View Full Version : what happened to my blendshapes?
02 February 2007, 01:34 AM
I'm using M.B to animate my face blendshapes from maya and I've hit a snag. I am following along the tutorial at area.autodesk.com/learning/tutorials/details/2496/ and it goes fine when I use the "surferdude" character example but NOT when I try to use my own maya character.
The problem occurs in M.B., during step2-pt.6, I would like to alt-drag my various blendshapes into "shapes menu" but lo and behold I cannot find my blendshapes anywhere in the schematic view!!
The tutorial tells me to alt-drag the "models" and I assume these would be the blendshapes, am I assuming incorrectly?
I did a little test to see if the problem started at the .fbx conversion stage out of maya, and I think this is indeed the cause. Because when I open my coverted .fbx file in maya I can see that my blendshapes have actually been removed from my top Reference (null/group) and placed outside its heirarchy(!) Why is this happening? I hope there is a more intelligent way to convert my .mb file into .fbx so that the blendshapes stay in the correct heirarchy... so that they can be located in the M.B. schematic view.
This is my first try doing this and would love to know how experienced folks deal with blendshapes in M.B.
02 February 2007, 11:31 PM
I understand this is a common problem and I've read lots of threads that talk about some various ways of dealing with this problem. Unfortunately posters have a habit of saying "thanks" when they find a solution but forget to post what solution worked for them. I'm am in the process of trying out all of these approaches one at a time so that I can find a solution. I would appreciate any input from someone who's experienced this situation and seen the light.
The main approach has been to troubleshoot the MB version with the FBX version and to this end I've updated both. So the problem persists even with MB7.5 and FBX2006.02(!)
I've read that blendshapes need to be named a special way, well what is it? From what I can decrypt the names must go something like "**_name" or "**:name" ... does this supposedly work like the bone names which MB automatically locates? This sounds promising...
Another point that I picked up was about freezing transformations in Maya, that it "kills" the blendshapes. I didn't see them killed in the .fbx conversion when I re-imported it back into Maya (just to test it) but it certainly doesn't show up in MB so I don't know if has been "killed". In any case I'm not freezing it and the problem persists.
That's it for now,
02 February 2007, 08:51 PM
I've been wrestling with my blendshapes as they come out of Maya 8.0 and go into MotionBuilder7.5 via FBX v.2006.02 and I'm 90% certain that the cause of my anxiety is from the FBX plug in. This is the message I get after converting my .mb scene to .fbx:
Warning: No Morph Target Found (1)
The morph Target object's controlling slider may not have been deleted. The morph target object for weight slider: BlendShape1.s.head.happy is not available in the scene. Ignoring this target.
To me this sounds like its ignoring my blendshapes, what gives? When I open my converted .fbx file in Maya I can see that indeed my blendshapes are gone, that's no good. How do I get that FBX plug in to keep my blendshapes so I can use them in MB?
05 May 2007, 07:03 AM
Context: Windows XP, Maya 8.0, Motionbuilder 7.5, FBX exporter v2006_11_1.
I was similarly encountering the "Warning: No Morph Target Found" errors when attempting an export to FBX within Maya, and when I imported the .fbx file into Motionbuilder there would be no facial blend shapes (aka morphs).
What I found is that Maya wants you to have, in your Maya file, copies of the original geometry objects that you used to create the blend shapes.
1. Loaded a .mb file containing a base character geometry + skeleton with no morphs applied to the geometry yet.
2. Imported a second .mb file containing 11 facial-morph geometries of the base character geometry.
3. Used "deform-->create blend shape" to add 11 morph targets to the base character geometry.
4. Deleted the 11 original facial-morph geometries since (in theory) I don't need them anymore, my morphs are on the base geometry now and I can use them fine within Maya.
5. Saved as .mb
6. Exported to FBX. The exporter gives "no morph target found" warnings for all 11 targets.
7. Opened the .fbx in MotionBuilder, added a Character Face, dragged the geometry into the character face "<drop model(s) here>" text line. No morphs show up in the "shapes" column.
1-3. Repeated steps 1-3 from above.
4. Did NOT delete the 11 original facial-morph geometries. I left these lying around in the file and didn't change their names.
5. Saved as .mb
6. Exported to FBX. The FBX exporter did not give "no morph target found" warnings at all.
7. Repeated step 7 from above. This time my morphs do show up in the "shapes" column.
So for the local MotionBuilder experts here: does this mean that when I create a character with dozens of morphs, that I'm supposed to keep the original deformed geometries lying around in my file instead of deleting them? That's a lot of extra junk geometry objects to keep in a file just so the FBX export works correctly. I don't appear to need these extra geometries to animate using the morphs within Maya itself.
05 May 2007, 03:34 AM
I would like to alt-drag my various blendshapes into "shapes menu" but lo and behold I cannot find my blendshapes anywhere in the schematic view!!
well first u need to alt-drag the modle (not the shapes ) shapes don't show in motionbuilder
also i think in 3ds max you don't need the morph targets (blendshapes) gemotry when exporting to motion builder, BUT in maya u must to have the blenshapes gemotry before you export them to motionbuilder.
also look at the video traning in 3dbuzz.com they explain every thing perfectly.
05 May 2007, 02:36 PM
I found that site to be very useful too! Many thanks for your followup, I've gotten the blendshapes working now, but I've discovered that MB is too limited in its generic shapes for me to use it. And the relations editor hardly makes up the difference. (did you see the 3dBuzz issue4 lesson at AREA? Sorry to say so, but the lip-sync stinks!) So I'm no longer using MB or blendshapes for my facial animation, instead I'm going the bone-route and just passing the mocap data directly onto my carefully weighted face. I guess I had to learn the hard way.
05 May 2007, 10:55 PM
It's a shame you gave up on the blendshape route. I was able to recreate the interface used in Jason Osipa's "Stop Staring!" book in Motionbuilder with GREAT success. I built the interface, originally, in MotionBuilder 5.5 but haven't had the time to upgrade it to 7.5 yet.
Don't bother using what Autodesk provides. Generate controls that manipulate your blendshapes manually and you will have ALL the flexibility, without any of the Actor Face interface limitations.
06 June 2007, 03:57 PM
That's a great book doing key framed animaton; and also using morphs certainly goes along with that traditional xsheet approach to get great results. But what I am trying to get working is just a different approach, no keys & no morphs. The main issue is that building morphs requires time and skill and when a production involves many characters there just isnt enough time and skill to go around to morph/key them all. The bone-face method is not working perfectly for me yet and I am trying to get my hands on some scripts for motionbuilder to properly register the face onto the head. Nobody seems to recommend this method even though this is the method favored by larger studios who already have proprietary scripts.
06 June 2007, 03:57 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.