zaterdag 12 september 2009
maandag 10 augustus 2009
Hacking Maya
Hacking is perhaps a bit too strong of a word. But it is defeniately possible to edit maya`s user interface to your own liking. A while back, I was annoyed with painting skin weights on a model with loads of influences (joints). Balancing the skin weights meant scrolling through the influence list and picking the joint I needed to paint. After some research on the net I found that it is possible to edit the user interface and add functions to maya. The entire GUI is constructed through MEL. So, finding and editing the appropriate scripts will change maya. I used it to build in a filter, which can shorten the length of the list considerable and make painting skin weights a lot faster and less cumbersome. I also added "lock all" and "unlock all" buttons.
I edited the artAttrSkinCallback.mel and artAttrSkinProperties.mel scripts found in the scripts/others directory. If you want to try and edit these yourself I`d suggest copying those to the scripts directory found in "my documents" and edit them from there. Any script in that dir will take precedence over the one in the maya folder.
I edited the artAttrSkinCallback.mel and artAttrSkinProperties.mel scripts found in the scripts/others directory. If you want to try and edit these yourself I`d suggest copying those to the scripts directory found in "my documents" and edit them from there. Any script in that dir will take precedence over the one in the maya folder.
donderdag 23 juli 2009
Pistons and sleeves
Here you see a short demonstration of a technique I am currently researching to quickly improve deformations on a smooth bound mesh. It involves using piston-like joints to pull the geometry out of the deep creases.
Pistons are basically 2 joints pointing at each other. I`ve constrained these so, that they only bend in one direction. For clarification I`ve parented some geometry under the pistons so you can see them move more clearly.
In the video you can see the three stages: in the front the joint chain used. In the middle the skin with deformation by the pistons. and in the back skindeformation with only the 2 main joints as influences.
For simple joints like elbows one piston would suffice, since this only has to bend in one direction. For joints with complete freedom like the wrist a sleeve of these pistons is needed. Twisting is still an issue when using this technique. However, in the case of the wrist the twisting would occur along the forearm anyway, so it isn`t a problem there.
Pistons are basically 2 joints pointing at each other. I`ve constrained these so, that they only bend in one direction. For clarification I`ve parented some geometry under the pistons so you can see them move more clearly.
In the video you can see the three stages: in the front the joint chain used. In the middle the skin with deformation by the pistons. and in the back skindeformation with only the 2 main joints as influences.
For simple joints like elbows one piston would suffice, since this only has to bend in one direction. For joints with complete freedom like the wrist a sleeve of these pistons is needed. Twisting is still an issue when using this technique. However, in the case of the wrist the twisting would occur along the forearm anyway, so it isn`t a problem there.
zaterdag 4 juli 2009
Something new
I always like to experiment with new imaging techniques. Especially when it involves light. Over the weekend I had an idea which, I figured, should work...in theory. As a proof of concept before going all overboard on a larger scale I created a very shaky construction held together by a lot of tape. I dont want to give away the exact technique, I want to experiment with it some more and make it my own first. But here are the results of the first trial. Proof of concept succeeded, the image are defeniately interesting enough to develop further imo.
These are in no way edited in the computer, only cropped.
dinsdag 2 juni 2009
Rig Demo
It has taken some time, but finaly the rigging system is ready for demonstration. I really learned A LOT making these scripts. Especially making the FK<>IK switching work meant researching how maya handles IK and FK and then making them match.
It also meant thinking a bit beyond just simply applying proven methods you can find on the internet. Because I had a very specific idea about how I wanted this system to work, I had do adapt the rest of the scripts to that. And since I had to find my own constructions that also made me think about the best solution for the end users, the animators. The main idea is to keep the screen clutterfree with as few controllers as needed while allowing the animator maximum controll. Also the most important part in any character, the spine, had to be carefully constructed. It needed to be flexible and allow independant chest and hip control. But most importantly, it needed to be solid; maintain its length.
The system itself is extendable so I can customise the rigs fully to the clients` wishes and incorporate new limbs all together. I based this first set of limbs on a biped-character, so now pretty much any character with any number of arms, legs, fingers, etc. would be built in no time. But limbs like fins, wings and tails are also possible once the scripts are done.
The facial rig is very limited right now, but is under development in my spare time.
It also meant thinking a bit beyond just simply applying proven methods you can find on the internet. Because I had a very specific idea about how I wanted this system to work, I had do adapt the rest of the scripts to that. And since I had to find my own constructions that also made me think about the best solution for the end users, the animators. The main idea is to keep the screen clutterfree with as few controllers as needed while allowing the animator maximum controll. Also the most important part in any character, the spine, had to be carefully constructed. It needed to be flexible and allow independant chest and hip control. But most importantly, it needed to be solid; maintain its length.
The system itself is extendable so I can customise the rigs fully to the clients` wishes and incorporate new limbs all together. I based this first set of limbs on a biped-character, so now pretty much any character with any number of arms, legs, fingers, etc. would be built in no time. But limbs like fins, wings and tails are also possible once the scripts are done.
The facial rig is very limited right now, but is under development in my spare time.
woensdag 6 mei 2009
RACING!!!!
This is a short animation I did with Lisa Hochstenbach and Martijn van Gurp for the dutch national television show `het Klokhuis`. We were asked to come up with a character and make a short sketch to fit a given theme: city cleaning. We all did our part in this animation, but I was solely responsible for all the CG-elements and compositing.
48hr film festival
I was asked to design a logo for a team participating on the 48hr film competition. Above is the result of this. I also did the end credits for the film. Below you see a still of what I came up with. Unfortunately this version wasn`t used in the end, but I did like how it turned out.
Backbones
I`ve been doing some research on spines. What would be a good spine to animate with? I learned most through a discussion on the CGTalk forums on a new spine technique, named the "Awesome spine". Main benefit of a IK-spine stup would be the independant hip and chest controll. Most IK-setups can do this and focus seems to have shifted to making such spines twist way beyond physical breaking point. The "Awesome spine" actually does do a good job on that, but it lacks solidity. As you independantly move the hips and chest, the length of the spine changes and is always stretchy. The parts would seem to be unconnected and drift in an animation. For this reason most animators still prefer a FK setup instead. With my "simple spine" setup I hoped to include all the benefits of a IK setup, while keeping the spine solid. For twisting I used a similar trick to the "Awesome spine" and upon further investigating I found more similarities. However, in the "simple spine" the spine retains its length and a nice curve, unless you increase the "max stretch"-attribute.
I haven`t gotten all the kinks out of it yet, but above video should give a impression about where I`m heading with this.
I haven`t gotten all the kinks out of it yet, but above video should give a impression about where I`m heading with this.
It moves!!
After a lot of messing around I finally fixed a solid FK<>IK switching system. It looks a lot easier than it is. Either that, or I have been doing something very wrong. The example in the video shows how it works on a finger like limb. But now that I`ve solved all the problems it can be easily adjusted to fit toes, arms and legs aswell.
It features FK<>IK blending and switching, a `finger curl`-attribute and a stretch limiter. With this attribute you can set the maximum scale you want the limb to have. Setting a small value, like 1.05, could be usefull for sticking the feet a bit more solid to the floor in a extended position of the legs.
It features FK<>IK blending and switching, a `finger curl`-attribute and a stretch limiter. With this attribute you can set the maximum scale you want the limb to have. Setting a small value, like 1.05, could be usefull for sticking the feet a bit more solid to the floor in a extended position of the legs.
Laying out the Topo
A WIP vid of the rigging scripts. It allows the user to connect all kinds of limbs together. This way you wont be stuck with a standard biped topology. The scripts first generate locators which indicate the joints. They can be moved around and rotated untill it fits the character.
The limbs can be further customized by `breaking` the joints (for smoother bending) or indicating that its supposed to be a stretchy limb (visually, not through a list of features you want on the final rig). The extent of the break can be adjusted by a slider. The the rig can then be assembled by just hitting a single button.
I`m currently at the stage of checking the joint orientations. Just before the final rig is assembled, the user can check all the joint orientations and adjust where needed.
dinsdag 10 maart 2009
Rigging Praradigm
The rig you see above I made last year. It is a rig for the upcoming short film of Floris Kaayk. For it I used a paradigm (set of rules and conventions) I made the year before that. I learned a lot from this project about rigging. Recently I started a project to script a rig-builder in MEL. This program would allow the user to build completely custom rigs with as many or few limbs as prefered within a few clicks.
When I started I had to consider a lot of things in advance. Since you can`t know what kind of rigs the user is going to build, everything needs to be really general and nodal.
The final system would have to be very flexible and easily expandable.
I guess I would call the paradigm I ended up with a layered-limbic-rigging-system. Layered, because eventually there will be multiple joint-chains operating on the same model. The layers I had in mind would be as follow:
FK and IK chains would operate on a CTRL-chain. These three layers will allow the animator to controll the movements. Then on top of that could be a DYNAMICS and NOODLE chain. These will allow for secondary motion and automatic followthrough. I still have to do some research whether this would best be solved with joints or skinclusters.
The limbic-part of my paradigm is that the rig-building will happen on a per limb basis. This will allow much more complex rigs than your standard biped or quadruped. My plan is to make a small MEL script for each type of limb. This will allow for easy addition of extra types of limbs as long as you keep in mind the name-spacing.
Based on this paradigm I formed the following namespaces:
limb_part_index_layer
The `limb` part is obvious, I guess. This specifies whether we`re dealing with a leg or an arm or a hand or a finger, etc. Each limb will consist of various parts. And most likely there will be more than one limb of the same type in a character...hence the need for an `index`. The layer indicates in what layer it is in, like TOPO, IK, FK, CTRL, DYN, NDL, BND.
The obvious omision in this name is the name of the character. This is entirely on purpose, because I`m assuming the characters will be referenced in the final scene. By referencing all object will be prefixed with the file-name.
When I started I had to consider a lot of things in advance. Since you can`t know what kind of rigs the user is going to build, everything needs to be really general and nodal.
The final system would have to be very flexible and easily expandable.
I guess I would call the paradigm I ended up with a layered-limbic-rigging-system. Layered, because eventually there will be multiple joint-chains operating on the same model. The layers I had in mind would be as follow:
FK and IK chains would operate on a CTRL-chain. These three layers will allow the animator to controll the movements. Then on top of that could be a DYNAMICS and NOODLE chain. These will allow for secondary motion and automatic followthrough. I still have to do some research whether this would best be solved with joints or skinclusters.
The limbic-part of my paradigm is that the rig-building will happen on a per limb basis. This will allow much more complex rigs than your standard biped or quadruped. My plan is to make a small MEL script for each type of limb. This will allow for easy addition of extra types of limbs as long as you keep in mind the name-spacing.
Based on this paradigm I formed the following namespaces:
limb_part_index_layer
The `limb` part is obvious, I guess. This specifies whether we`re dealing with a leg or an arm or a hand or a finger, etc. Each limb will consist of various parts. And most likely there will be more than one limb of the same type in a character...hence the need for an `index`. The layer indicates in what layer it is in, like TOPO, IK, FK, CTRL, DYN, NDL, BND.
The obvious omision in this name is the name of the character. This is entirely on purpose, because I`m assuming the characters will be referenced in the final scene. By referencing all object will be prefixed with the file-name.
Abonneren op:
Posts (Atom)