Vertex groups for hair (length etc.): effect not always satisfactory #54069

New Issue

Tom Telos · 2018-02-14T09:14:18+01:00

Tom Telos commented

2018-02-14 09:14:18 +01:00

System Information
iMac 27-in late 2015; 16GB; AMD Radeon R9 M395X 4GB

Blender Version
Broken: recent builds of 2.80 and 2.79 (feb 2018); 2.78c; 2.75a. Worked: none yet

When you open the attached .blend file,
VgLen.blend you will see two objects with particle-based hair.

The lower one has a vertex group for length, that covers the entire object,
whose weights go linearly from 1 to almost 0, and its effect on hair length
is not quite what one would expect, but close:
weights of 0.9 and 0.8 appear to have the same effect as 1.0,
and the shortest hairs are about 1/4 the length of the longest, as opposed to almost 0,
but there is, clearly, a linear dropoff from vertex weight 0.8 on downwards.
This I observe with interpolated child particles, no kink,
or with kink = Curl, Radial, Wave, or Braid.
Note: Kink = Spiral does produce a length dropoff that looks impeccable!

The upper object is a simplified scalp from my WIP here: https://sta.sh/21968qcn7oqe

It also has vertex groups for density, length, and kink, but here the discrepancy
between the expected hair lengths and what one sees is much greater.
In this case, even kink = Spiral fails to satisfy length expectations; moreover,
less than half the hairs receive a spiral, even when I’m not using any vertex group for kink.

**System Information** iMac 27-in late 2015; 16GB; AMD Radeon R9 M395X 4GB **Blender Version** Broken: recent builds of 2.80 and 2.79 (feb 2018); 2.78c; 2.75a. Worked: none yet When you open the attached .blend file, [VgLen.blend](https://archive.blender.org/developer/F2270385/VgLen.blend) you will see two objects with particle-based hair. The lower one has a vertex group for length, that covers the entire object, whose weights go linearly from 1 to almost 0, and its effect on hair length is not quite what one would expect, but close: weights of 0.9 and 0.8 appear to have the same effect as 1.0, and the shortest hairs are about 1/4 the length of the longest, as opposed to almost 0, but there is, clearly, a linear dropoff from vertex weight 0.8 on downwards. This I observe with *interpolated child particles, no kink,* or with kink = Curl, Radial, Wave, or Braid. **Note**: Kink = Spiral does produce a length dropoff that looks impeccable! The upper object is a simplified scalp from my WIP here: https://sta.sh/21968qcn7oqe It also has vertex groups for density, length, and kink, but here the discrepancy between the expected hair lengths and what one sees is much greater. In this case, even kink = Spiral fails to satisfy length expectations; moreover, less than half the hairs receive a spiral, even when I’m not using any vertex group for kink.

Tom Telos commented

2018-02-14 09:14:18 +01:00

Added subscriber: @ttelos

Mai Lavelle commented

2018-02-15 09:46:10 +01:00

Added subscriber: @MaiLavelle

Mai Lavelle commented

2018-02-15 09:46:10 +01:00

I'm not entirely sure what the cause is, but it seems to related to the number of hair segments interacting strangely with interpolation. There are plans to replace particles with a new system, so not sure if this will end up getting fixed in the current system or not. In the mean time increasing the 'steps' parameter should give you a better result.

Tom Telos commented

2018-02-15 22:11:11 +01:00

Thank you, Mai.

The “Steps” parameter is specifically for kink=Spiral, and appears to affect only the presentation of the spiral,
but the “Segments” parameter that shows up near the top of the Particles Properties does influence hairs
in remarkable ways. My .blend file has it set at its default value, 5. If you select the scalp object
(named HeadLenTest) and bring Segments down to 4, you will see that, in response,
many hairs now extend their endpoints into the scalp, even though the object’s normals all point outward.
This effect is most obvious with kink=Spiral: you will then notice that all the hairs that extend inwards
have their spiral on the “in” side, which may provide a partial explanation to why, when Segments=5,
many hairs appear to have no spiral.

When Segments=2, no hairs project into the scalp, and the range of lengths appear to be ok.

Raising the Segments value from 5 to 10, or 20, appears to have no effect other than to shorten the hairs.

Thank you, Mai. The “Steps” parameter is specifically for kink=Spiral, and appears to affect only the presentation of the spiral, but the “Segments” parameter that shows up near the top of the Particles Properties does influence hairs in remarkable ways. My .blend file has it set at its default value, 5. If you select the scalp object (named HeadLenTest) and bring Segments down to 4, you will see that, in response, many hairs now extend their endpoints *into the scalp,* even though the object’s normals all point outward. This effect is most obvious with kink=Spiral: you will then notice that all the hairs that extend inwards have their spiral on the “in” side, which may provide a partial explanation to why, when Segments=5, many hairs appear to have no spiral. When Segments=2, no hairs project into the scalp, and the range of lengths appear to be ok. Raising the Segments value from 5 to 10, or 20, appears to have no effect other than to shorten the hairs.

ronan ducluzeau commented

2018-02-17 14:05:07 +01:00

Added subscriber: @zeauro

ronan ducluzeau commented

2018-02-17 14:05:07 +01:00

The “Steps” parameter is specifically for kink=Spiral, and appears to affect only the presentation of the spiral,

Mai is not talking about this setting present in Children panel.

In Display panel and Render Panel, there is also a "Steps" parameter.
This steps are set to a value of 2 or 3. If you increase them to 10, you obtain a better render result.

Steps are controlling visual segments of hair path.
Step to 0, you obtain a straight line. Steps to 1, you see a polyline of 2 segments. Steps to 2, a polyline of 4 segments, Steps to 3, a polyline of 8 segments. etc.... It is a power of 2.

You can define 20 physical segments that can be edited in Particle Edit mode and would be used by hair dynamics simulation.
If number of steps stays equal to 2, path stays visually a polyline of 4 segments.
And any particle emitted from face is not emitted from a vert. It has a weight higher than 0. So, first step is shown for most of particles.

Increasing number of steps on a path of same length means reducing length of this first step.

Increasing default value for "Steps" parameter, everywhere, to 4 would probably be more pleasant for users and lead to less questions.

> The “Steps” parameter is specifically for kink=Spiral, and appears to affect only the presentation of the spiral, Mai is not talking about this setting present in Children panel. In Display panel and Render Panel, there is also a "Steps" parameter. This steps are set to a value of 2 or 3. If you increase them to 10, you obtain a better render result. Steps are controlling visual segments of hair path. Step to 0, you obtain a straight line. Steps to 1, you see a polyline of 2 segments. Steps to 2, a polyline of 4 segments, Steps to 3, a polyline of 8 segments. etc.... It is a power of 2. You can define 20 physical segments that can be edited in Particle Edit mode and would be used by hair dynamics simulation. If number of steps stays equal to 2, path stays visually a polyline of 4 segments. And any particle emitted from face is not emitted from a vert. It has a weight higher than 0. So, first step is shown for most of particles. Increasing number of steps on a path of same length means reducing length of this first step. Increasing default value for "Steps" parameter, everywhere, to 4 would probably be more pleasant for users and lead to less questions.

Sergey Sharybin commented

2018-02-19 17:54:11 +01:00

Added subscriber: @Sergey

Sergey Sharybin self-assigned this 2018-02-19 17:54:11 +01:00

Sergey Sharybin commented

2018-02-19 17:54:11 +01:00

Caused by 4f1c0a1. Will give a closer look on what's wrong in that commit later.

Caused by 4f1c0a1. Will give a closer look on what's wrong in that commit later.

blender-admin commented

2018-02-20 10:47:01 +01:00

This issue was referenced by 87e37224c7

This issue was referenced by 87e37224c74a4336278e27b0de9e2223701439ae