Maxwell 2.5 is out. General opinion of the past years was more or less:
forget it, too slow.
Yet, at closer look, I must say: don't believe the hype! This is a
fantastic renderer!
Interior scenes: no problem.
Multilight: Tweak intensity and brightness of light emitters after the
rendering is finished. Cool!!!
Resume render. Who else can do this?
Simple setup! More or less the only parameter is Sampling level.
It's not only about rendertimes, but also setup time.
Had a lot of "fun" already tweaking abstract settings in mental ray or
vray to squeeze out a decent image. 100 testrenderings until you are
ready to go? Especially with stills, you might be better off simply
letting Maxwell render a little longer and worry about something more
interesting like the picture itself.
So far, so good... but
SI's Render Region is not supported.
A-ha, there's "Maxwell Fire":
http://www.maxwellrender.com/fire/index.php
Wow, exactly what we need!!
Realtime preview, perfect renderings... how long have we been waiting
for this??
BUT: the Softimage plugin does not support Fire either!
Why??? max and Maya's do, why not Softimage?
Here's what Mihnea Balta in the Maxwell Forum says to this:
" From a workflow point of view, integrating FIRE makes sense when
there's a
built-in material editor. In Softimage we use referenced materials because
the UI toolkit is too limited to support the complex structure of the
Maxwell material. In this case, since materials are edited with MXED, we
can't update the FIRE window while you're tweaking a material. You would
have to click edit, make a change and close MXED to see the change in
FIRE, and that's hardly interactive.
We could probably build a material editor and embed it in Softimage using
the platform UI API (Windows API, basically, since we don't support the
plug-in on Linux anyway), but it would be quite a bit of work. However,
we've taken a look at the event system in Softimage, and it's not good
enough to support an interactive renderer. It's not even possible to be
notified in a reliable way when an object gets deleted.
Given all this, we've decided we won't support FIRE in Softimage.
...
Arnold's materials are shading networks composed of nodes with relatively
simple attributes and UI, so that's easy to integrate in Softimage. Since
they are made of XSI nodes, it's possible to track changes to them and
update the preview render. I'm still curious how they handle geometry
modifications and object removal, since those are not shown in the video
(my guess is that they don't). We had a prototype FIRE implementation in
XSI, but we dropped it after we saw how many things are broken, badly
designed or don't exist at all in the relevant APIs. The only thing which
worked more or less was moving the camera around, but even that's
difficult to do if you group the camera with something and move the group.
Maxwell has a single, monolithic material with lots of functionality, so
it requires a more complex UI. We could split it into nodes (actually,
that's how it's implemented in Maya and Max behind the scenes, but with a
custom UI on top to hide it), but editing a material would be considerably
more difficult than it is in MXED. For example, in order to add a layer,
you would have to create a layer node, create a BSDF node, and connect
them to the correct slots. Moving a layer from the bottom of the stack to
the top is a simple drag'n'drop operation in MXED, but would mean
disconnecting all the layers and reconnecting them in the correct order in
a network-based implementation. I think we would end up with something too
clumsy to be useful. Skipping the Softimage SDK entirely and building our
own editor window is most likely possible, but as I said, it takes some
work, and I don't know when/if we will have time to do it.
"
Anything official to say to raise our spirits?
Best,
Eugen
I think they just did not try hard enough. I don't remember receiving questions from them since XSI 7. And remember that XSI 6 was the first version with a real render API. It had it's limitation for sure.
They are right in saying that our UI API is missing some bits, but still they can work around it by doing a native window, use PyQT or create a custom display host.
Guillaume Laferriere
There are more render options than I thought:
mental ray (of course. iray upcoming??)
Arnold (beta, fast)
Maxwell
VRay (beta, slow)
3Delight
Indigo
Fryrender
Arion
Octane
LuxRender
TheaRender (interesting! You wrote the plugin, Steven?)
finalRender (beta seems to be stalled.)
The more, the merrier, but only with Render Region support and interactive
preview things get really 21st century, second decade!
Softimage should open up the render API doors as wide as possible for 3rd
party developers, and communicate this ambition.
Looks like there's some work to do, see the Arnold hotfix.
If Softimage and Nextlimit are just waiting for the next move of the other
side, noone wins!
Cooperation??
Am 19.01.2011, 20:13 Uhr, schrieb Steven Caron <car...@gmail.com>:
> "...if *thousands* of angry softimage customers..."
Best regards,
Vlado
--
-------------------------------------------
Stefan Kubicek Co-founder
-------------------------------------------
keyvis digital imagery
1050 Vienna Wehrgasse 9 Austria
Phone: +43/699/12614231
--- www.keyvis.at ste...@keyvis.at ---
-- This email and its attachments are
--confidential and for the recipient only--
> So what's holding you up from making this fantastic renderer available
> for this fantastic host app?
Nothing much at this point, I guess. There are some little bits and
pieces that are still missing, but they will be done soon.
Best regards,
Vlado
With your effort you will definitely attract a good chunk of the archvis
crowd (which is huge) away from max to Softimage.
I know the resentments from Softimage officials about this chapter, though
I don't really understand them.
How does "box business" (archvis) interfere in any way with the ambitions
towards high-end simulation/effects/character anim?
Actually, everything you really need for archvis can be useful for
projects more on the "top-notch" side:
good modelling features (check) and a modern renderer (mental ray: no
check).
When Vray for SI is out, for me as an allrounder there won't be one reason
left to touch 3ds max for anything except data exchange...
Best,
Eugen
-----Original Message-----
From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Eugen Sares
Am 20.01.2011, 16:20 Uhr, schrieb Marc-Andre Carbonneau
<marc-andre...@ubisoft.com>:
Speaking only for the rendering part of the SDK (as it was the initial subject of this thread), I really don't think that there is a major issue now.
My only experience in this area was when I was an active beta tester of SitoA, the Arnold plugin for Softimage. I also contributed in the development of things like the support for the XSINormal shader, the Standins, etc...
The Rendering SDK was looking good to me. And before that, a plugin was developed for a TV serie (Pokoyo) using the old Softimage API (I can imagine the pain at this time) ! So I guess, the issue is not 100% completely in the Softimage SDK camp.
I think that Solid Angle developers (the Arnold company) are really working hard to integrate as many feature as they can in their Softimage plugin. It is not a little task to implement every rendering feature of a 3D application. On top of that, the 3D application always evolve and can add more features to implement in the rendering plugin (like support for rendering Strands when ICE was released...).
frenchViz
From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Gene Crucean
Sent: Thursday, January 20, 2011 2:31 PM
To: soft...@listproc.autodesk.com
Subject: Re: Maxwell and Renderers in general...
Hahaha.
On a serious note. Why has the SDK been such an issue for so long? There has to be logic behind it right?
:)
:)
Guillaume
<marc-andre...@ubisoft.com<mailto:marc-andre...@ubisoft.com>>:
> <vl...@chaosgroup.com<mailto:vl...@chaosgroup.com>>:
>>> <vl...@chaosgroup.com<mailto:vl...@chaosgroup.com>>:
Tried hair instancing, but even then the hair is generated at rendertime,
which of course does not work either.
Is it possible to make an (animated) geometry "snapshot" of hair, to bring
it over to Maxwell brute-force?