oDDity

Well here's the thing you don't understand about keyframe animation - it's not about copying reality. If you wanted that, then just use mo cap. Animation is about filtering out and keeping only what's essential to making things seem real, and then exaggerating or diminishing and cleaning it up. That's why it can be done cleanly and effectively on one layer using a small strategic number of controls and channels. It's more of an impression of movement than photorealism you should be after. That's where the real skill lies. It's pointless trying to layer ever more subtle rotations and translations on top. If you do the main layer effectively, it's not needed. It's like people in sculpture who think the details are the most important part, so you see these shit model with terrible proportions, no form, no rhythm, no weight, they've just spent 90% of the time intricately sculpting details of wrinkles and pores on a model that is never going to be any good.. A good sculpture looks good at every stage of development, even the earliest layout, and the same should be true of animation.

It all depends how effectively they're being used. They can be used effectively by animators who already know how to animate properly - that is, in a disciplined fashion where they keep control of every channel on every control - but they are also open to abuse by lesser animators who think layers are the be all and end all and just use them to key the crap out of everything and toss more layers on top as a quick fix. What those animators are literally doing is constantly using layers to correct the sloppy animation they did on previous layers, because the previous layer is now fucked and messed up with lots of redundant keys and channels which shouldn't even be there. To me, that's not easier, because now you have tons of layers with tons of keys and channels crapped all over them, instead of just one layer with nice neat keys on the specific channels you need. Animators should have to learn to animate properly first, without layers, and then learn how to use layers only to compliment real sklills.

Well this is the point, the normal map size needs to be decided on a case by case basis, rather than being batch reduced to a certain resolution.

It'll not reduce their visible quality. Games like RtCW used 256 and 128 textures and it still looks damn good. Normal maps don't suffer from size reduction as much as the detail in colour maps. The 1024's were never intended as final game resolution, it was just in case they had to be edited etc Taking a quick look through the textures, there's a lot of 1024 normal maps that could even be knocked down to 256, considering the low frequency information they're carrying.

You Will NOT notice smaller normal maps. The size of them is insignificant compared to the actual texture.

It'd be crazy to do them all as dxt3. Most of the textures don't have an alpha channel, so you'd be doubling the size for nothing. As for normal maps, if they have to be left as targas, then all the 1024's should be reduced to 512 ,adn the 512's to 256. The difference will not be noticed.

NAh, you barely notice the difference. Also, are mip maps really necessary, since you can shave another 200kb off that by not having them. Doesn't sound much, but over 1000 textures, that 200mb saved.

You only need dxt3 if there is an alpha, otherwise dxt 1 will do, which is half the filesize. A 1024x texture should be around 600kb.

Now you've reduced your argument down to something reasonable like 'layers are helpful and make things easier' rather than 'it's virtually impossible to do some animations without layers', I agree.

This is where you need to learn to keep control of you channels and not all the bad habits you've picked up in MB of keying the crap out of everything. When I shake my fist vigorously, all I get is some y rotation on my hips. This leaves x, y and z translations free for larger movements, and you still have x and z rotation free. You have 6 channels to work with for each bone/ik control, and professional animators use them wisely and separately, and use each channel for the appropriate movement, setting as few keys as possible and using weighted curves to get from one position tot he next. Your assertion that most animation, even the simple noob sort of animation you've done, cannot be done without layers, is borne out of pure ignorance of professional animation techniques. I agree layers are a quick fix, they can let you be sloppy and lazy and get away with it, but they're in no way essential for good complex animation.

To take your earlier example of a shaking fist, you should be using FK and not IK for that. You would animate the main shaking fist from the elbow and hand and the secondary overall arm movement you wanted to add, you would do from the shoulder. Your problem is you only know how to use IK so you have to do everything by moving the fist. The arm is a three piece system of overlapping movement, with delay and followthrough, and IK is useless for that.

I did a quick video tut of animating a (rough) walk cycle with this new rig. Should help you get around in maya. You use the shift+middle mouse button for selecting and on the timeline and MMB for dragging tangents in the curve editor. Also MMB on the timeline to jump around without moving the animation to that time, so you can set hold keys. http://www.pixelwerks.be/pictures/walktut.7z

Ha, listen to this guy would you. He didn't even know what motionbuilder was until I told him and made some tutorials on how to work it, and now he's trying to tell me how it is. You do not need layers to do overlap and secondary motion. Layers are just the the easy lazy noob way of doing it. Real animators do not need layers because they keep complete control of each specific channel of their animation and do all their work with curves. When you use curves properly you do not need many keys, just the main pose keys. You're way out of your depth here, so don't bother arguing any more.

Believe me, the vast, vast majority of professional animators do not use motionbuilder, they use maya or some othwer 3d app. In fact, virtually no one uses motionbuilder for keyframe animation. It's purpose is correcting motion capture. They find no problem doing feature animations exponentially more complicated and subtle than a warm hands loop without motionbuilser layers. Those layers really just make you lazy and give you a bad habit of keying the crap out of eveything. In maya you keep it clean and only key the specific channel you need, so the y rotation of the upper arm for example, and you keep a nice clean curve for that in the graph editor and and do any amount of subtle work with it there. You set only the few keys you need on the specific channels you need and it makes editing easy.