Octane 1.5 features revealed!

[darkline]

Actually multipass motion blur gives the most accurate MB possible, you just need enough samples to get it smooth and without banding, this can vary depending on the speed of movement within a frame.
The thing with an unbiased renderer like Octane is that you don't have to render every pass at full quality because averaging say 10 frames together at the end will give you approximately the same result as having rendered a still for 10 times as long so effectively if you are wanting to render and average 10 frames per finished frame you can drop the max samples you would use on a still frame by a factor of 10 and still get a similar convergence on the final frame. (in reality you may need a few extra samples but it's not as bad as you may first think) I have done tests of this and it seems to work quite well.

Please see my thoughts regarding Vector MB here: http://render.otoy.com/forum/viewtopic. ... ur#p107840

T.

I have tried this but I don't agree unfortunately. I rendered a scene with 10 passes and dropped the sample rate by a factor of 10 so I was essentially rendering the same time per frame. results were far inferior. I remember Karba saying it wasn't ideal because octane samples identically each time, therefore you're wasting calculation time per pass as it essentially re-calculating the majority of information on each frame. That's wasted time. So while I would say rendering 12 passes does not = x 12 rendering times, depending on your tolerance you're still looking at a factor of 5x or more render time. At least that's my experience, noisy areas take a lot longer to clean up using multi-pass. And as you say to get a streak free image you have to up the passes even further making render time worse.

Look, I guess it will work fine for some (I have used it with some limited success myself), but my preference is RSMB in post, each to their own. RSMB will fail on complex scenes too for sure, but if you're rendering on an alpha channel it's much better...

All Im trying to say is this is not a production friendly solution, but a time consuming workaround. Each scene has to be taken on it's merits as which to use.

Octane MB will be by far the most accurate and fastest method when available (if done like in iray), nothing will give results like it and I hope it's here soon

[j7th]

MB in open source intel embree 2.0:

http://embree.github.io/
http://embree.github.io/data/embree-sig ... -final.pdf

[mbetke]

I really don't need a roadmap for my daily work. bu for personal interest it is nice to have.
I have to decide a Yes or No if a client asks me if I can do a certain job.
I can't tell him to come back some months later where I may could do his project based on a roadmap.

But for long-term clients I sometimes present them better images (like after Instances or PMC has been added) and tell them "Hey the tech I use has now better features! You like it?". If they like it I can make a bit more money out of a picture or make them happier clients which just some checkboxes turned on on my side.
So even there I don't need a roadmap.

[Proupin]

I remember Karba saying it wasn't ideal because octane samples identically each time, therefore you're wasting calculation time per pass as it essentially re-calculating the majority of information on each frame. That's wasted time. So while I would say rendering 12 passes does not = x 12 rendering times, depending on your tolerance you're still looking at a factor of 5x or more render time. At least that's my experience, noisy areas take a lot longer to clean up using multi-pass. And as you say to get a streak free image you have to up the passes even further making render time worse.

Yes, I have also observed this is as well, it should be a simple fix from Otoy's part, to randomize the 'seed' on every pass... then it would be a very reasonable way to render with Motion Blur. I hope they read this and work on a simple fix.

[Karba]

I have to decide a Yes or No if a client asks me if I can do a certain job.
I can't tell him to come back some months later where I may could do his project based on a roadmap.

You always have to decide - NO if it is NOT here right now.

[rappet]

I have to decide a Yes or No if a client asks me if I can do a certain job.
I can't tell him to come back some months later where I may could do his project based on a roadmap.

You always have to decide - NO if it is NOT here right now.

I cannot resist to quote jedi Yoda: "Decide you must, how to serve them best."

[Karba]

I cannot resist to quote jedi Yoda: "Decide you must, how to serve them best."

Average CG project takes about 2-4 weeks.
Developing of such things as MB or displacement takes 6-16 weeks.

You just have no chance of waiting for new features during the project. You definetely will fail.

[RealityFox]

Liking the new features, I didn't get to watch the whole video so I may have missed this:

When will I have to pay again for the new version of Octane? 2.0 I guess? And how much? I'm not really that interested in updating to 2.0 based on the feature list. I know people have been screaming for MB and I to would like to use Octane for animation in the future. I personally just want displacement, mb, better texture controls and multiple UV layering. I understand that those features are off into the future and I probably won't see them until 3 (or maybe even 4 sadly).

But I can't pay 120 bucks every year for Octane until I get those features. So will the cost be cheaper for those who just want to update? maybe 50 or 100? I mean I know 2.0 has some really nice feature sets but at the same time I want to just wait for my displacement and UV to happen.

First world problems :/

[FrankPooleFloating]

Um.. has someone posted a 2.0 feature list? Please show me it. Now.

[sdwhitton]

a node for offsetting bitmaps sounds very good, good way of reducing 'repeats' over a large surface, by mixing same bitmap, but one offset 50% say, using a turbulence / noise map as the mix channel

the 'millions' of emitters function looks good too

thanks guys