FPrime: can't stop posting images

[mbaldwin]

I find I'm almost compulsively dragging out my old models and gleefully tweaking them within new scenes.

Thanks Worley! just wasting time...

y'all are wellcome to post your tests as well.

[multipasser]

well this should be in gallery than.
Because you dont give any other info.
We need time in lw vs. Fprime.

And how many poly's, what render modes ect..

[CB_3D]

nice pic!

[brtk]

All right, why not. This is too an old model:
http://w1.806.comhem.se/~u80600094/LW/Bilder/Fprime/Porsche_FPtest_02.jpg
Polycount: ca677k (tesselated), rendertime: ca 30min @ A2600+, 1 arealight, 3 luminous polys, MC 4 bounces.

brtk

__________
Mr Worley, hope someone, somewhere goes nuts right now :p

[mouse_art]

You make me envious, i become my fprime not until monday or tuesday. :( ;)


btw nice renders :)

[3DBob]

Nice work guys - I think Fprime will release the stuggling artists potential from lightwave - You can just do things now that you would have avoided before - and the time penalty is not prohibitive.

3DBob

[mav3rick]

here is my little contribution to fprime scene:) man this fprime is like a drug...... it is absolute joy to work with it...... i spent lot time ... playing........... i forgot of my gf exsistance:) hell ppl are gonna loose merryage cause of fp:)



ok i stop

here is impala that come with lw i played with surfaces and some fast lightning and all prepared for render in 5-7 min.. render took me less:) heheh so 2 monte carlos and 4-6 minute render and 7 iterations and here it come results
www.null.hr/temp/impalanew.jpg

here is good old mustang scene from lw on 3 monte carlos 6 iterations 2-3minute render and motion blur
www.null.hr/temp/mustang.jpg

and some object friend riplee gave me on test,monte carlo 3 and 5 min render
www.null.hr/temp/malice2.jpg

hear ya doodlers
must go play moreeeeeeeeeeeeeeee moreeeeeeeeeeeeee moreeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

[Architook]

I love seeing everyone's test pictures, can't get enough.
But one thing is that you can't see from looking at posted pictures is how the scene feels when working with the objects live in LightWave. That FPrime interaction is why it's so fun to play with them! But I still love seeing the final output anyway, especially radiosity renders.

[Signal to Noise]

mbaldwin: Your pic just reeks of hilarity!:D

[minus]

Made this a few days ago. Just let it bake in the previewer overnight.
http://www.opnotic.com/wip/blurstager/blur_stag_fp.jpg
3 bounces, blurry floor, no "real" lights in the scene. (just luminous surfs)

[mbaldwin]

minus: looking good!

signal: thanks. throwing these models together now makes me want to do a whole series of different big heads in small rooms.

It vaguely reminds me of the huge olmec head mr. burns gave to homer simpson.

[dwburman]

I'd post my latest FPRime renderings but I'm afraid they're fairly boring. It's a hallway :o

I did put something in the WIP section a short while ago. Nothing great but I kinda like it. A city at night thing. It reminds me of a miniature set.
http://vbulletin.newtek.com/showthread.php?s=&threadid=20049

[minus]

mbaldwin
Seriously your render is very comical. I've laughed out loud now a number of times when I see it. =) It really is rediculus and a great test of the render engine all at the same time. =)

[CB_3D]

minus, this looks very good! I never get this kind of result in MC without any LW lights. Luminous polys never seem to light the scene enough for me.

[minus]

CB_3D, For a long time I was averse to Luminous polies... but through experimenting I've grown to love em. In the end actually... I didn't change their properties so much as figure out their abilities.. I don't go over 100% luminosity usually... they are great of course in reflections if you "unseen them by camera". Usually though I have to lessen the blackness by over half sometimes in the Floating Point image controls. I didn't have to do that in this above pic here just because of the sheer number of luminous polies all around the room.

Here is a zoomed out view to show more of the room. It was originally 1600X1200 and could use some more rendertime yet... not sure if I'll give it anymore though as I'm working on something else now.

http://www.opnotic.com/wip/radtest/glow_ball_blur_resized.jpg

[CB_3D]

So, if i understood correctly, youuse Luminous polies that are hidden to the camera to boost the illumination?

thx

[mbaldwin]

minus: I like the luminus effect. great work!

here's another test of mine. I know I don't have any times posted as per the render. I will say that setting up procedurals is very fast in FPrime. It's got me experimenting again with surface settings--lots of fun.

-m.

[minus]

CB_3D
I didn't unseen them by camera in the above pic.
Using unseen by camera for luminous polies is something I will do sometimes, usually if I'm just doing a render of a single object on a blank stage. If it's the only light in the scene... it's a cool effect as it shows up in reflections really nicely. Example pic:

http://www.opnotic.com/wip/Light_blur_stripe.jpg

mbaldwin That looks great... Are those water droplets or reflections that look like they could be?

[mindseye]

Here's my first "final" paying job with Fprime..... I let it run overnight to 3200x2700 pixels. Made it to 9.0 quality level by morning. 1 area light and 1 bounce. It's kinda random about the clip map issues... some scenes I have show the clipmap problems and some don't... this one didn't. This is a resized sample from the 3200 pixel file. But the full size is super clean. With Fprime I had the light and radiosity settings done in 15 minutes... I love it.

[sinthetique]

Originally posted by minus
Made this a few days ago. Just let it bake in the previewer overnight.
http://www.opnotic.com/wip/blurstager/blur_stag_fp.jpg
3 bounces, blurry floor, no "real" lights in the scene. (just luminous surfs)

Did you obtain the blurry floor with a regular Lightwave technique, or did you use a plugin. Do you mind sharing how you did it? Thanks!

[Architook]

You don't need a plugin to make blurry reflections, it's built into Lightwave. On the surface panel, go to Environment, and set Blurry Reflections higher. It works in native lightwave, and even faster in fprime.

[Architook]

There's also built in blurry transparency in LightWave, which is very cool indeed and I always wonder why I don't see more of it used. I think the blur options are just overlooked by most animators.

[Zafar Iqbal]

Aaaaaaarg.. i can't stand all these fast and cool renderings.. plz donate me some monetos so i can get me a copy of FPrime.... :cool:

[ACLOBO]

Well, ok, first of all I do love Fprime. However, I dont' seem to be having quite the impressively fast results that others are getting. Now, I know my system is by no means top of the line, but it is still a decent machine. I have a Dell p4 2 2ghz (the fist available as a matter of fact I know the 400 mhz bus is slow). I also have 768 megs of ram in this machine and it does well for all my needs at the moment. However, I have done a couple of test renders myself and just finished one with radiosity with the new 1.02 Fprime patch. Here are my examples. There are roughly 230k polys in the scene.....

First pic is with version 1.01. I am not sure about how many bounces - I think 2 or 3 and it took about 3 hours to get to this level I believe (on and off) - so you can see where I am wondering how others are getting faster and better results with radiosity. This one cooked in previewer....

The second pic is after the 1.02 update and again is Monte Carlo with 3 bounces. In each scene there are only 3 luminous polys lighting the scene (you can see their reflections on the car) Level 8 render from renderer. I think this one took between 1.5 and 2 hours (not sure I was watching tv). I will try to do some more tests where I pay attention. But I hope you see my point. :P

Thoughts anyone? I do have every little seperate piece of geometry in its own layer, so I don't think optimizations have anything to do with it..... It just seems to me that I should get comparable results to other people...

Also, version 1.02 does seem to be a little zippier - but my point and question is if I am doing something funky that might be slowing things down. Seems they should be a little faster at realization and not still take hours and hours to render radiosity at 640x480. Still a great setup tool none the less!

-Adrian:p

[sinthetique]

Originally posted by Architook
You don't need a plugin to make blurry reflections, it's built into Lightwave. On the surface panel, go to Environment, and set Blurry Reflections higher. It works in native lightwave, and even faster in fprime.

How, exactly, did I never see that? Thanks!!! :)

[ACLOBO]

Originally posted by Architook
There's also built in blurry transparency in LightWave, which is very cool indeed and I always wonder why I don't see more of it used. I think the blur options are just overlooked by most animators.

I think I can comment on that observation. In real world productions, ray-tracing is often not used as it is a slow process to do when you have a TON of shots that need to be done. Instead things are rendered in many passes and I think that the good artists are doing that stuff in post. The average lightwave user probably renders everything in one go (a final pic). As such, they have a large picture to look at with all this stuff going on - instead of individual passes that can "remind" one that this element is too soft or hard or bright or whatever. So, yeah - to make a long post longer, it is overlooked because until a plug-in like Fprime came along, we were all bogged down into just trying to get our lighting right, let alone worry about little things like blurry eye candy! :D

-Adrian

[TyVole]

Originally posted by Architook
There's also built in blurry transparency in LightWave, which is very cool indeed and I always wonder why I don't see more of it used. I think the blur options are just overlooked by most animators.

I see blurry reflections, but where do you set blurry transparency? Do you mean Additve Transparency?

[WizCraker]

Rendered in FPrime 1.0.2

Was FPrime always able to render Caustics? Notice the Sphere on the right.

1 MC bounce 1 Area light Left, 1 Spot Light Right. Scene was the Raytrace Test in the LW Content Directory.

http://www.imagereality.com/Images/NewtekImages/RayTraceTestFPrime102.jpg

[lasvideo]

The sphere on the right seems to be refracting light through a transparent surface.I see no evidence of caustics. And yes...Fprime has always had the abilty to preview/render raytraced refraction.

[wacom]

Originally posted by ACLOBO
Well, ok, first of all I do love Fprime. However, I dont' seem to be having quite the impressively fast results that others are getting. Now, I know my system is by no means top of the line, but it is still a decent machine. I have a Dell p4 2 2ghz (the fist available as a matter of fact I know the 400 mhz bus is slow). I also have 768 megs of ram in this machine and it does well for all my needs at the moment. However, I have done a couple of test renders myself and just finished one with radiosity with the new 1.02 Fprime patch. Here are my examples. There are roughly 230k polys in the scene.....

First pic is with version 1.01. I am not sure about how many bounces - I think 2 or 3 and it took about 3 hours to get to this level I believe (on and off) - so you can see where I am wondering how others are getting faster and better results with radiosity. This one cooked in previewer....

The second pic is after the 1.02 update and again is Monte Carlo with 3 bounces. In each scene there are only 3 luminous polys lighting the scene (you can see their reflections on the car) Level 8 render from renderer. I think this one took between 1.5 and 2 hours (not sure I was watching tv). I will try to do some more tests where I pay attention. But I hope you see my point. :P

Thoughts anyone? I do have every little seperate piece of geometry in its own layer, so I don't think optimizations have anything to do with it..... It just seems to me that I should get comparable results to other people...

Also, version 1.02 does seem to be a little zippier - but my point and question is if I am doing something funky that might be slowing things down. Seems they should be a little faster at realization and not still take hours and hours to render radiosity at 640x480. Still a great setup tool none the less!

-Adrian:p

Ceck for N-gons and tripple them if you find them.

I have almost the same machine/setup as you and I've seen HUGE improvments in speed with FPrime. I have a Dell 1.8Ghz Dimension 8200 with 768MB RDRAM. Try posting your scene and files and I'll try and render it...and see what's going on with it.

[WizCraker]

Originally posted by lasvideo
The sphere on the right seems to be refracting light through a transparent surface.I see no evidence of caustics. And yes...Fprime has always had the abilty to preview/render raytraced refraction.

Strange the LW render did not render it. That is why I was assuming it was Caustics.

[lasvideo]

Check to see if, in the render panel, the selection for "Raytrace
refraction" is checked.

[WizCraker]

Originally posted by lasvideo
Check to see if, in the render panel, the selection for "Raytrace
refraction" is checked.

I will check when I get home. Brought the Laptop to work but left the dongle at home.

[Zafar Iqbal]

Im curious: Does FPrime's refinements eleminate interferences from say... a large and systematic object (like a grid disapearing into the horizont)??

[WizCraker]

Refraction turned off, have a Light Probe in Imageworld, and the caustic like effect only shows up when using MC with FPrime 1.0.2

[T-Light]

Here's a quicky using Tracer-Radiosity in the Benchmarks directory. The picture on the left (no AA) is from the Lightwave renderer with caustics set to on, The picture on the right from FPrime 1.02, also with caustics on.

Nothing. :( . Pity, you had me going there.

WizCraker, where is the scene file that you're using in your test?

[T-Light]

Another test.

This time using CausticRingSlow.lws from _Features/Caustics.

I've tried pushing the intensity within FPrime to achieve a caustic lighting effect. The 'effect' can be achieved, but only slightly. If the intensity is pushed much higher, the ground washes out and the affect dissapears. I agree with you that your test looks to have caustics, but it could well be as Lasvideo says, simply refraction (Albeit, a rather powerfull example of it).

[T-Light]

It's a bug.

It's a pretty good looking bug, but a bug none the less. I found your test scene and tried various kinds of lighting. The effect only shows with image world and HDR images. I then loaded up a few other scenes but couldn't reproduce the effect. After going back to your original scene, I noticed the scene is completely enclosed. AKA, As far as LW and the LW renderer is concerned, the HDR image can't light your scene. For the most part, FPrime recognises this too, that is, everything apart from glass.

I remember seeing something in Worley's original FAQ stating that plugins such as image world may work or may not. I suppose this is one of those 'may not' moments.

Shame :(

Damn that's a nice looking bug.

[WizCraker]

Originally posted by T-Light
WizCraker, where is the scene file that you're using in your test?

It is some where in the LW Content Directory it is called RaytraceTest.lws. I'll post the scene file when I get home.

I find that you get the effect when using ImageWorld as I rendered out with it turned off and it did not do it.

[ACLOBO]

Is image world a separate plug-in or am I just not seeing it in my plugins folder? I can see textured environment - is that it? I suppose not since Image World is not spelled, nor sounds like Textured Environment.... :p

-Adrian

[WizCraker]

It is under the Scene Tab option Backdrop then under Add Enviroment.

[ACLOBO]

I must be high or something. I looked and looked and there are only three options under the add environment tab. How could I have missed it - it is right about textured environment..... I simply think that my brain enjoys playing tricks on itself at times. I don't see how I could not see it as it was there all along....

Now for another question..... What is image world? :D :D :D

-Adrian

[WizCraker]

Got a Reply back from Worley on this Caustic Effect.

The "caustic-like" effect is correct, it IS caustics, and it IS "caustic-like". LightWave has an interesting design where caustics are computed ONLY from lights. Fprime doesn't support this.
But the interesting LW design
does NOT compute caustics from OBJECTS (like a luminous one) or background HDRI maps.
There's a reason for this, doing those caustics is expensive and slow, but FPrime's radiosity is so fast I left it enabled. LightWave takes pains to deliberately turn it off, again I expect for speed reasons.

This is kind of one of our design decisions, so we try to MATCH LightWave, or should we do the Right Thing even though LW is not?

We don't document this because usally it's subtle, and who knows, we may change the way it works. And since we DON'T support forward light caustics (yet?) it would confuse people. So better to leave it as an undocumented feature. :-) We could make a toggle switch to turn it on and off but that locks us into the design forever.. and we're being very careful with the FPrime GUI, trying to keep it easy and streamlined.

*Edit: posted the entire email instead of just parts.

[T-Light]

Hello WizCraker.

Yup, I wrote in with this too, and had the same response. Unfortunately, I think I may have been as clear as mud in my description.

The effect rendered by FPrime in my earlier example of 'CausticRingSlow.lws' certainly seems to reciprocate this 'subtle effect'. What interested me about your image was that it was so dazzling. When I found your scene and managed to replicate the effect, I found I couldn't repeat it with other scenes.

The reason I went on to say it was a bug, was because, although the transparent sphere is showing the effects of the ImageWorld plugin, it would appear that the rest of the surfaces in the picture are not.

As the scene is an enclosed environment (4 walls, ceiling, floor), the transparent object, like the rest of the surfaces, shouldn't react to ImageWorld.


I tried a few experiments with the scene, to see how it reacted under different conditions.

Hope this makes things a little clearer. :)

----------------------------------------------------------------
Scene - Scenes/Setup/Raytracetest.lws

Same scene for all images.
All images rendered in FPrime.
All images created with MC, one bounce, 100% intensity.

----------------------------------------------------------------

1) Image created without ImageWorld.
Normal render - as expected (same as LW).

2) Image created with ImageWorld + HDR light probe.
Light now emanating from transparent object (other surfaces unaffected).

3) Scene lights set to 0%. Other settings as in (2).
Ghostly light now shown emanating from sphere.

4) Lights as in (1) and (2). Wall surface made 100% transparent.
Now that all surfaces in the scene can react to the HDR image, The subtle effect of caustics around the base of the sphere has dissapeared.

[minus]

On that transparency blurring thing... I can't find it either... Did you mean "Refraction Blurring" ?

[Architook]

Tlight- what happens if you have your ray recursions set higher? How about MC bounces higher?

I have a guess.. when transparency runs out of ray recursions, it shows the background color. If that background from your HDR, it would be very bright. So if the "glow" goes away when you increase ray recursions and MC bounces, that would explain it.

I'm just pulling this out of air, but I once had a problem in LW where ray recursions made my glass windows glow. That wasn't with radisity or fprime though.

[T-Light]

Hello Architook,

I know where you're coming from, but ray recursions in this case isn't the answer. I've just ran a quick test on lower ray recursion's and the transparent object does indeed get brighter, in fact, it gets so bright that anything under about 5 recursions, the image is unusable.

The images from my last post were set to 24 recursions.

The ghostly light effect (sorry for the simplicity - there's no obvious visible light source in the image) would seem to be caused by fprime taking into account all surfaces (including transparency) and making a correct judgement about their appearance within the given lighting and scene parameters.

BUT...

The light passing through transparent objects, is taking into account HDR lighting, even though the surface is in an enclosed environment and couldn't possibly be lit by any outside sources.

I must emphasise 'light passing THROUGH transparent objects'. The transparent surfaces (reflection, diffuse, specularity etc) all react in the same way as other object surfaces within the scene.

Clear=Mud=Me.
Sorry, if I confused the issue earlier, after all, this whole thing started with a debate on caustics within FPrime.

Please let me state - this is no longer just about caustics. :)

Here's a close up of the transparent sphere. (The only really relevant image from the last set I posted).

All lights set to 0%
Ray recursion set to 24
MC 1 bounce, intensity 100%
Walls (including Ceiling), and floor set to 0% transparency.( There's no light getting in!)
HDR light probe image used within the ImageWorld plugin.

[Architook]

TLight, so what about using more MC bounces, though? If you have that set to 1, then you're getting the same effect as a ray recrusion of 1.. maybe radiosity can't go through more layers than MC bounces?

Does the ghost change when you have MC=3? 10? fprime doesn't seem to be slowed toomuch by higher levels.


The fact that main transparency gets brighter because of the HDR is the clue that makes me think this is the same effect, but for radiosity.

I like puzzles.

[T-Light]

See,

Now I'm looking at these posts after a night out with the family, a few beers, and a bottle of good wine.

Laptop's locked up for the evening (with the dongle), I'll run the tests tommorow morning first thing. I have a suspicion here that your suggestion of increasing MC bounces will only make the surrounding objects brighter, but we'll see. (I saw that earlier and didn't test it - my apologies- just didn't think it would affect the glow itself)

By the way, the scene is in the standard LW install folders, everyone can run these (please see my previous posts - I'm a few sheets to the wind here)

Till Tommorow. :D

[T-Light]

Bless your cotton socks Architook.

You were 100% correct on your idea of increasing MC bounces.
I stand corrected. :)

The images below were taken with all the settings from the above scene (previous posts), Only MC bounces are varied.

ps It's still a bug. :p

[WizCraker]

Originally posted by T-Light
ps It's still a bug. :p

Actually it is not a bug. If you Re read the email that Worley sent, which I originally posted on page 3 it is a feature that is in LW but LW turns it off to speed up rendering.

But the interesting LW design
does NOT compute caustics from OBJECTS (like a luminous one) or background HDRI maps.
There's a reason for this, doing those caustics is expensive and slow, but FPrime's radiosity is so fast I left it enabled. LightWave takes pains to deliberately turn it off again I expect for speed reasons.

This is kind of one of our design decisions, so we try to MATCH LightWave, or should we do the Right Thing even though LW is not?

We don't document this because usally it's subtle, and who knows, we may change the way it works. And since we DON'T support forward light caustics (yet?) it would confuse people. So better to leave it as an undocumented feature. :-)

[mbaldwin]

nice to see the experimentation continues. I think I've come up with a glitch--double-sided polys with transparency masks(images) showing up correctly in reflections of other surfaces. I'll have to test further and post.

In the meantime, the fun continues. here's more procedural-only surfacing. The biggest time saver for me is FPrime essentially removing all the blind alleys. immediate feedback to scene specific texturing leading to fast completion of shots. The 3d creation process for me is finally humming along.

[T-Light]

I would have thought the novelty of fprime would have worn off for me by now, but I still can't pull myself away from it.

ps That's a terrific image. :cool:

[3DBob]

Great Image Mr Baldwin

3DBob

[T-Light]

"Actually it is not a bug." - WizCraker.

OK, I'll try one more explanation then I'm letting this go. :)

Please understand WizCraker, It's not that I didn't read Steve Worley's comments on FPrime's built in caustics, I posted the previous examples and comments 'Because' I read and understood Steve's email thoroughly.

The points I was making on the glow effect/bug were as follows.

----------------------------------------------------------------------

1) FPrime's built in caustics react to all light sources.

--> The glow effect in your image is only reacting to the HDR in the 'ImageWorld' plugin.


2) Built in FPrime caustics become brighter with increased Monte Carlo Bounces, (Please see the pictures I posted from the 'CausticRing' scene for an example).

-->The glow effect in your example dissapears if you increase Monte Carlo bounces.


3) FPrime's built in caustics are generally "subtle" -SWorley. (Again, please see the 'CausticRing' scene for just how subtle).

-->The effect in your initial image is comparatively bright.


4) FPrime's built in caustics will diminish as available light is reduced.

-->The effect in your image is unchanged, even with all available lights switched off!


----------------------------------------------------------------------

Either we agree that we are talking about two different things, or we'll have to agree to disagree.

Cheers.

[policarpo]

Originally posted by mbaldwin
nice to see the experimentation continues. I think I've come up with a glitch--double-sided polys with transparency masks(images) showing up correctly in reflections of other surfaces. I'll have to test further and post.

In the meantime, the fun continues. here's more procedural-only surfacing. The biggest time saver for me is FPrime essentially removing all the blind alleys. immediate feedback to scene specific texturing leading to fast completion of shots. The 3d creation process for me is finally humming along.


Oh man, this image is so tight!

Add some crusty moss to the heads...and a small female figure walking between the two giant heads....

ROCK!!!!

[policarpo]

This is kind of what I was seeing as the color space for the image.

I hope you don't mind me tweaking it, i just love the image though.

Thanks for inspiring me!!

[mbaldwin]

Poli,

thanks for the kind words--no probs here with your tweaks. It's amazing what a good coat of paint will do for your objects. Here's a look at the same scene unadorned:

[mbaldwin]

and here's a look at a slightly different angle. I have to say, Fprime has let me take 3d 'sketch' sessions farther along than I would have without the plug-in. Without it, who needs the tedium(life's too short)!?
With it, at the very least I wind up learning a few things(like there's a hell of a lot you can do just with bump maps, procedurals and a few gobo shadows to sell a shot). And in the end, I have a few useful textures for my files.

thanks for the comments,

-m.

[policarpo]

Humm-a-da humm-a-da!

Baby!

Now add some localized high spec to make parts of it look wet.

Awesome work!!!

[silverlw]

I also had to try 10 bounces and lumeobjects and this is what i got after 7 hrs of rendering. This is after 380 refinements. Fprime badly need some sort of noisereduction. This image came out well but the time for it to refine would drop dramatically with NS i think. I would probably have settled for 25-40 refinements if i had NS like in Lightwave. and then this image would have taken 10-15 minutes instead.
http://hem.bredband.net/b223277/7hrs10bouncesLumeObj.jpg

[Panikos]

silverlw.

Noise = lack of light
Is like shooting with a video camera with insufficient light.

Light helps in having more consistent imagery.
Lum-polys work, but nothing works as well as a real lw-light.
Additionally, if you manage to reduce radiosity effect with a light, you get faster results.

Lum-polys are good for filling purposes, and not for subtituting light sources.

"Noise" or grain is something relative. If you have a RGB image of a real movie, do you think that it would be as clean as 3D ?
Noise or Grain are part of the game if you involve radiosity.
3d imagery without radiosity look too "CG"
The image that you attached looks fine to me.

Noise Reduction digests details.

[mbaldwin]

silverlw,

looks good! I know what you mean about noise--that lw's standard f9 render seems to deliver a less grainy image a lot faster than fPrime's preview render. Have you done the same image with fPrime's non-interactive renderer? or is that indeed what we're looking at?

No complaints here, though. I'm still giddy from the drastic speed increases to the other aspects of lightwave workflow and rendering. I would love to see Multiprocessing support on the next update. I hope Worley can also deliver a version that works with G2 and other shaders!

-m.

[mbaldwin]

And while I'm chiming in again, 'thought I'd go waaaay back in the archives and post the first model I ever made. I dusted it off and tweaked the scene in no time flat.

I envy the 3d youth of today--oh, if only I had fprime(and a G5) back when I was starting out. Just think what I'd be creating now.

Onward!

-m.

[silverlw]

It's with Fprime preview. Have you seen any improvement with Fprime render concerning noise?

[CoryC]

Originally posted by silverlw
It's with Fprime preview. Have you seen any improvement with Fprime render concerning noise?

FPrime render is a higher quality render. The preview has to sacrific some quality for speed and interactivity.

[wacom]

Originally posted by silverlw
I also had to try 10 bounces and lumeobjects and this is what i got after 7 hrs of rendering. This is after 380 refinements. Fprime badly need some sort of noisereduction. This image came out well but the time for it to refine would drop dramatically with NS i think. I would probably have settled for 25-40 refinements if i had NS like in Lightwave. and then this image would have taken 10-15 minutes instead.


The more bounces you have, the more noise you'll get too. Try this at 2-4 bounces and a higher intensity level and see what you get.

[tokyo drifter]

Originally posted by Panikos
silverlw.

Noise = lack of light
Is like shooting with a video camera with insufficient light.

Light helps in having more consistent imagery.
Lum-polys work, but nothing works as well as a real lw-light.
Additionally, if you manage to reduce radiosity effect with a light, you get faster results.

Lum-polys are good for filling purposes, and not for subtituting light sources.

"Noise" or grain is something relative. If you have a RGB image of a real movie, do you think that it would be as clean as 3D ?
Noise or Grain are part of the game if you involve radiosity.
3d imagery without radiosity look too "CG"
The image that you attached looks fine to me.

Noise Reduction digests details. But CG is all about faking it. Noise is realistic but it would be faster and more flexible if we could get clean noiseless renders and add noise in post if needed. The more bounces you have, the more noise you'll get too. Try this at 2-4 bounces and a higher intensity level and see what you get. Wacom, That's a good tip, I'll need to test that out.

[WizCraker]

Actually noise is what we see in film we are use to it so therefore if we see something that doesn't have it we say "hmm that looks fake."

[silverlw]

Did a experiment to "clean" this image myself and that looked alot better. Hopefully worley is bizzy adding noisereduction right now;)
http://hem.bredband.net/b223277/7hrs10bouncesLumeObj_filtered.jpg

[Panikos]

If you had a scene with textures, noise reduction would digest a lot of details, mostly bump maps

[3DEagle]

Hi Silverlw,

You did some kind of strange exeriment...the picture is now a lot better...what did you do....washed it in Photoshop?

Enjoy Sunshine in Life

Ole

[silverlw]

I used an external noisereductiontool called Neat image.
Neat image homepage (http://www.neatimage.com/examples.html)
Neat image is very god at cleaning up noisey images and can do batchjobs. Unfortunatly i have no other solution until Worley implements gammacorrection and noisereduction in Fprime.

[silverlw]

This is a silly test i did with 13iterations (very noisy) and after that i filtered it with Neat image.
With textures and other lights in the scene i would probably accept the quality at about 15-20 iterations, if Fprime had some sort of noisereduction. I really miss lw's noisereduction and if anyone have tips for how to postprocess the noise away in a better way im all ears:p

1radbounce 13 iterations with noise1radbounce 13 iterations with noise (http://hem.bredband.net/b223277/Unfiltered%20radroom%2013iterations.avi)
1radbounce 13 iterations with (less) noise (http://hem.bredband.net/b223277/Filtered%20radroom%2013iterations.avi)