population on object problem

Started by otakar, May 02, 2016, 01:46:59 PM

Previous topic - Next topic

Kadri


It is strange indeed.
Your file does have the same problem too actually Ulco.
It is just that to see it more easily you have to use 45 degree rotation.
Anything less closer to zero and more closer to 90 makes it harder to see and using different scales for the population hides it too.
But when you hit the sweet(bad) spot it looks like Otakar's image for example.



Kadri


I tried changing the population angle too but that didn't worked out.
Not in my test at least.

I just tested a basic sphere OBJ file now.
I thought maybe the objects surface features could be attributing to this problem.
But no...Even a sphere population gets rowed up(?) when you rotate it in the X or Z axis.
Only the Y axis is without any problem like in the tests before.

bobbystahr

Quote from: Kadri on May 06, 2016, 01:34:53 PM

I tried changing the population angle too but that didn't worked out.
Not in my test at least.

I just tested a basic sphere OBJ file now.
I thought maybe the objects surface features could be attributing to this problem.
But no...Even a sphere population gets rowed up(?) when you rotate it in the X or Z axis.
Only the Y axis is without any problem like in the tests before.

after numerous tests  myself find this to be true
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

otakar

Thanks for the confirmation, everyone. Oshyan, the coordinates make no difference, both the object and the population origin have the same coords. Population size is sufficient to cover the whole object regardless of rotation. The instances DO get generated, it's their placement that seems to be issue (see my first render) - all bunched together in a few spots. And yes, I did try to match the population plane rotation to the object's rotation without success (this does not make sense anyway). So from my perspective this seems to be a bug.

bobbystahr

just found this figure lying around and then saw the spam from eon and tickety boo I was playing again...
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

Matt

There's definitely a bug here. From what I can see so far, it seems to be worse at small scales, perhaps relating to the size of the spacing between instances.

Matt
Just because milk is white doesn't mean that clouds are made of milk.

KlausK

Hi everyone,

curious stuff, here are some test renderings.
Distribution of the population object is set to 0.0 on all random values to see more clearly where on the populated object it gets placed.
What position information does the populator read initially from the populated object? Polygon center? Point on a Polygon?

A few things I found:

- The populator does not seem to update the position values of the object it has to populate correctly - if at all.
It actually looks like it only repopulates objects on the positions which are identical after the populated objects rotation values are manipulated.

- Only changing the rotation values on the populated object influence the populator. That`s what some people already found out.
Maybe it has something to do with local and global values not calculated correctly.

- A pre-rotated object imported from LW for example gets populated correctly although not completely. (see attached picture)
As soon as you start to rotate the object in Terragen the same populaton error occurs.

- Rotating the object in Terragen in steps of 90 degrees works on all axes.
It will be populated more or less correctly.
But again not the whole object gets populated when you rotate by 90 and 270 degrees.
That is in a right angle from the original rotation values. It somehow seems to only see the "upper" half of the body.
Are the values of the objects "lower" half negative values Terragen does not compute?

I tried to figure out a way to read and write the vector / scalar / normal values of the rotated object and feed them to the populator node.
But my skills are veeery limited. And of course I am only guessing what might be the problem here...
Is that at all possible?
Maybe some blue node guru in the forum could give it a try?!

You may all have a good laughand make  fun of me if I completely headed in the wrong direction with this ;)
cheers, Klaus

ps1: the first picture shows an object where I only exported the centerpoints of the polys of the object I created in LW.
It is then populated with a card and a rock (Terragen objects)
ps2: posts will get shorter sooner or later, hopefully
/ ASUS WS Mainboard / Dual XEON E5-2640v3 / 64GB RAM / NVIDIA GeForce GTX 1070 TI / Win7 Ultimate . . . still (||-:-||)

Dune

Interesting experiment. I guess Matt will find the culprit from these.

Regarding Are the values of the objects "lower" half negative values Terragen does not compute?: I would say so; if you want to populate on the 'underside' of an object, you need to add another population and reverse its location (X=180º)

bobbystahr

Quote from: Dune on May 10, 2016, 09:50:41 AM
Interesting experiment. I guess Matt will find the culprit from these.

Regarding Are the values of the objects "lower" half negative values Terragen does not compute?: I would say so; if you want to populate on the 'underside' of an object, you need to add another population and reverse its location (X=180º)

I think I recall Matt mentioning that TG *drops* objects onto the target object so no there's other way to do the underside.
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist

Kadri

#39
Quote from: KlausK on May 10, 2016, 08:10:45 AM
...
- A pre-rotated object imported from LW for example gets populated correctly although not completely. (see attached picture)
As soon as you start to rotate the object in Terragen the same populaton error occurs.
...

Can you check that ones again please Klaus?
I had tried that earlier and just got curious and tried it again.
I have no problem to populate a pre rotated imported object to populate in Terragen.
Actually it doesn't make any sense that it should have an impact.
Terragen or any other program can't see any difference (if it was rotated or not in another program) in such an object.
But i was just curious if the surface features of the objects do have any impact on populations and tried it anyway.
It populated just as any object without problem until i rotated it in Terragen of course.


Kadri


Klaus just a guess but if you have the "Object spacing in a,b"  value at "1",
you can use much much smaller values actually by entering it manually in those fields.

KlausK

@Dune & bobbystahr - that makes sense of course. Thanks for the hint.

@Kadri - I think I meant the same thing.
In picture "KlausK_PopulateObject_Problems_1.jpg" the example in the middle is the object I rotated in LW, saved and exported it as lwo.
No surface or UVs on it. Terragen imports it as object with all transform and rotation values at zero.
So you are right. It does not care how it was transformed in the 3D app before.
Due to the fact that the exported objected is pre-rotated before exporting it however, it does not get populated completely all over, mainly
at the "underside" so to speak. I tried it with other objects as well and got the same result.
The other thing...
I know about the object spacing values. I wanted to populate the object with one rock per point or polygon
or whatever point in object space Terragen uses to place the rock on the bellshape. To see more clearly
where the populaters are put. That`s why I took out as much as possible randomness in the placement.

Or am I not getting what you mean?

cheers, Klaus
/ ASUS WS Mainboard / Dual XEON E5-2640v3 / 64GB RAM / NVIDIA GeForce GTX 1070 TI / Win7 Ultimate . . . still (||-:-||)

Matt

Quote from: bobbystahr on May 10, 2016, 09:53:56 AM
I think I recall Matt mentioning that TG *drops* objects onto the target object so no there's other way to do the underside.

You can populate on the underside of an object. You do this by rotating the entire population so that it projects in a different direction, instead of downwards. Oshyan gave a demo of it in this video around 54m52s:

https://www.youtube.com/watch?v=n7Rp4Go8jLs&feature=youtu.be&t=54m52s

Matt
Just because milk is white doesn't mean that clouds are made of milk.

Kadri

#43

Klaus i think i misunderstood what you said.
If the problem in that image was only about the underside of the object you have to use another population as Matt said for this.

I am happy that Otakar made this thread.
I had problems in the VR challenge with populations. Because of time constrains couldn't look further into this.
Funny that this problem comes up only now.


bobbystahr

Quote from: Matt on May 10, 2016, 06:53:45 PM
Quote from: bobbystahr on May 10, 2016, 09:53:56 AM
I think I recall Matt mentioning that TG *drops* objects onto the target object so no there's other way to do the underside.

You can populate on the underside of an object. You do this by rotating the entire population so that it projects in a different direction, instead of downwards. Oshyan gave a demo of it in this video around 54m52s:

https://www.youtube.com/watch?v=n7Rp4Go8jLs&feature=youtu.be&t=54m52s

Matt


Yes I recall that being part of your original long time ago post on that...
something borrowed,
something Blue.
Ring out the Old.
Bring in the New
Bobby Stahr, Paracosmologist