sprockets Shelton's new Char: Hans It's just donuts by ItsJustMe 3D Printing Free model: USS Midnight Rodger Reynolds' 1950s Street Car Madfox's Pink Floyd Video Tinkering Gnome's Elephant
sprockets
Recent Posts | Unread Content
Jump to content
Hash, Inc. - Animation:Master

OdinsEye2k

*A:M User*
  • Posts

    724
  • Joined

  • Last visited

Posts posted by OdinsEye2k

  1. One animation idea I could think of would be the finger of God sweeping through that blue and touching the point where the Earth is, and then have a flash of light bring the world into being. Maybe a subtle halo around the planet as the 'afterglow' of this event in the still version...

     

    That being said, can I please ask what a 'Creation Museum' would have in it? There can't be any exhibits explaining how creation works (I wouldn't think, because there is no plausable mechanism to use). Or is it more of a devotional, looking for ways to connect with God through the original event of creation?

  2. Ya know, this is one of those super-duper math problems that might be fun to work out sometime.

     

    I finally took a class on dynamics which uses the quaternion/Euler angle concepts intensely.

     

    My long-standing suspicion on the performance of smartskin vs. fan bones has been the math underlying both operations.

     

    For a fan bone (this is the brilliance in cogs is that in most dimensions except roll they show this very dramatically), you are rotating points about a center of motion. There is a "sweep angle" the points go through to follow their neighbors.

     

    For smartskin, you are dealing in Cartesian rather than "spherical" co-ordinates (I'm using these terms a little sloppily I think with Eulers, maybe not). So, at any given position, you have 3 functions:

     

    x = f(x,y,z,w)

    y = f(x,y,z,w)

    z = f(x,y,z,w)

     

    These work fine on their own, but summed together (maybe even blended), I suspect the coupling terms in quats between w and the x,y,z vector causes behavior that does not match the rotation of the neighboring points.

     

    So, I think the basic premise of "fan bone first, tweak with smartskin" is probably correct due to keeping the errors small.

     

    But, it would be cool to try and derive a way to make everything match....

     

    I'll have to take a look at it sometime in a couple of days when I'm not busy studying but still have this stuff on my mind.

  3. Okay, well - this is the best place to go for vector math, but since that isn't available ATM - I'll try a different trick:

     

    For the shadow on/off thing, I'm not sure if this is doable with Expressions, but try setting an Expression on shadows on/off (now, keep in mind that this will probably nuke a part of the reason for skylighting in the first place) and just letting it be equal to 0 or 1 (maybe 100). If this changes the property, you're in business.

     

    Just do a =Floor(Cos(Rotate.X) * Cos(Rotate.Y) * Cos(Rotate.Z) / 0.9 ) or something like this. Note that this works if "top" is the Euler angle (0,0,0). This Expression will essentially go to 0 once the light moves away from this angle by very far. To adapt this result to any other Euler angle set, just add a - offset to each term, i.e.:

     

    Cos(Rotate.X - 45*Pi()/180) should set the proper point at 45 degrees - the Pi/180 term is degrees -> radians conversion.

     

    A similar formula (but without the Floor and division part) should work for dimming the light as desired - you should multiply by a starting value for the top light.

     

    Is this what you are looking for?

  4. Do you mean like change the light throughout the day, as in a rapid play-through of the day with the sun rising in setting in short order?

     

    Or are you talking about the random positioning of the lights within the skylight? And adding dimming and color changes to enhance the effect?

     

    If it's the random bit, I would suggest a central value + Rand()*some constant to make things jostle around the average value.

     

    If you want more consistent behavior, like simulating the effect of clouds as your skylight bounces around, that would require something more specific. If you further detailed what you were after, I could make some theoretical suggestions.

  5. Not only is Yves an excellent person to have on the team anyways, but if you compare his works to those listed on the "Meet the Hash Team" roster, you see that this is a huge addition to the team.

     

    Looks like in the 3-d world here, the Red Sox draft A-Rod (just smile and nod - this is a good thing B) ).

  6. Hey all,

     

    I did a revision of the Expressions tutorial to be more focused on specific examples of phenomena rather than going function by function down the list of functions available. I've tried to go more right-brained on this one, but may not have gotten too far.

     

    Anyhow, there are a few basic tricks in there - a lot of the advanced rigging type tricks await the ability to use GetWorldPos and GetWorldDir on bones.

     

    http://www.prism.gatech.edu/~gtg724n/math.html

  7. The 'quick cut' technique with two adjacent keyframes will not work if you are using motion blur, by the way. It makes kind of a neat streak transition.

     

    Multiple cameras are probably the purest way to go - you just have the nuisance of having to run a separate iteration of Master if you want to render all the shots overnight (there's still no batch render for single-user, right?).

  8. Lighting and tesselation in video games is also very ignorant in terms of surrounding features from what I remember in an interview (more like a technological Crossfire) between graphics card programmers and 3d software writers. There are a bunch of tricks that reduce the dependency of the light and shading of any given triangle on its neighbors, so the entire shading process becomes parallelized instead of serialized (as in ray tracing, as one bounce can only be calculated after its predecessor bounces). Once a computing problem is parallelized, it becomes pretty simple to solve by simply throwing a wider computational pipe at it (think Beowulff cluster). This then requires a lot of pre-loading and tricks to re-enter the lighting information that is lost from this simplification.

     

    I think Martin's expression of getting a suit at WalMart for $80 is dead on. The complaint I heard in the article from the ray-trace programmer is that graphics card methodologies are too poor in terms of sophistication to do rendering without the attendant hacks.

  9. One final thing to note for the A:M market, if you are indeed interested in that:

     

    AM users tend to be for making their own content, and prefer tutorials and tools (I believe the Anzovin Setup Machine is more than likely AM's best seller after AM itself) to pre-made models - which never seem to be in the buyer's style anyways.

     

    Also, as for UV mapping, AM does it in a slightly strange style, but once you have it figured out you can see that it is fairly powerful and can be managed the same way a "conventional" UV map is (see the Zandoria UV tutorial for that one).

  10. And never trust an artist to draw anything metal correctly :P

     

    Shouldn't the conduit have the spare wires poking out of the top rather than the sides (if it's supposed to show a work in progress)? Or is the conduit slotted? In that case, I would add a little black gasket or two to ensure that those sharp edges don't strip out the wire.

  11. I guess the biggest question is how to do the syntax for the GetWorldDir method correctly. The other part about the temporary variable poses may be something to try and work out later.

     

    Unfortunately, I'll be away from my animating computer for the next couple of weeks - but I'll send in a project file to Hash when I get back with some illustrations of what I want to do. Thanks for the reply Noel.

     

    Bjorn

  12. Do you have a side view? The particles don't seem to be eminating from the holes but rather dropped from above.

     

    Also, maybe a rough displacement map might work out better than just the colored decal.

     

    As you said a rough draft - all the pieces are there with a lot of potential.

  13. Well, the last big presentation I gave involved AM animations of our space vehicle design with the Knight Rider soundtrack in the background after we declared that terrorists had taken over the International Space Station. A fantastic moment that was probably a little over the top for an academic conference - if we didn't have the excuse of still being in school (and the umbrella of Georgia Tech tradition), we probably wouldn't have had such a good reception. A couple of the judges still grumbled about it a little.

     

    However, I do guarantee that everyone at the conference remembers exactly who we are and what we had done. On the other hand, no guarantee that that was exactly a good thing....

  14. Okay guys,

     

    I've burned about five different days trying to figure out how to do my "secret sauce" cart expression trick. I've gone through everything (not much) that currently exists, official and unofficial. I'm hoping either a Hash writer or uber-user sees this.

     

    Okay, here's what I've decided I want to do:

     

    -Establish 3 percentage properties as "variables" - containers for intermediate values I plan to calculate. One to calculate the wheel radius (distance between the location of two nulls), the distance between rotation point and rear wheels (again, distance between two nulls) and the distance between rotation point and the front wheels.

     

    The ingredients:

    -GetWorldPos Expression. I want to get the world position of two nulls, take the difference of those positions (which I assume will be a vector) and then norm it to get a distance. That way I get an "expressions rig" that will recalculate the distances just by moving nulls. Preferably I would be able to set this up in an on/off relationship to turn on the "rig." What is the form of the variable I should be tossing in here? Bone.Transform? Bone.Transform.Translate? Bone.Transform.Translate.X?

     

    -Empty percentage user properties. I don't want to make a relationship on these guys, I just want them to hold a value to keep the Expressions cleaner. I know how to make these, I'm just hoping that using them does not introduce some quirk that I don't know about.

     

    -Expressions within relationships. I've gotten it to work before, but everytime I tried and set one of my "variable" all I can get are "syntax errors" even though I aquire target names by clicking on the properties in question within the relationship I am trying to build this rig in. This method usually works for me, hence the confusion.

     

    Anyways, all of this is just to get something that will account for how much ground the wheels will cover when the model rotates. Getting things to happen when it's moving in a straight line is relatively easy. Luckily, motion can be decomposed into rotation and translation.

     

    Thanks in advance - this is the final piece to my long-winded Expressions and mathematics tutorial.

  15. Hey everyone,

     

    Trying to round out the Expressions Tute some more - one big question I came across is - does anyone know what the notation for a vector is?

     

    I've tried:

     

    (1,1,1)

    {1,1,1}

    [1,1,1]

    {1;1;1}

    [1;1;1]

    (1;1;1)

     

    to no avail. Any thoughts?

     

    Thanks,

    Bjorn

  16. I've been trying to use Conform with a recent version of AM and I get a crash as soon as I hit the 'ok' button. This wizard worked up until v11b I believe, but I haven't used it very often, so I may be doing something wrong that I shouldn't.

     

    I have a mesh grid built around (0,0,0) and another grid built at (0,0,10) and am telling the wizard to move along the vector (0,0,-1) (crash) or to the point (0,0,-10) (crash).

     

    Can anyone please let me know if they are having similar trouble or what I am doing to kill the plugin?

  17. Final try if you can't get the percentage thing to come on. This may not work, but that's desparation for ya:

     

    Close all windows. Click on the sprites material and then go to the properties menu. Click on the "over life" submenu below opacity. If the timeline falls to change to percentage, I believe you can still input data. However, each second is one percent. So, to get to one hundred percent, you would need to advance to one minute forty seconds (one hundred seconds).

     

    Again, I am not positive that AM will interpret this correctly, but this is the same way they used to encode bone rotations for smartskin in the data files, so it may work.

×
×
  • Create New...