I was doing render tests with the TWO rabbit and trying to figure out why he's so slow.

Is there a reason his hair materials have the hair splines set to 6 CPs? That's a lot for such short hair, isn't it?

I think I made him but someone could have altered him since. But yes, 6 cps does sound alot. Maybe his head requires that many? Can't remember.

This is interesting...

Hair set to 6 CPs (the way the hair materials are saved in the SVN) - render time 1304 seconds

Click to view attachment


3 CPs - render time 676 seconds

Click to view attachment


1 CP - render time 363 seconds

Click to view attachment


There are slight differences in appearance, but it's not apparent that one is better than another.

The next mystery is why using a z-buffered light is only slightly faster than a ray traced light. In other tests I've done with hair z-buffer was much quicker.

I redid the fur on the rabbit. I don't know why I used 6 cps, but the more you use, the better the transitions between settings along the length will be.

You mean...color transitions... or width...or something different...? Just wondering what you mean there... HEY! That's a good looking rabbit there!

Mostly color transitions.

Perhaps it depends on number of rays, distributed in passes & how many passes? when the hair material is also set to cast shadows.

Transparency and length of hair might also have contributed to larger discrepancies in other tests. I believe I have also found thicker,shorter, hair is faster than thinner, longer, or using intricate images for emitters (complicated alpha channels with varying transparency). Perhaps your other tests used more complicated hair?

Here is example where more complicated emitters take longer - my guess is that the discrepancies in render times with different methods (ray trace versus z buff) would be greater for those that take longer with z buffered

As far as I can recall, I've never done hair with more than 2 cps, as I found it hard to groom. So interesting to know there is a render hit with more cps.

here's a comparison of 2 vs. 6

realtime mode looks smoother with 6 but final render looks the same to me

Click to view attachment

How could you do those curvy hair styles with only 2 cps?

Here's a v13 render, the version I redid the rabbit fur in. There is definitely a difference. It renders the same in v15, multipass ON or OFF.

Actually it's really 3 - counting the cp at the base plus 2 along the length. - That's more like what I meant to say. I believe I've always used the default number that comes up - whatever that is (I think it's 2). Always wondered why/how others did use more.

I've discovered there are 'trix' to getting grooming mode to work with a lot of CPs... so you get the full 'tactile-response'... things like (temporarily) set the pre-roll to zero... turn OFF dynamics, and hide any other hair/geometry that is impertinent.

I did'nt know you could set that setting to 1...how could hair have only 1 CP? It has a beginning and an end...thats 2, no? I'm always finding crazy stuff about variuos settings in A:M...like for years you think something only goes from 1 to 100% and then you learn it can go beyond 100 or even accept a negative value...

Good tests, Robert!

A few tips regarding grooming:

1) Always turn off Dynamic Options in the Hair System while grooming. You can turn it on for testing but try to get into the habit of grooming with dynamics turned off.
2) Start simple. Turn Density down to 1-10 in the Hair Emitter and to 100 in the Hair System. Groom and then increase the Density in the emitter a little. Turn it back down again to give a faster reponse while grooming. When grooming is finished you can bring the density back up and play around with both density sliders to get the right look.
3) Always make certain that hair/particles are turned off in any other window that you have open. (Close any unnecessary windows).
4) If possible keep the number of Control Points down. (2-3 is usually enough. Very long wavey hair may require more).

If you you find yourself waiting for A:M to catch up you have probably ignored one or more of the above.

On my 'muscle gurl' character with the long wavy hair I have 7 control points... the higher you go the better the collision-detection works BUT the hair tends to stretch-out and recoil-back in actions.
http://www.youtube.com/watch?v=F6zXxpc_EA8

Collision Detection is probably THE reason to use higher # of cps - I usually stay away from collision detection, and set an angle limit instead for dynamic hair. (not perfect - but good enough for me)

As for wavy, curly - I play with the kinky properties (still using only 2 cps). But then again I'm not as KINKY as Matt of course.

I find having to groom more than 2 cp's just an unnecessary hassle. But I haven't done a hair STYLE that would require it. I didn't see anything in Matt's girlie hair style that would require more than 2. Braiding might need more than 2. But lazy ol' me would probably do hair type braids an entirely different way (short hairs groomed to look like braided pattern)



Rather than monkey with those factors while grooming - I find it easier to just set the real-time display density percentage factor to some low enough number. Then you don't have to readjust the actual densities when you go to render (or do render lock on screen). Also makes for faster/easier real-time response when checking animation with hair on.

If I used only 2 in the gurly hair it would pass-thru the shoulders and be stiff as a board. I always compare it as thus: If you want your hair to bend like a plastic comb...use less...if you want your hair to bend like a piece of wet spaghetti, use more.

Another good criterion for choosing # of cps (besides collision detection)

hmmm... Tin Servant hair also has 6 CPs. Maybe that's why I can't get anything going with his hair in it.

In hindsight his hair could have been done as helmet hair since it's completely rigid in his shots.