Announcing Redshift - Biased GPU Renderer

[ActionArt]

I can second that. It is coming along VERY NICELY. This will be THE render engine to get when it comes out, no doubt. I'm so glad I got on the test team, it has been a great experience.

[Hirazi Blue]

Coming along nicely enough to leave its alpha state soon? I seem to recall someone commenting that calling it an alpha was somewhat false modesty anyway.
;)

[ActionArt]

In some ways yes. I think it's because they're still adding features which technically still makes it alpha. There's a bit to go yet to get all the great stuff they have planned in the first release. Lots of work but going quickly.

Makes me realize just how pathetic the mentalray/AD integration is. It'll be 2050 before we get the current mr feature set integrated and by then nobody will care.

Thank the lucky stars for Redshift! It'll save us.

[Cacoman]

Well as Action Art said, I believe it will come to beta after all the features supported by other render engines are there.

Although Redshift is lighting bolt fast compared to other renders to achieve the quality Redshift is delivering, they are still improving speed and attending our requests. Don't know how many hours a day this guys work, but it's insane.

I've seen some new (but old ;) ), faces in the group.

Maybe if you write them...

[yashugan]

I also would like try it, but i suppose that if they decided a close number of tester, there is a reason. For shure, any way, i'll bought it. XSI need really something FASTER than the superslow mental ray.

I have only to wait. but looking other render time of develope i really ahve few hope to see it soon.
Any way my best wish!!!! Keep strong!

[scaron]

are there any alpha/beta testers of redshift here willing to answer these questions about the plugin integration with softimage?

they are almost all related IPR, or how interactive the scene updates from changes done by the user...
-if you have a large scene with many objects and you do a modeling change on one object, does the entire scene need to be reprocessed/exported?
-is instancing supported? through ICE?
-what happens when you do modeling changes to an object which is instanced?
-is there a fly through mode? ie. do you have to re export the scene every time the frame changes even if it is static?
-what happens if you import an softimage model from disk? does the entire scene need to be reprocessed/exported?

thanks!

[doca]

I cant answer to all your questions but here is some info:
-I think that there is no whole scene export if you change something. I am working on huge scene, about 15 million unique polygons, with animated trees (leafs are animated, ice instanced, so there is only 10-20 unique polygons for leafs), tons of instance models (none of them goes to xsi scene polycount), animated characters... I believe there is a few hundred millions of rendered polys, maybe even billion, all leafs with translucency... First frame is exported in a less than minute, all others take about 20-30 sec, and for every frame it takes new cache. Every frame has motion blur and dof enabled. Render time is about 1.5-4 minutes, without gi, about 2-6 with gi (brute force gi). It is rendered with GTX titan, but, I must say that render time is exactly the same on GTX 780, so thre is no need for titan. For those scenes memory consumption is about 1.3-1.5 gigabytes (again, no need for titan).
Hope, this helped.

[Draise]

Impressive results!

[scaron]

I cant answer to all your questions but here is some info:
-I think that there is no whole scene export if you change something. I am working on huge scene, about 15 million unique polygons, with animated trees (leafs are animated, ice instanced, so there is only 10-20 unique polygons for leafs), tons of instance models (none of them goes to xsi scene polycount), animated characters... I believe there is a few hundred millions of rendered polys, maybe even billion, all leafs with translucency... First frame is exported in a less than minute, all others take about 20-30 sec, and for every frame it takes new cache. Every frame has motion blur and dof enabled. Render time is about 1.5-4 minutes, without gi, about 2-6 with gi (brute force gi). It is rendered with GTX titan, but, I must say that render time is exactly the same on GTX 780, so thre is no need for titan. For those scenes memory consumption is about 1.3-1.5 gigabytes (again, no need for titan).
Hope, this helped.

thanks for your response, but its too vague. if you start a render with the region (interactive preview) and you select and object and start doing modeling changes. does it take the same amount of time to start rendering as it did initially? or is it less? not interested in motion blur or depth of field. this is more a question about how tightly its integrated into softimage and how clever it is on interactive changes. i didn't want to bother the redshift guys directly, but i might have to message nicolas anyways.

thanks
s

[Draise]

I've had some very immediate results with the Render Region preview. I found it almost more responsive than Mental Ray (obviuolsy because it's faster ) I found it very responsive. The only bug in that one was to make sure the render preview was set to "render current pass" when using multiple render engines and passes, because switching engines causes an SI error (not the render engine).
Also, node and shader intergration is almost seamless, with some added custom shader nodes to work with (but not too many). Most of what's there can be used without problems, from what I've tested.

I'm tempted to just buy some good CUDA cards and change the studio's workflow to Redshift! Indeed "Alpha" version is just being modest.... though I haven't tested it heavily just yet.

[Cacoman]

Well the integration with Softimage is amazing. Imagine MR but Redshift.

The changes in the interactive preview are submitted at once. Sometimes I turn to look at the render preview and it's already done.

They have a progressive rendering mode. So if you turn this on, it works like the Fire from Maxwell but a lot faster.

For me Redshift once it has covered every aspect of Softimage is what MR should have be from a long, long time.

After modeling and animating and just before the lighting and shading process, I took a deep breath...

Redshift au contraire, is not only light speed but fun also!

And believe me when I make this statements. I've been working with Softimage since Softimage 3D 3.5

Cheers!!!!

[Falam]

Is a Geforce Titan really beneficial over a 6xx/7xx series graphics cards, in terms of 3D and/or GPU rendering ? My conclusion is no.

[Cacoman]

Well over the 6xx series the Titan renders faster.

Recent tests with the 780 I think, are rendering at the same speed of the Titan, according to one user.

[Falam]

Titan and the 7xx series are almost identical in speed, you could go with one or the other, while the 6xx is next one down.

[Cacoman]

In fact 5xx series are rendering faster than the 6xx

[Falam]

Which card is best overall and which is strictly a arms race between two companies ?