View Full Version : Absolute morphs

01-03-2008, 02:08 PM
Ok, I was messing with the LW DTS Exporter to get morphs over to Torque.
I got the relative morphs to go over but I had a problem with them, so I figured i would try to use a "Absolute" morph.

I can't get an "Absolute" morph to work at all inside of LW. I create an absolute morph in modeler and it shows up, but I can't get it to show up in layout. The morph isn't even recognized anywhere in layout. I've been trying to figure out how to just use an absolute morph in just LW. I'm geussing it doesn't work.

Has anyone ever used an "Absolute Morph":bangwall:

01-03-2008, 03:59 PM
A little plug-in, that is dead now, called SmartMorph is the only thing that has EVER used what is in the absolute morph dataset.


Sad but true. SmartMorph is ... (sadness tells me to be done with this note...)

01-03-2008, 05:01 PM

Geuss i can just stop trying now. I've been trying to figure out what I was doing wrong for the longest. think i hurt my head

01-03-2008, 11:05 PM
Smartmorph is amazing and to my knowledge...not dead:

01-07-2008, 07:45 AM
Maybe this is common knowledge to most LWers', but what is the difference between a relative and an absolute morph? And why would I pick one over the other?


01-07-2008, 09:14 AM
..... and relative and absolute morphs are used in LW but only in Modeler afaik

Relative morphs are used/created when you want that morph target to be modified when you modify the base model eg if you have a face model and create a new smile morph target for it(with the relative option on) any changes you make to the base or original position will update the new smile target that was created)

Absolute does not get affected by changes made to the original base position however .. :)
and yes smart morph is great from what i have seen ;)
almost like XSI morph system ......

01-07-2008, 09:59 AM
The difference is in what information is stored. Firstly, as endomorphs are vertexmaps they store some information for each point (or only on a subset of points; usually only the once that where edited). Relative morphs will store the relative coordinates, ie the offset between the base position and the morphed position, how to get from one place to the other (a vector/delta). Absolute morphs will store the absolute coordinates, ie the actual morphed position (a point in space).

So if the (unmorphed) base position of a point is at <1,1,1> and the morphed position is at <2,1,1> the relative morph will store <1,0,0> whereas the absolute would store <2,1,1>.

When to use which!? That's a tough one, since we haven't had native access to absolute morphs we haven't really had the chance to learn when to use them over relative ones. Allow me to speculate! I imagine that the choice should be based on how you want morphs to blend. Relative morphs (usually) blend additively, whereas absolute morphs would probably be used to blend by averaging between all morphs. As such I actually imagine that absolute morphs could be a better choice for realistic facial animation because by averaging you would never risk deforming the face too much. No point would move further than it has in any of the active morphs. With relative morphs the sum of all active morphs could very easily result in a point being offset too far. Lets say we need to blend two morphs which both has the mouth go wider. The character could be having his picture taken, thus smiling while saying 'cheese'! So we would activate a smile morph and the phoneme 'e'. If the sum of those relative morphs are brought beyond 100% we'd probably get abit wider mouth than we'd like. Naturally we could lower the smile value everytime he says the letter 'e'. But this is not optimal. We would like to keep him/her smiling just as much all the time and just animate the phonemes without having to worry about a particular phoneme not working with our mood/expression morphs. With absolute morphs we could keep the smile at 100% and activate any phoneme and it would never stretch beyond the limits of the face (as dictated by all active morphs; given that all our morphs where within the limits). So far so good... . Unfortunately we probably want the mouth fully closed for m:s, b:s, p:s and so on. With the smile morph there and everything averaging that won't happen. So again we will have to adjust the smile, but now for different phonemes. (Perhaps via the Node Editor we could setup some morphs to receive extra bias or cancel out other morphs!?)

So I guess we could sum it up as relative morphs blend additively whereas absolute blend by averaging. Assigning a value over 100% will allow a relative morph to displace further (very useful at times), whereas for absolute morphs it would increase that morphs bias (if blended with something else). Blending relative morphs risk stretching beyond the limits and blending absolute morphs risk not moving enough. Take your pick. :)

01-07-2008, 08:03 PM
Thanks for the info everybody, great explanations!


01-31-2008, 11:27 AM
Seems that absolute blending of endomorphs is a *far* easier coding task. So can it be done? Why can't Morph Mixer deal with this??

01-31-2008, 02:19 PM
you could make an 'absolute morph' by storing 1 frame of an altered pose as an mdd then blending between those instead of morphs.
An mdd is basically a sequence of absolute morph targets.

01-31-2008, 04:52 PM
Wow.. all this time I didn't know the difference between relative and absolute morphs.. now the names make perfect sense. I guess I don't use morphs enough.