4090 Redshift

[jamination]

I am currently using a 3090 with Redshift, can anyone say if the 4090 works as well. No answers to the question on the Redshift forums sadly.

[avalon_70]

I just saw the question now, I usually just frequent the Cycles page.
Quick answer: No.
The latest version for Softimage is 3.0.45 which supports only the 3000 series.
You're already lucky to get the 3000 series going, unfortunately with my license I'm stuck in the 1 series, because it's not possible to buy the old license!

[jamination]

It does not appear to be true. I own a 4070 and it runs fine with Redshift and Softimage. I would not have tried had they said on the Redshift3d forum there was no reason it would not work.

[avalon_70]

It does not appear to be true. I own a 4070 and it runs fine with Redshift and Softimage. I would not have tried had they said on the Redshift3d forum there was no reason it would not work.

This is great news! and also unexpected.. 😱
What Redshift version do you have?

[avalon_70]

It does not appear to be true. I own a 4070 and it runs fine with Redshift and Softimage. I would not have tried had they said on the Redshift3d forum there was no reason it would not work.

Of course we're weird... we're already 4 cats, if we don't even answer each other... well...

[jamination]

I am using 3.028. I rolled back a bit from the final realease (3.045?) because I was seeing bad artifacts in renders, also it would not render .pic files which is not a huge deal but...If I forgot to change it to .png then the render would not output anything.

[FXDude]

It's not very likely that older versions of Redshift would stop working on later/newer GPU's,
or not that much more likely than MentalRay stopping to work on newer CPU's.

( but I wonder ... can we still buy older Redshift lisences ??? )

Otherwise by the way, I also have a working raytraced "pseudo-GI" in the works for Mental Ray
(using a very wide colored AO bounces in the architectural shader )

Making "fast GI", ( or fast enough ) and good looking GI,including in interior contexts !
( ~20 min for the Classroom in 1080p )
___________

Because a very "big reason", if not the ONLY reason why something like Arnold came about,
was the basic inability for reliable light transport using mental ray or even Vray at the time,
in basically ANY circumstance where animation would be involved.

Otherwise either "global illumination" OR "FinalGathering" were even completely unusable in animation UNLESS using a fixed map,
or UNLESS using FG in"precise mode" , but that was ALSO unusable, only because you could let it sit there rendering almost forever,
and there would STILL be just WAY too much noise.

So "Fast enough raytraced GI" was arguably the whole point of Arnold,
maybe also for the whole "ASS File" render asset workflow which was MUCH more fluid, workable, and memory efficient, than using MR .mi files,
being what allowed for "billion polygon" renders.
( "Fast enough" meaning "not "infinitely long" renders,
yet Arnold can also take ALOT of time to get "clean enough" renders, also because it's "unbiased".
specially with stuff like Hair, and specially of course in interior lighting contexts. )

____________________

Also otherwise, I find that Redshift sort of "nailed it", in terms of balance between speed and "biased-ness"

Because it's not only because a renderer is GPU based that it's automatically "fast",
maybe like Arnold GPU can be faster for previews with lower samples,
but getting "clean final renders" can take just about as much time as using CPU, except that using CPU can have WAY less issues overall.

AND also-also otherwise in XSI, of course there is also Sycles !
which I personally have yet to try newer versions which came with lots of stuff lol

( sooo many things to get to .. like haven't still tried to render things like strands or particles and/or volume stuff in Sycles ... uhm YET! )

By the way, old versions of something like GPU renderers, if old versions were good at things like GI, or maybe a few other things,
then any other "new features" can mostly be "fluff", or "accessory" to what's may be "the main important things".

But ANYHOO .. let us know if encountering other "roadblocks" with either newer versions of redshift on newer GPUs,

and Cheers!

[jamination]

Before 3.017 (or around there) the Nvidia 30000 and 4000 series will not work for sure. I believe it is noted in the release notes.

[FXDude]

Thanks for that info, I'll probably do some testing for the latest working version around that version number.
Otherwise for me I don't even know which version I have installed
( from more than 4 years ago lol ... while still working even very-very well on a 2070 RTX ! )

[jamination]

No, to answer you question you can not buy "old" licenses of Redshift. The license server has changed to a new Maxon version that does not support any Softimage version.

[FXDude]

Stupîd man-made (artificial) crisisees.

( I'd say also qualifying the ENTIRE Softimage situation )

is this "progress" ? or "evolution" ???

Because we arguably have been "de-evolved".

Stephan Haidacher, Babette Khan, Tim Borgman, Vitaly Bulgarov, ... and ALOT of VERY good artists ....
... what are they up to today ?
they're not dead lol ... but most of their Vimeo pages went silent (no new material) post 2020
otherwise FILLED with absolutely AMAZING XSI stuff,

Some have some new Houdini material ... but you know ... Houdini is no XSI,
or you need Maya AND Houdini to do XSI stuff .... which can easily be like just "too much"...

... specially that Maya can be rather pretty quirky / awkward / crashy as heck,
and Houdini despite all it's power, is probably one of the most user un-friendly or un-intuitive pieces of software ever devised,
like they aren't even trying.
Otherwise, it can be an almost endless endeavor to start animating characters in Houdini,
or to quickly model stuff without even caring about node trees.

( despite EVERYTHING being possible in Houdini )

Yet I wonder how many people basically just switched fields entirely (or became specialized in one thing)
because of this whole debacle. ( I'm also maybe one of them lol )
____________________

I personally consider that we have like NO DEFENSES against the whims, or desires ...
or that we are at the mercy of whatever "market leaders" may deem "best" for "the rest of us".
( same with the whole "rental only movement" , & maybe same with python 2 )

Like why not retire C++,
would be a total disaster because of a totally arbitrary decision.

And most of us would go
<< oh .. guess theres no C++ anymore ... oh well !! time to move on ! >>
( as if it was due to "natural disasters" )

But what is there to do ... send letters ?

Sometimes it feels like we are little lab rats in a cage,
hoping that lab experimentalists won't slowly kill us all,
( figuratively of course lol )
or to suddenly cut-off water supply on a whim
( because of reasons )