blendshapes….and more blendshapes

Doing blend shapes.  Well more accurately done doing blendshapes.

knee_deep_inBlendshapes

..and….doing multicam from the playblasts.  Why am I doing this?  Well, the last scene comes in at 1700 frames.  This will be somewhere around 90 GBs of frames, and that’s not mentioning how long it would take.  So, using the edit points created from the multicam stuff, I’m figuring out what parts of the scene to render from each camera.

Also to whomever is reading this, some good advice to pass along.  Do not stay up until 5 a.m. working on stuff, it will do bad things to you.

Interrogation scene

Wanted to pop a screenshot up onto here to show progress.  Blendshapes are good and online.  Doing some preliminary lighting now, need to animate and then setup the camera angles too.  It’s interesting how much slower the viewports run with viewport 2.0 and lighting turned on.  Quadro FX 4800s.

Setting_up_lighting

This scene takes place after the virtual reality construct crashes out.

Slight edit and update.  I went to add aim constraints to Julia’s eye balls.  Only to discover it wouldn’t take the constraint, but Bills eyes would.  All characters are referenced in.  So after opening up Bills rig and mesh and comparing it to Julia’s I discovered that not only did I group the eyes, but somehow I bound them to the rig too in Julia.  In Bill’s the eye balls were simply parented.  I noticed that in Julia’s rig and mesh, the transform handles were grayed out, indicating that the eyes were bound to the rig, rather than grouped or parented.  Feels like a rookie mistake but, I am learning as a student as an art school.  So in the end, I unbound/unskinned the eye balls and they went to wherever in the scene file, found them and put them back where they’re supposed to be.  Then parented to the joint nearby rather than binding or skinning them.

Interesting note, Maya will not let you constrain something already constrained, if it’s affecting the transform.  It was flagged in the script editor, which is what led me to take apart the referenced file.  I feel fortunate that I did not need to rebind the ENTIRE mesh.  Next, resuming the Aimer constraints to the locators so I can direct where the eyes are pointing.

Knee deep in blendshapes

Soooo…..I’m now in blendshapes for the interrogation scene.  To my knowledge in LightWave they’re called Morph Targets, in Maya it’s Blend Shapes.  The idea behind it is to manipulate a mesh to whatever you want, but ideally to form various facial expressions.  It’s all per vertex, so adding or deleting geometry won’t work.  You can combine them though in varying amounts.

knee_deep_inBlendshapes

The heads and eye brows are different states of the face mesh and eye brows.  One facial expression posed per mesh.  The sliders manipulate how much of the blendshape I’m using for the character, mixing and matching is possible and to be utilized.

Progress, sorta

Though I’m not moving through the pipeline as fast as i wanted, I’m going pretty fast.  Finally have thee last character modeled and rigged.  Posing him correctly to hold an assault rifle is more challenging than I thought.  I may add a helmet later on but, really want to focus on getting the firearm modeled.  I ended up using myself as reference along with google images to see the body posture of how to correctly hold an assault rifle firearm.

One of the times I went camping, in Caro Michigan, I was taught some basic firearm etiquette along with how to correctly hold a rifle.  Most of the key is posturing your body to correctly handle the firearm recoil.  Guns are deadly weapons, they are not something to be taken lightly and do require a healthy focus and respect on safety to operate in a responsible and lawful manner.

I have thought about the idea of cutting a few corners with the gun design though.  It doesn’t need to be hyper realistic, it does need to be stylized.  Something with a blocky design to it should be fine.  And just as in real life, if it looks intimidating and aids in deterring a criminal from being violent, I don’t think that would be a bad thing.

Currently part of the pipeline is waiting on a classmate to get done modeling a warehouse so that I can build that scene.  He’s using the same warehouse in his senior thesis.  So, I will be working on the very last scene of the film where the FBI agent talking to Bill and Julia and keeping with the idea of doing a nonlinear pipeline in the process.

FBI_Character_rigged_bound

Progress

With the help of old friends from Novi for voicing characters and operating a video camera.  The live action parts have been shot.  Though I had to record their voice performances at different times, I was able to sync stuff up in post.  I’d post screen shots and documentation but I’m not in senior studio right now, will be in the morning.

speed rigging/binding/weightpainting

I don’t think I’ve done the process this fast but don’t have a lot of options.  Massive use of what I call ‘sleight of hand’ but it’s working.  Have a visitation and funeral to attend, someone family has passed.  So kinda had to do this in a hurry.  The idea behind slight of hand is only doing the absolute necessary for a shot.  No extras at all.  The film Jaws kind of inspired the term to me, in the film the shark is almost never on screen, but you know it’s there.  I did have opportunity to tweak and change the character design to differentiate them but, not very much.

Criminal character is rigged and bound ready for animation in the scene, surfacing still needs to be done.

Production_shot_speed_bindingrigging_Criminal_character Production_shot_speed_bindingrigging_Criminal_character_2

I did add a mask for the character to help define it more and lean it towards being a mischievous trouble making character, they are a criminal.

Volley Ball scene

This scene is far simpler than the initial one.  Ran into some of the same render mess though.  Mental Ray doesn’t seem to like high frequency detail a whole lot.  The ground plane is a procedural shader mix but, it goes on forever on purpose.  This also means there’s lots of room for moire patterning.  I originally decided to make the camera move and make the shot more dynamic but in the end, I had to keep it static becuase of the moire patterning and flickering.

Interestingly, I’ve noticed this same issue with game engines in first person shooters, so, what do they do?  Somehow they have the distant areas blurred, so no flickering.  Another idea i could try is Depth of Field and blur stuff that way but that will make post even more complicated than it already is.

It’s supposed to be a casual game of volleyball, nothing serious, rather dinky, nothing complicated.  There is a referree offscreen that catches Bill hacking however.

This shot is using nDynamics for he net, it is online and working as the screen shot implies.

Production_shot_55ish