Is there a way to get the UV position of a point on a curve?

[magneto]

Hi,

This is the opposite of what I asked last time. Basically I have 2 curves that are totally different by their length and number of points.

What I am trying to do is to get the normalized UV coordinate of a point on Curve A, say point 3 is 0.271, and then get the corresponding point position in Curve B. In this case I would use primuv and feed it the UV coordinate I have which is 0.271.

But I don't know how to get these normalized UV values for each point.

I looked at the UV Texture SOP and used the Arclen Spline Length and other spline options and then normalized these values to 0-1, but the coordinates do not match, when I used Curve A for primuv.

Can I calculate these UV values by measuring the length up to each point, dividing the result by the total curve length? I am not sure how to measure up to a point though since the arclen function requires UV coordinates.

Anyhow is there a way to do this?

Thanks

[zarti]

to me it looks like you do not have good relationship with measure SOP ..

just call it .

=)

[magneto]

Thanks zarti. I know about the Measure SOP but how can I use it to get the UV values of points? It would only give me the total length like arclen no?

[zarti]

But I don't know how to get these normalized UV values for each point.

{ WoW ! nonCommercial version doesnt let me Paste code .. Haa ! another handjob for me }

fit( $MAPU, 0, point( `opinputpath( '.',0 )`, $NPT-1, "uv", 0 ), 0, 1 )

--

here is a file which seems to work

- with or without UVs ( Measure SOP assigns distance as attrib to each point ) ..

- NURB / polys ..

--

getProportionalLocation.hipnc

--

.cheers

[magneto]

Thanks zarti. You can copy using the NC version. Just open the external editor (Alt+E)

So I looked at your file and found that you used the expression you posted in attribcreate1, which is connected to curve1. That was exactly what I was doing, but in this nodes until attribcreate1 there is no Measure SOP. I noticed you use it in other parts of the file.

But the values calculated using my_distance doesn't give the correct values

Try this expression:

primuv("../attribcreate1", 0, "P", 0, point("../attribcreate1", 6, "my_distance", 0), 0)

and you will see that returned value will not be the same as Point 6's X value.

As for the Measure SOP, I noticed you were using Blast SOPs. Were you trying to delete everything but 1 segment and then measure its length? Because if you measure after deleting, the length of the segment changes.

Thanks again

Edited September 9, 2012 by magneto

[jkunz07]

Here's some methods I've used before

Assign_U_Coordinates_to_a_Curve_ToPublic.hip

[zarti]

You can copy using the NC version. Just open the external editor (Alt+E)

oh thanks , ill try to remember that .

--

Measure SOP , which i mentioned in the first reply , was not part of the NURB's case . it was in the branch of polygon case and have used it before to measure segment lengths ( while in this one to measure ' how far ' is each point from point 0 ) . so , i could find the ratio to feed into Carve SOP or primuv expression ... but ,

But the values calculated using my_distance doesn't give the correct values

Try this expression:

primuv("../attribcreate1", 0, "P", 0, point("../attribcreate1", 6, "my_distance", 0), 0)

and you will see that returned value will not be the same as Point 6's X value.

.. but it should give exact values if segments are equal ( points distributed uniformly ) .

so Carve and primuv exp. obey to something else but not 'physical lengths' .

sad . and dont know Why that ( ??! )

edit: i tried to use a curve with all segments equal to each other and my method works .

--

As for the Measure SOP, I noticed you were using Blast SOPs. Were you trying to delete everything but 1 segment and then measure its length? Because if you measure after deleting, the length of the segment changes.

that Blast-ard escaped the cleanup i made of some nodes i was testing inside foreach . it is useless actually .

--

so maybe a solution wd be to evenly resample curves ( in a parallel branch maybe ) to get the most accurate result possible without affecting original ones .

Edited September 9, 2012 by zarti

[magneto]

Thanks a lot guys.

@John: I looked at your file, cool techniques but when I tried to feed the Red color values to primuv expression's u argument, it's still not exact to actual point positions. Were you able to see identical results on your tests?

@zarti: I see what you mean. Uniform works for sure. But if the points are uniform, $PT/($NPT-1) would work too because the u coordinates would also be uniform.

Are you sure Carve and primuv doesn't use the physical length? Since I don't know the exact U values of points, I can't verify 100%, but my tests show the U values you use for Carve and primuv seem to be precise by length.

I think that's why finding the U coordinate is tricky

[zarti]

her is the proof Carve ignores physical length of curves ( primuv() should be the same ):

- the segments are equal ( build with snapping to grid ON )

- set to display the nodes in black .

the Carve is animated with linear() interpolation , while as you can see the speed is affected by points density .

--

getProportionalLocation_proofOfNonLinearity.hipnc

--

edit : if you create an UVTexture with texture type = Average Spline , it works here .

put this in attributecreate2 node :

$MAPU / point( opinputpath('.',0), $NPT-1, "uv", 0 )

but im guessing you already have UVs .. or not ??

edit2 : attachment added ..

--

getProportionalLocation_workingWithAverageSpline.hipnc

--

Edited September 9, 2012 by zarti

[magneto]

Thanks zarti. Your idea works when the curve order is 2, when I increase it to 4, the values shift again

I am not sure if Carve and primuv is supposed to be linear. Perhaps someone can shed a light? I have another idea to try so if it works, I will post it here. If it doesn't work, I will post it here

[jkunz07]

@John: I looked at your file, cool techniques but when I tried to feed the Red color values to primuv expression's u argument, it's still not exact to actual point positions. Were you able to see identical results on your tests?

Attached file includes how I was planning to fetch point position based on normalized U value. I don't have any experience with the primuv expression, but in the file providing a U value between 0-1 returns an exact point position on the input curve. As far as the U coordinates being assigned to the curve, those are still an approximation.

Assign_U_Coordinates_to_a_Curve_ToPublic_v002.hip

[magneto]

Thanks John, very useful. I looked at the file again. How do you compare your generated U values to return the matching XYZ coordinates? Like which expression are you using?

[jkunz07]

Thanks John, very useful. I looked at the file again. How do you compare your generated U values to return the matching XYZ coordinates? Like which expression are you using?

In the v002 file I just uploaded there's some added sops colored in like a mustard yellow. Using a point sop I'm shuffling P into Cd and $CR into $TX. $TY and $TZ are zeroed out. This creates a normalized line in the positive X direction.

With the add sop I'm creating 1 point with a slider range 0-1 this slider sets $TX for the newly created point. Using a vopsop and pcfilter the nearest point's color is retrieved (which is actually the position).

Kinda confusing, probably could be simplified, but it's what I came up with.

[petz]

i didn´t go throught the whole thread too carefully but can´t you just use "unituv" or $PT / ($NPT - 1) for that purpose?

petz

curve1.hipnc

Edited September 10, 2012 by petz

[magneto]

Thanks guys.

@John: I will have to inspect your file more closely. I couldn't find time yesterday

@petz: $PT / ($NPT - 1) only works for uniform points on a curve. If they are all over the values do not work.

I looked at unituv from the docs but not sure how it works? I checked your file but the U values seem to be the same as $PT / ($NPT - 1).

[stabby]

Hi, If you're working with polys, then you should convert your curves to nurbs, order 2.

Use a basis SOP and change U parameterization to 'Approximate Arc Length'.

The carve SOP and primuv etc. will use the new parameterization.

EDIT: should probably add that changing the parameterization of nurbs with order > 2 will change the shape of your curve a little.

Thanks guys.

@John: I will have to inspect your file more closely. I couldn't find time yesterday

@petz: $PT / ($NPT - 1) only works for uniform points on a curve. If they are all over the values do not work.

I looked at unituv from the docs but not sure how it works? I checked your file but the U values seem to be the same as $PT / ($NPT - 1).

Edited September 13, 2012 by stabby