PDA

View Full Version : LW 11.5 / FPrime?



Builtdown
10-29-2012, 06:44 AM
Hi,

Ive been using LW 9.6 + FPrime for years and have enjoyed their co-operation.
Now Im planning to upgrade to 11.5 but I have heard that FPrime dont work with it. Is this true?

If it doesnt, then I would like to know how fast LWs Viper render Monte Carlo radiosity compared to FPrime?

If FPrime does wok with 11.5, does it work with instancing?


Thanks,


- Builtdown

erikals
10-29-2012, 07:32 AM
FPrime is dated, nope, doesn't work with instancing.

more info here > http://forums.newtek.com/showthread.php?123895-Worley-Labs

50one
10-29-2012, 07:35 AM
Hi,

Ive been using LW 9.6 + FPrime for years and have enjoyed their co-operation.
Now Im planning to upgrade to 11.5 but I have heard that FPrime dont work with it. Is this true?

If it doesnt, then I would like to know how fast LWs Viper render Monte Carlo radiosity compared to FPrime?

If FPrime does wok with 11.5, does it work with instancing?


Thanks,


- Builtdown

Hey man,

I'll keep it short. 11.5 is not released yet, Fprime is dead, it will probably work with 11.5[most of the shading nodes are not working anyway since 10 I think] What you need the Fprime for anyway these days? If LW3D group will decide to upgrade the VPR window to be un-dockable and with similar options to save as Fprime - than I would say it will 10X better than Fprime was[at least most of the shading nodes is working and pixel filters I suppose]

erikals
10-29-2012, 07:46 AM
doesn't seem to work with 11...
http://forums.newtek.com/showthread.php?123895-Worley-Labs/page9

jasonwestmas
10-29-2012, 08:10 AM
If you're a worley guy just stick with 9.6 for rendering and all will be well. But if you want more features you'll have to ditch the worley stuff.

Pavlov
10-29-2012, 08:34 AM
let's stop this VPR - better -than - Fprime thing. This is simply not true. If uou render with cache irradiance (interpolated) VPR is very nice, but when you render UNBIASED Fprime still whips the floor with VPR. And coincidentially, these days unbiased is becoming the standard for many reasons - i.e. no fancy cache tricks for animation and so on.
So there are good reasons Fprime should be upgraded and made instance compatible. Otherwise let's hope VPR (and the whole LW engine) gets a huge boost in unbiased rendering.

Paolo

jasonwestmas
10-29-2012, 08:42 AM
Yes, in a LW context I would still use Fprime for animated GI.

geo_n
10-29-2012, 08:43 AM
And fprime is still 100% easy to setup flicker free renderer. The bruteforce renderer is still fast in todays standard. If I would chose between up to date fprime or gpu octane, I would choose fprime any day.

erikals
10-29-2012, 09:46 AM
well, just to inform newbies, Fprime could'nt get the Octane quality by far though.
but for flicker-free renders, it's quite welcome.

LW11 should be alright too from what i've heard, (haven't used it much) as far as render speed goes...
always use post filters to reduce flicker btw...

but note, even a bigger problem in LW are GI splotches, not GI flicker.
(as flicker quite often can get heavily reduced in post using e.g. Neat video)

quote from Rob... (October 18, 2012)
"...areas I am not completely happy with. Part of that is that we are working on updates to our rendering engine – and how it handles data – so that those formats can be used better..."

http://www.fxguide.com/featured/rob-powers-the-guy-at-the-wheel-of-the-new-newtek-lightwave/

Pavlov
10-29-2012, 09:52 AM
erikals - splotches are caused by biased rendering (interpolation). Unbiased gives noise as only artifact.
In next few years we'll see interpolated GI disappear, since machines and engines can render unbiased GI in acceptable times now, and GPU is a huge boost in this direction.
On this aspect i feel LW engine is quite behind the lines, while Fprime could compete with any CPU unbiased engine even today. That's why i'd like Workey to put his hands on it again.

Paolo

erikals
10-29-2012, 10:36 AM
yep..

Builtdown
10-29-2012, 11:21 AM
Thanks for the comments.

So basically I would be disappointed with LW11 renderer compared to FPrime. Maybe I will keep working with 9.6. I have never had flickering or splotches with FPrime.

By the way, one main thing I love with FPrime is the possibility to select the render quality level when doing draft animations. Is this possible with native LW11 renderer?

dwburman
10-29-2012, 12:30 PM
No, the native renderer does not have the same level of controls since it is not an iterative renderer like FPRime.

You can set VPR to draft mode and set it to half resolution and just do a preview render, but you might be better off lowering your camera and AA settings and just doing a render.

What you use really depends on your needs.



If you want a previewer that works with everything in the surface editor but the shader nodes that require preprocessing (mostly SSS nodes) including everything on the shader tab use VPR.

If you want something to preview Hypervoxels and volumetric lights and works with the new native instancing use VPR.

If you want a quicker, iterative, unbiased renderer use FPrime.

If you want to accurately render/preview alternate camera types (advanced camera, liberty3d ubercam cameras) use FPrime (VPR is a viewport renderer and only renders things the way they are displayed in the viewport).

If you want to use the built-in linear workflow tools, so you don't have to set everything up manually, use VPR and the native renderer.

Generally, VPR is a better previewer except for the lake of a floating window and zoom/pan in the previewer. FPrime's rendering workflow has not been duplicated, so if you're used to rendering in FPRime, then you might want to stick with it.


My guess would be that Fprime would work in LW11 as long as you steer clear of any of the new features/shaders while using it, but I don't have the last version of FPrime to test and I didn't read the threads posted above.


Ultimately, I'd suggest downloading LW11 and trying it out for yourself. You can register for a 30-day trial license that gives you complete access to the program (after which it reverts to the very limited Discovery Mode). You won't be able to test FPrime in it unless FPrime looks for your dongle by itself instead of asking LW which dongle it's attached to. If Fprime does it's own query, you could start LW11 without the dongle then plug the dongle in before launching Fprime.

imshadi
02-26-2013, 06:02 PM
My experience is that LW11 does indeed produce faster results than FPrime using Brute Force MonteCarlo. All I had to do was to reduce the number of rays and let the antialias passes do the job of killing the noise. At our studio we found that we could produce brute force MonteCarlo only with Lightwave. We already dropped the idea of doing it on Mental Ray or VRay because we could not get the speeds we need. I stlll think Lightwave's shaders could evolve just a little to match some of the cooler stuff VRay can do like with plants. But generally I have returned to Lightwave for final renders and feel very satisfied. I can even use full 3D motion blur with MonteCarlo (with MR or VRay that is unheard of at least with our resources.)

I really would hope though that Newtek developed an iterative renderer like Fprime, because it was just too convenient, fast, and the fact that you can refine a render's quality is a time/resource saver that makes all the sense in the world.

I have tried to use fprime on LW11, but my problem is that G2 does not seem to work, and I miss not only the SSS on G2, but the ability to display Color Corrected images on Fprime and have linear workflow previews. (If you want to know the settings I posted them here http://farm8.staticflickr.com/7051/7062869797_b04775fc59_b.jpg)

erikals
02-26-2013, 06:21 PM
...by Brute Force MonteCarlo you mean without Interpolated, right?

jasonwestmas
02-26-2013, 06:31 PM
Hmm, I guess I'd have to do my own tests but I think brute force MC in LW is still pretty darn slow compared to other render engines. Good news if it is in fact faster when comparing quality but I doubt it.

imshadi
02-26-2013, 06:46 PM
Absolutely.

imshadi
02-26-2013, 07:32 PM
I use 10 rays for the direct rays. And one to two bounces. I set the motion blur to dithered (photoreal sometimes doesn't work) set the thing to a wild 32 Motion Blur passes each of just one sample (each motion blur pass is also a geometry/shading sample), no adaptive. My render times at 720p have been from 30min to 1 hour per frame which is great compared to 2 - 3 hours per frame on simmilar VRay scenes that do not use actual GI but occlusion shader tricks, and absolutely no motion blur there of any kind. I was astounded to realize how fast the LW native renderer had become. There is still a lot more to try, and my scenes have up to now been mostly exteriors, which is a special and convenient case, but still these times are unmistakably amazing. Here's a sample of the scenes that I'm working on.

111979

I have to add that I had a lot of difficulty at first dealing with Lightwave's simpler scheme of controlling sampling. I felt like I had lost control somewhat, and I'm still not sure whether I can obtain great render times in all cases given some difficulties I've had with scenes dealing with lots of reflections, crazy instancing and many lights, but I am becoming positive that the renderer is now faster than ever and definitely It has resurfaced as a realistic option in my professional work. Let's see if it will stay that way.

erikals
02-26-2013, 07:53 PM
nice, that's pretty fast. AA looks very good too.

wonder, for lighting,
did you use a HDRI or Sibl > http://www.hdrlabs.com/sibl/how_it_works.html
or did you use DPont SunSky... ?

ivanze
02-26-2013, 08:37 PM
imshadi, Que imagenes tan buenas!!!

Nice renders!!

jwiede
02-26-2013, 08:52 PM
I have tried to use fprime on LW11, but my problem is that G2 does not seem to work, and I miss not only the SSS on G2, but the ability to display Color Corrected images on Fprime and have linear workflow previews. (If you want to know the settings I posted them here http://farm8.staticflickr.com/7051/7062869797_b04775fc59_b.jpg)
Yep, I miss the G2(SSS)+fPrime combo quite a bit. Regardless of whether brute-force MC on LW is now faster by itself, I do not find brute-force MC _and_ SSS to be faster than what G2's SSS + fPrime offered. Further, because of fPrime's iterative nature, it offered options that LW's brute-force MC simply does not, such as the ability to work (for comp, etc.) with partially-converged solutions while finishing the convergence in parallel. With LW's MC it's an all or nothing proposition in terms of render cost, no "splitting the difference" like fPrime offered. Alas, fPrime's days appear to be over.

I do find it interesting that you're finding LW's brute force MC to be faster than Vray's. What version of the Vray core did you test against, and what kinds of surface materials were used in the testing? Any materials with significant refr/refl, blurry refl/refr, or SSS?

Pavlov
02-27-2013, 01:57 AM
yes, i confirm, we do the same.
For exteriors we're using MC, non interpolated, 2/3 bounces, 3-4 rays, and a good amount of AA. It's quite fast, and VPR is a bolt (yes, VPR is much faster with no interpolation and few rays, than interpolated). For exteriors it's a dream, for interiors not as much. I hope NT works a little on unbiased GI in next update.

Paolo

lwanmtr
02-27-2013, 05:34 PM
Nice, I'll have to try those settings on my own scenes....lol..funny the info you can get by randomly veiwing a thread.

Hail
02-28-2013, 03:40 AM
yes, i confirm, we do the same.
For exteriors we're using MC, non interpolated, 2/3 bounces, 3-4 rays, and a good amount of AA. It's quite fast, and VPR is a bolt (yes, VPR is much faster with no interpolation and few rays, than interpolated). For exteriors it's a dream, for interiors not as much. I hope NT works a little on unbiased GI in next update.

Paolo

Interesting.. but the question is how come its faster than Vray?
Perhaps could it be due to lack of proper optimisation in the Vray setup?

Pavlov
02-28-2013, 10:22 AM
where did i wrote it's faster than Vray ?

Paolo

jasonwestmas
02-28-2013, 11:40 AM
I think Pavlov meant that with fewer rays he can get a faster render in VPR within LW-Brute Force than he can with interpolated turned on. . . with the same rays I suppose.

erikals
02-28-2013, 11:56 AM
ah, didn't quite get that at first...

ivanze
02-28-2013, 04:01 PM
imshadi was the one who said that was faster than Vray, not Pavlov.

Pavlov
02-28-2013, 04:26 PM
exactly. using 2-3 rays and a good AA (i.e. 8-20, AS 0.05, keeping base samples at 1) you see what's happening in your scene much faster (and better) than ANY interpolated setting. The difference becomes huge in case of heavy instancing.
When you have a lot of bounced/indirect light in your scene, things start to change and at a given point interpolated is better.

Paolo

Hail
03-01-2013, 03:20 AM
where did i wrote it's faster than Vray ?

Paolo

Ooops! wrong quote, sorry dude, I meant to quote imshadi.:)

lino.grandi
03-01-2013, 06:58 AM
My experience is that LW11 does indeed produce faster results than FPrime using Brute Force MonteCarlo.

That's my experience as well.

Taran-Q
04-24-2013, 06:55 AM
I am just very stuck to the method of selecting my materials in the fprime window :), besides that it's still a very quick way to just see how the scene globally will look like. Final renders are done with Kray :D

Pavlov
04-24-2013, 07:05 AM
If bounces are few, this is true. In exterior uncached MC is OK. In interiors, Fprime probably had better refining algorythms, since it cleaned faster than LW's MC.
Paolo