PDA

View Full Version : LightWave v9 Status Update posted...



Chuck
04-21-2006, 09:40 AM
Since no one has looked at it and it has been up a while now, maybe we better link to it! :)

>>> LightWave v9 Status Report, 4/21/06 <<< (http://www.newtek.com/forums/showthread.php?t=49791)

ufo3d
04-21-2006, 09:53 AM
We are getting close to our release date, and I know that many of you have a lot of questions, so let’s get right to it.
the status report is sweet, but, how many time we hear these words "getting close" , "soon"? it seems we might not see final version until siggraph , q3 of 2006?


LightWave v9.0 is now in code freeze, and that will be evident with the next Open Beta build, which the Open Beta group will receive early next week. With that build all of the features that we plan to ship in 9.0 will be active


New nodes give you the power of sub-surface scattering and other shading models, another unannounced feature.

the modelling tools should be in next build? we have been waiting for the video of different shading models and modelling tools in layout for very long long time............:devil:

Tiger
04-21-2006, 10:59 AM
I`m so happy I bought Lightwave!
Thank you Newtek :thumbsup:

Titus
04-21-2006, 11:12 AM
If the speed improvement is real (10 - 20 times), then is just worth the upgrade.

colkai
04-21-2006, 11:25 AM
I'd be inclined to take the speed improvment comment carefully.

Rendering speeds have improved by as much as 20 times faster

That doesn't mean every scene will render that fast, nor indeed is that a set in stone figure. Kinda like saying your car does 51mpg, but in traffic, in urban areas, you're more likely to get 33mpg.

mav3rick
04-21-2006, 11:25 AM
as i said previous for lw8.5 i dont care if i need to wait till siggraph i just want 9 to be ready and well rounded not just for us but for dev team to complete code and not left messy code here and there.

lino.grandi
04-21-2006, 11:42 AM
I'd be inclined to take the speed improvment comment carefully.


That doesn't mean every scene will render that fast, nor indeed is that a set in stone figure. Kinda like saying your car does 51mpg, but in traffic, in urban areas, you're more likely to get 33mpg.

We're talking about opengl performance, not render times.

Titus
04-21-2006, 11:57 AM
Rendering speeds have improved by as much as 20 times faster, and the more you throw at the rendering engine, the faster it gets.

Anyway, this sounds like marketing speech, something you need to take with a grain of salt.

Yog
04-21-2006, 12:39 PM
Anyway, this sounds like marketing speech, something you need to take with a grain of salt.Possibly not.
Memory management and memory flushing can take a considerable time on larger scenes. If you can improve memory flushing speeds by 20% it will make a big difference if it takes a minute or more on huge scenes, but if it currently only takes 0.5 secs on very simple scenes you will hardly notice a 20% increase in flushing speeds.

colkai
04-21-2006, 01:05 PM
We're talking about opengl performance, not render times.
where? last I checked, we were commenting on Render speeds, or did I miss summat? :stumped:

metahumanity
04-21-2006, 01:32 PM
That renderspeed increase almost seems to good to be true.

10-20x ??? Can it be?

Chuck
04-21-2006, 01:51 PM
That renderspeed increase almost seems to good to be true.

10-20x ??? Can it be?

This does need to be kept in perspective, in particular that the speed increase is specific to raytracing. The renderer's raytracing core is now highly optimized for better performance as the scale of the scene increases. In very small scenes you can even be a bit slower, but most scenes will render faster to some degree. As noted in the feature list the average would be about twice as fast - and an average means you are in the middle of a range. The Open Beta users have tested a variety of scene types and sizes, and identified what the expectations should be for quite a number of different use cases. I think the 20X figure was the record for speed increase between 8.5 and v9 among the scenes they've tested so far.

metahumanity
04-21-2006, 01:55 PM
I highly doubt it! Unless your computer gets hit by lightning the split second you hit the render button.:D


LMAO

Titus
04-21-2006, 01:57 PM
Possibly not.
Memory management and memory flushing can take a considerable time on larger scenes.

That makes sense, but also this is too good to be true. I can expect more modest figures in terms of real speed and performance.

I think LW is a fast app, it was only a matter to use new rendering algorithms developed in the last years to make it fast as a lightwave :).

metahumanity
04-21-2006, 01:57 PM
This does need to be kept in perspective, in particular that the speed increase is specific to raytracing. The renderer's raytracing core is now highly optimized for better performance as the scale of the scene increases. In very small scenes you can even be a bit slower, but most scenes will render faster to some degree. As noted in the feature list the average would be about twice as fast - and an average means you are in the middle of a range. The Open Beta users have tested a variety of scene types and sizes, and identified what the expectations should be for quite a number of different use cases. I think the 20X figure was the record for speed increase between 8.5 and v9 among the scenes they've tested so far.

Thx for clarifying, Chuck.

So, from what i understand, tree heavy scenes will benefit a great deal. Cool.

What about radiosity and GI? Any significant increases?

private
04-22-2006, 03:21 AM
I am not doing the open beta. Has screamernet been touched? If not, why not?

Matt
04-22-2006, 02:31 PM
Honest, to the point, open, optimistic ... love it NewTek!

:)

Captain Obvious
04-22-2006, 02:54 PM
The reason the new renderer is so much faster than the old one is probably that it's a raytracer instead of a hybrid scanline/raytracer. As far as I know, raytracers have much better complexity scaling than scanliners. As an example: take a simple metanurbs object in Layout, and increase the subdivision rate. This makes the OpenGL much slower fairly rapidly. But if you render it in a raytracer, like Fprime, you'd barely notice a difference from 1000 polygons to 100,000 polygons. But this is just me guessing, so don't quote me on it! ;)




What about radiosity and GI? Any significant increases?
The radiosity in Lightwave is raytraced. Any optimizations made to raytracing will speed up the radiosity by the same amount.

joao
04-22-2006, 03:20 PM
its always nice to see big improvements in rendering for polygon heavy scenes.... but unless opengl in modeler is improved i don't see how i can create these heavy scenes in the first place - rendering and opengl display have to be improved side by side.....
just my thoughts
joao

Captain Obvious
04-22-2006, 03:28 PM
I don't know about the rest of you, but I always build my scenes one object at a time, and position them in Layout.

stevecullum
04-22-2006, 06:52 PM
Has screamernet been touched? If not, why not?

I can only assume it must have been, otherwise, how would it render the new nodes and other features?

Chuck
04-22-2006, 09:59 PM
Open Beta folks, please remember the NDA is still in force. We've removed the messages referrring to a thread in the private Open Beta forums.

Nemoid
04-23-2006, 01:11 AM
its a good reading and i think Lw is going in the right direction, and 9,0 will show increased performance.
From what i read it will be a huge step ahead and a good starting point as the base of a future great development of the app , this time, in terms of performance, and some of the toolset is great. APS for example.

I also loved the adoption of real studio config files. sometimes we criticize Lw UI and workflow so much, but configuration is also very important.
having a real production config will help alot.

This being said, what i also think is that a considerable work has to be done in the 9.x cycle, to get better CA performance, better rigging workflow and in general, a faster, more organized workflow however.

Too many similar tools in modeler needing integration between each other, more interactive tools required, integration of CAD tools/precise modelling for archiviz behaviour too.

but i think many of these points will be addressed during this cycle, and that the core will be enhanced even during this period.

So Nt keep up with the good and hard work !