FA+
Shop
Why U No UPDATE?
12 years ago
Well, the long and short of it is that my current project doesn't
involve furry characters. So with all the spaceship stuff already
on my account, I just don't want to push the issue and post more
super bald tech stuff.
While I would like to continue the furry character line and make
a good mesh, I don't have the time for even more parallel projects.
Since January, I had two releases of the Space Mouse controller (and
there are still a few issues I need to correct), and I was building
and animating for that new project
(mind you, all of it in my free time, oh sometimes I wish I would be
able to make money from it so at least I didn't have to work on so
many fronts)
which, I guess, will be presented in due time.
But animating. Oh, animating.
It's not just that you have to actually animate (well, rig, weight,
bind, morph, assemble a controller, and THEN animate). You also have
to create the backgrounds. And while I am a stout apologist of the
theorem that you don't need as many detail in animation, because you
don't have as much time to look at each frame, shot, and scene...
you still DO need detail, and when you switch the scenario every
twenty seconds, that's still a helluva modeling work for stuff that
is just standing around.
So, you need to build a library of STUFF so you don't need to come
up with everything from scratch every time. Which means that you better
choose a style, size, degree of detail, and polygon count early
on because otherwise you may end up with STUFF that just doesn't look
good together (also a danger if you buy or download or collaborate with
others on assets). You need to develop a feeling for the necessary
amount of detail, and for the frequency you can actually reuse an
item.
People do notice. On the Star Wars forum I read how people find it
funny that in The Clone Wars every planet has the same kind of wooden
cart. Well, yes, some people do have too much time. But it emphasizes
that you need to have an eye on your backgrounds too. Just don't be
lazy.
And then of course there is the render. Yeah, this fits in nicely
with my last post about new computers and stuff. Sadly I don't have
GPU rendering or distributed rendering, and using my older machine
as render node is only of limited usefulness (since the new one is
so much faster than the previous one that the gain would be, uh,
a speedup of about a quarter). So, here I sit and render.
Tried global illumination with three bounces and subpolygon displacement
at first. Ended up reducing to two bounces and bumpmapping. Still I
estimate the finished render to be at 36 hours for 250 frames (10
seconds of animation). At half HD (quarter area); no motion blur;
a really simple scene; no blurry transparencies or reflections, no
SSS or translucency, no refraction or caustics.
Duh.
I wish I were more prolific with the VRay material system. I should
render that out in VRay instead of the internal renderer (even if you
CAN animate in GI now...). But that depressingly reminds me of how
much of a noob I am in so many fields. VRay among them. Must spend more
time on my tutorials; I want to render all-HD soon...
Shucky-darn.
involve furry characters. So with all the spaceship stuff already
on my account, I just don't want to push the issue and post more
super bald tech stuff.
While I would like to continue the furry character line and make
a good mesh, I don't have the time for even more parallel projects.
Since January, I had two releases of the Space Mouse controller (and
there are still a few issues I need to correct), and I was building
and animating for that new project
(mind you, all of it in my free time, oh sometimes I wish I would be
able to make money from it so at least I didn't have to work on so
many fronts)
which, I guess, will be presented in due time.
But animating. Oh, animating.
It's not just that you have to actually animate (well, rig, weight,
bind, morph, assemble a controller, and THEN animate). You also have
to create the backgrounds. And while I am a stout apologist of the
theorem that you don't need as many detail in animation, because you
don't have as much time to look at each frame, shot, and scene...
you still DO need detail, and when you switch the scenario every
twenty seconds, that's still a helluva modeling work for stuff that
is just standing around.
So, you need to build a library of STUFF so you don't need to come
up with everything from scratch every time. Which means that you better
choose a style, size, degree of detail, and polygon count early
on because otherwise you may end up with STUFF that just doesn't look
good together (also a danger if you buy or download or collaborate with
others on assets). You need to develop a feeling for the necessary
amount of detail, and for the frequency you can actually reuse an
item.
People do notice. On the Star Wars forum I read how people find it
funny that in The Clone Wars every planet has the same kind of wooden
cart. Well, yes, some people do have too much time. But it emphasizes
that you need to have an eye on your backgrounds too. Just don't be
lazy.
And then of course there is the render. Yeah, this fits in nicely
with my last post about new computers and stuff. Sadly I don't have
GPU rendering or distributed rendering, and using my older machine
as render node is only of limited usefulness (since the new one is
so much faster than the previous one that the gain would be, uh,
a speedup of about a quarter). So, here I sit and render.
Tried global illumination with three bounces and subpolygon displacement
at first. Ended up reducing to two bounces and bumpmapping. Still I
estimate the finished render to be at 36 hours for 250 frames (10
seconds of animation). At half HD (quarter area); no motion blur;
a really simple scene; no blurry transparencies or reflections, no
SSS or translucency, no refraction or caustics.
Duh.
I wish I were more prolific with the VRay material system. I should
render that out in VRay instead of the internal renderer (even if you
CAN animate in GI now...). But that depressingly reminds me of how
much of a noob I am in so many fields. VRay among them. Must spend more
time on my tutorials; I want to render all-HD soon...
Shucky-darn.
Check this out: http://www.wowhead.com/spell=28740 (Has also a 3D view).
It's a rat. It has so few polygons, but it's still a rat. Just plain textures seems to play very significant role in rendering quality, than other advanced stuff. I see often amateurs use thousands and thousands vertices for small objects and that makes me laugh. It looks no better and increases render time a lot :>
The stuff got too advanced. I guess it's be better to learn how to use fewer rendering tools/features good, than trying to use all at the same time blindly and hoping that it will do everything automagically for you :s
p.s.: I have no idea what I'm doing
Game engines are pretty specialized systems. They do realtime very good, I have seen some astonishing usage of realtime AO, effects, even fur. They also make nifty use of the graphics cards' advanced calculation abilities to realize dynamics effects.
However, they are also limited by principle. Since they have to produce 60 or 80 frames a second, they rely on tricks, like pre-rendered GI baked to textures, alphachannel materials to simulate clothing layers, premature interruptions for localized AO, constraints for the number of polys you can use in a moving object, all that stuff. It's not the same as a true CGI renderer - a game just cannot afford to provide a full GI solution with 3 bounces.
The older the game - or the lower its demands of the hardware - the more limitation the game engine has. Really early 3D shooters didn't even allow vertically overlapping polys - no bridges there - and their enemies were flat sprites. A modern high-end game does have thousands of polygons actually, and comes close to renderers in their viewport display, but they still use reduced light solutions.
Watching games and renderers evolve in parallel is pretty interesting. Not only do the games drive much of the hardware development - they also absorb much of the renderers' look by developing simpler, faster solutions suitable for realtime display. Renderers on the other hand do use many tricks also found in games, like baking light solutions, to get the render times down. And they try to utilize the hardware that has been created for gamers' needs. Blender even integrates a game engine with its main package. You can get the render times down in a normal renderer too, if that is your goal. Hours of rendering time are actually not a good sign for the setup or your choice of the light solution.
If you are into film-making with game engines, try machinima, which does precisely this. It has a great advantage in its realtime feedback (depending on the engine and the controller), although you may run into issues when it comes to closeups, effects that are not predefined, subtle motion (facial expressions) or weird camera angles.