Hi Eric:
No worries, I love being able to be corrected and learn more if I am
mistaken over stuff! Although I hope I'm not going too much off on a
tangent considering this is the XSI mailing list :P
- Shaderforge (or strumpy, whatever you prefer) is more or less
essential imho , especially for prototyping stuff to see if it's even
the right path to take (plus it's a nice way to have easy access to cool
stuff like vertex deformation for cool effects without even stepping
into Mono.) The problem is that it's not nearly as extensive as it needs
to be (really, still no GPU noise function?!), and most of the time you
also run into roadblocks that require you to go and make your own
shaders anyway...and then you're left with trying to figure out two
seperate shader systems. The bigger point is...why isn't this something
that Unity devs focus on instead of just bringing in a 'new' PBR
all-in-one shader?
- We did not use the Avatar for certain obvious characters that would in
no way fit a Biped even remotely (the fox, for example, is treated as a
Generic Avatar.) The nice thing is that your animation should come in
1-to-1 as you animated it, since this bypasses the Avatar system (and
thus the RoM stuff entirely), although there are sometimes odd things
that happen with gimbal flipping in animations (but those can be solved
easily). The bad thing, and correct me if I'm wrong here, is that you
lose all IK features. You also lose the ability to retarget animations
from character to character. Which, for the fox, we didn't mind so much,
since we didn't have any other characters that would share his
animations. As for IK on the fox...that's why we made his paws so tiny :)
As for 2 spine bones and the devs' reasoning behind it (it's not even
documented X_X):
http://blogs.unity3d.com/2014/05/26/mecanim-humanoids/
- Particle Playground is nice. However in testing it severely degraded
performance and we decided not to investigate further into optimizing
our engine for it. I did use it for some of my own personal freelance fx
work and it was alright; like I mentioned, I ended up more scripting
effect behaviours that I specifically wanted, and combined them with
Shaderforge shaders on meshes to be able to control more effectively the
final look without trying to figure out "why isn't this working?"
I think we just have to agree to disagree on the part of how Unity
handles metadata for files though :P To me, it's absolutely mad how the
devs are fine with the way the GUID/UUID system works, which to me
defeats the whole purpose of them in the first place (and was an endless
source of frustration for us since there's really no way around it,
since we cannot predict how exactly Unity would generate the new GUIDs,
and under what circumstances.) Since it applies to everything, from
shaders to levels to models...it's just a terrible unnecessary roadblock
to collaboration owing to the old implementation of the unity asset
server, and should be given a serious rethink.
Anyway, that's as far as I'll go on the subject of Unity; I was glad for
the experience tbh. (It makes me much more thankful other engines exist)
I'd love to discuss more about it!... but we should probably have that
conversation off the list. :X
(p.s. Thanks! Hope you enjoy the game when you get to playing it!)
On 1/20/2015 10:55 AM, Eric Turman wrote:
@ Siew Yi Lang
I felt compelled to respond to your review of Unity because of certain
inaccuracies and/or misconceptions. I hope that you do not feel
slighted as this is not a personal attack at all.
Although my day job now deals with commercial/film/vfx, I have worked
on several AAA game titles in the past and I am currently wrapping up
an indie game with two other people that we made in Unity. I'll admit
that I agree with some of what you have stated but not nearly as
strongly or vehemently. One point I will agree in is that while they
have added a bunch of new features they do need to fix what is broken.
That can be said of many software packages...I'm biting maya tongue
right now to stay on topic and not to go off on a certain package.
*Shaders:* I agree that ShaderForge makes shader creation much easier
and it is worth the added expense. Shaderforge
<https://www.assetstore.unity3d.com/en/#%21/content/14147> has enabled
us to develop some very specialized and unusual non-photo real shaders
that could have been a stretch for us otherwise. That being said, we
are friends with a programmer who codes his own non-node based shaders
for multiple platforms in Unity without a problem and he has helped us
with such shaders for other projects.
*Mechanim: *I'm going to mostly disagree with you here and my
experience is nearly opposite of yours with Mechanim. We definitely
did not use legacy. We were able to, without scripting, make our own
avatar. You do not need to use the built in avater, just don't use it.
Additionally, Unity did not retarget it to two bones for us. Unity
Mechanim dealt just fine with our custom skeleton setup. Animation and
hierarchical blending also worked just fine OOTB for us with our
custom shadow rig skeleton. After glancing at how Unity wanted the
shadow rig to be organized, our animations came in without a hitch
from Softimage. I did not have to do anything with ROMs. I mean no
disrespect, and no hard feelings are intended, it's just that I know
firsthand that it works.
*Shuriken: *I am not going to argue with you on this one. Unity's
built in particle system is, in a word, weak. Fortunately, an asset
called Particle Playground
<https://www.assetstore.unity3d.com/en/#%21/content/13325> comes to
the rescue and, as I understand it, its code is exposed if you need to
tweak it.
*Pipeline:* I'll agree that the hundreds and thousands of pieces of
metadata was a bit cumbersome on the check in/checkout. But this is a
technical industry, the artists are working with technology and it is
not unreasonable to expect them to have a modicrum of technical
knowhow. The artists at Element X were sat down and launched with
instructions on the use of SVN. After a few days, they were shown that
how they had misunderstood/cut corners were doing things with SVN were
causing problems and shown the procedure once more. After that, we
didn't have any issues. Its a partial matter of training the artists
what to do and the consequences of not following procedure, and it is
another matter of making certain that they are not going to be lazy
about following those directions. For teams of more than a few people,
custom created pipeline tools can help the artist make certain that
they are not making mistakes.
*On a side note:* I think that Never Alone
<http://neveralonegame.com/> is a beautiful game on many levels.
*Congratulations Siew Yi Lang!* I'm very much looking forward to
playing it on Steam <http://store.steampowered.com/app/295790/> once I
clear off my plate a bit. Any large production is quite an ordeal, but
I hope that you can feel proud to have worked on it.
Cheers,
-=Eric
--
-=T=-
