon. It creates a curve as best fit through the points it then creates as you click your way along where the curve is to go.

I have a line and wish my curve to start tangent to the line at a position I click, then to best fit the points the tool makes thereafter. There is no start tangent or end tangent option to do this though.

It is mathematically possible to start tangent then best fit.

I then try Ctrl hover Osnaps and pick One shot tangent fromā€¦feeling pleased with myself I remembered these.

  1. select curve for tangent trackingā€¦I select my line
  2. pick a pointā€¦I click on the line where i want the curve to start
  3. Start of Curveā€¦eh ?..the line has gone white and my cursor runs up and down it, but I just picked the start in step 2ā€¦oh well I will click the line again just fwd of the last clickā€¦I get a line going off at any angle, not tangent.

Steve

I would make use of the ā€˜TAB Lockā€™ for this.

  1. Select END and NEAR OSNAPS
  2. Click the END of the line you wish your curve to be tangent from.
  3. back track along the Line with NEAR activated.
  4. Tap the TAB key
  5. move off in the direction you wish the first point to be and click

Noā€¦ It creates a NURBS curve of the specified degree that has control points where you click. Nothing to do with ā€œBest Fitā€ at allā€¦

ā€“Mitch

1 Like

Use TangentFrom for the second, not first, control point. The second control point determines the direction of the start of the curve.

Try this:

Make sure you can select the starting point of the curve twice. Place a temporary point if needed.

Curve

Select the starting point for the curve.

Start the TangentFrom one-shot object snap.

Pick the line or curve you want the new curve to start tangent to.

Pick the location you want the new curve to be tangent from. (This is why a temporary point may be needed.)

Pick the location of the second control point. It will be constrained along a line tangent to start point.

Continue to pick control points.

1 Like

Hi,
I see this thread I created and hope it provides an answer to my need for best fit arching curve start and end tangentā€¦as show by the green curve, the result of using InterpCrv then curvature graph !

see attached.
I wish for the ā€˜intelligentā€™ (it seems to know what I want !!) tangent curve that InterpCrv gives with StartTangent (ditto end tangent) combined with an ability to then best fit a curve of arching form (not arcing but arching) through points and ending tangent.

I was to say Deg2 but deg2 can have a kink.

Considering the controlPoints curve, and this thread, just to see what it gives, maybe something that can be rebuilt afterwards, I find the tangent part is not suitable.

using the TangentFrom method whilst making the curve tangent, sees the curve kick out (red), using InterpCurve sees the curve follow a better path (blue), it seems more ā€˜intelligentā€™ when it sees where the rest of the curve has to go and creates the curvature I am after, however that tool is too precise for the points, I want the type of curve that control point makes, a sort of best fit, though as pointed out by Mitch, its placing CPs at the points.

In fact I want an egg shaped curve, so no meandering at all, should I use Control point curve then rebuild with less points maybe, or InterpCrv and curvatureGraph or is there a better tool to create an arching curve best fitted through points ?

I also whilst managing the TangentFrom method of davidcockey could not get it to end the curve tangent, what is the method there ?
Though the results are not as good as InterpCrve, it would be worth knowing.
contol point curve start and end tangent.3dm (38.0 KB)
Steve.

When you have points that are as bad (uneven) as these, you should not be using InterpCrv at all. You should be using BlendCrv, and simply visually adjusting the settings until the curve passes close enough to the points for your satisfaction. If you had maybe two points to run through, I would say use InterpCrv. A much better workflow is to start with BlendCrv. If that gives you an acceptable curve for your data set, great. If that gets you close, but maybe not exactly what you want, then you can either ChangeDegree and point edit, or Rebuilt with more control points and point edit.

I spent the extra 30 seconds to finish it off. Workflow was:

  1. Start with using BlendCrv to get close.
  2. Change degree to 5.
  3. Turn on curve points.
  4. Point Edit using Gumball to achieve final fit

.contol point curve start and end tangent_skyg.3dm (62.7 KB)

You can always check the deviation between a curve and your points using PointDeviation. The final curve is within 0.5 mm of all the points, is smooth on curvature graph and tangent to the top and bottom curves.

1 Like

Hi skyg
Thanks. This is very important and useful now to me as I find myself doing this a lot.
What curve did you draw and how so as to choose it for BlendCrv as starting off without any curves sees the command ask for a curve to be selected.
Were the tangent start and end created using BlendCrv or created prior as part of the process of creating the curve for it to be fed ?

Steve

hI:

for more info, look at the help file for BlendCrv, thereā€™s an animation.

Level I, page 99
Level II, Chapter 4.

-Sky

skyg, why rebuild as degree 5 with presumably a single span and 6 control points including the ends rather than rebuild as a degree 3 with 6 control points and 3 spans? Both are curvature continuous.

Hi David- Degree 3 curves that have more than 4 points will have internal knots - the curvature graph will hiccup there in most cases even if the curve is G2 as you say. A single span 5/6 curve is cleaner and has a nicer flow of curvature.

-Pascal

What Pascal said with regards to curvature combs. Also, personal preference to simply use single span surfaces, as I find they are much easier to point edit and sculpt. I also find that I can generally get degree 5 surfaces to match better than degree 3 where you have multiple surfaces meeting (say a corner with 4 surfaces). Also - I didnā€™t Rebuild. I used ChangeDegree. I personally like using it much more, as the shape of the curve is identical to the input (tangency is maintained, not that you canā€™t fix it up with Match after the fact, but thatā€™s just one more thing to do). BlendCrv will make single span curves by default - so theyā€™re degree 3/4 points if you simply use tangent, and degree 5/6 points if you use curvature. So if you change degree up to 5 from 3, you have the exact same curve but more points to allow point editing. Make sense?

Hi skyg,
Do you use ChangeDegree and make deg3 into deg5 for your aerofoils ? (the article has vanished off www)

The ability of -_Rebuild to match an existing CP structure is good.

The BlendCrv method here is proving a godsend on the subject matter I posted about, :grinning:

Steve