PDA

View Full Version : I want to render like Maxwell Render



Filmdesigner
05-13-2009, 03:03 AM
I have accepted a private competition to beat a friend who uses Maxwell Render. I have openly told him Lightwave 9.6 can create the same renders if not better.

In vain or not, I have challenged him. I now need to learn how to use Lightwave better to create interiors at a level far beyond what I am creating now. My interiors are looking terrible and flat at the moment.

Can anyone help with tutorials, sites or anything that will aid me in the whupping?

Thanks all.

flakester
05-13-2009, 03:21 AM
Can't go far wrong with this pretty comprehensive guide here: http://www.except.nl/lightwave/RadiosityGuide95/index.htm

Written for 9.5, but I'll be surprised if it doesn't help you understand a whole lot more about radiosity solutions. :thumbsup:

Hope that helps.

--
flakester

ingo
05-13-2009, 04:07 AM
The best thing to compete with Maxwell will be Kray, just take a look at the forum at http://www.kraytracing.com/forum/

Matt
05-13-2009, 04:12 AM
For true realism, you won't beat Maxwell.

For render time with _acceptable_ / _passable_ as real, you can.

biliousfrog
05-13-2009, 04:24 AM
Maxwell is physically accurate where as Lightwave is a brute force solution so it will be partly down to you to create the realistic lighting. Don't expect to just put an emissive object in a room and let Lightwave do all the work...it works for Maxwell but not Lightwave, they're two very different renderers. It's a trade off between accuracy and speed, Maxwell has the first and Lightwave has the second.

The best thing to compete with Maxwell would be Fry, it's the same type of renderer (and more user friendly IMO)...but it wouldn't be a test of Lightwave's renderer then would it. ;)

archijam
05-13-2009, 04:25 AM
Scene setup will help you greatly .. go for HDRI lighting and reflections, blurred reflections, and Occlusion pass etc.

Also, try to compose and select the positioning, materials and objects carefully to allow for some subtle DOF.

If you want it to be 'all LW' use Exception's old trick - tacke your rendered image, map it to a poly or set as background, then re render with all post-effects - vignetting, corona, virtual darkroom, etc. etc.

Much faster and much more control :thumbsup:

Now go get 'em tiger (good luck...).

Jockomo
05-13-2009, 06:46 AM
You know I hate to break it to you; but the true talent lies in the artist, not the software.
Your competition will only show the strengths and weaknesses of your skills, not the 3d renderer.

MooseDog
05-13-2009, 08:14 AM
scene set-up as stated will be your focus. here a rather long discussion on different approaches for interior lighting with some pretty good results:

http://www.spinquad.com/forums/showthread.php?t=25361

good luck:thumbsup:

Tranimatronic
05-13-2009, 09:17 AM
My biggest tip would be - put a strong light in somewhere and add the bloom shader. I do most of my kitchen interiors as early morning type lighting. If you can work from a photo do so - try to add some of the mistakes added by poor photography that people are used to seeing (like the bloom). Otherwise what Jockomo said.....

Matt
05-13-2009, 10:55 AM
My biggest tip would be - put a strong light in somewhere and add the bloom shader.

I add bloom in Photoshop, copy the layer, crunch the histogram down to concentrate on the highlights, blur it, set the layer to screen mode, and adjust the transparency until it looks right.

adamredwoods
05-13-2009, 11:08 AM
Also, show us what you have so far and we'll point you in the right direction.

But I agree with everything said here, mostly it's about the user, not the software.

biliousfrog
05-13-2009, 11:45 AM
try to add some of the mistakes added by poor photography that people are used to seeing (like the bloom).

Unfortunately, that just makes images look more CG because professional photographers would try anything to reduce those defects such as lens flares, chromatic aberration, depth of field, noise etc. You're confusing what you perceive as photo-real with real photo's...it's a common mistake.

...that's not to say that those effects shouldn't be used, they have their place. If you want to replicate a bad or ameteur photo those effects can increase the realism. Sometimes you need to show people what they expect rather than what they would really see. The problem is that, for architectural and product work, you're competing with professional photographers.

I'd strongly recomend reading some architectural photography books and learning about their techniques. The photo's are usually pin-sharp with good contrast, little or no noise and perfect exposure.

jasonwestmas
05-13-2009, 11:50 AM
Flatness can be remedied usually with good lighting, specular, reflection and color composition. show us some screen shots.

Tranimatronic
05-13-2009, 01:24 PM
biliousfrog - yeah - but as you already mentioned Maxwell does the physically accurate stuff - wouldnt you be better trying to add photographic mistakes the people are used to seeing ? Maybe go for this kind of feel instead of trying to fight fire with fire.

Mattoo
05-13-2009, 08:14 PM
-If reality is the goal then a good understanding of a linear gamma workflow is a big help.

-Also, using the Energy Conservative material nodes will help take the guess work out of it and keep things fairly accurate (technically).

-Almost every material has some degree of Fresnel or glancing angle effects (colour shift and/or increased diffusion).

-Don't use CG lights, use luminous objects and rely on blurry reflections instead of speculars. Correcting for gamma (mentioned above) with this method is quite important to get the correct light fall-off.

- Don't use ambient occlusion or any baked on lighting. Rely on Monte Carlo GI alone.


It'll be a beast of a render but it'll be getting a lot closer to what Maxwell is doing under the hood. Good luck.

Filmdesigner
05-13-2009, 11:30 PM
Thanks All. I intend to do the job I can do but as expected the tiny fraction of LW I use will not be enough and reading and learning is in order.

I create most of my renders using the most basic set-up I can. This is causing trouble as I haven't had the time to learn the newer features of LW and have relied a lot on Exception's tutorials to make sense of the new Radiosity Engine.

I am deciding what scene I will use and will post thumbnails shortly.

Thanks All

clagman
05-14-2009, 08:27 AM
-If reality is the goal then a good understanding of a linear gamma workflow is a big help.

-Also, using the Energy Conservative material nodes will help take the guess work out of it and keep things fairly accurate (technically).

-Almost every material has some degree of Fresnel or glancing angle effects (colour shift and/or increased diffusion).

-Don't use CG lights, use luminous objects and rely on blurry reflections instead of speculars. Correcting for gamma (mentioned above) with this method is quite important to get the correct light fall-off.



Will probably be difficult to obtain the look of IES lighting without using CG lights. I suppose one could actually model a reflector and use MC caustics to reflect a luminous objects light.

adamredwoods
05-14-2009, 11:57 AM
-Don't use CG lights, use luminous objects and rely on blurry reflections instead of speculars. Correcting for gamma (mentioned above) with this method is quite important to get the correct light fall-off.


Great tips. Why would using luminous polygons be better than say, an area or sphere light?

biliousfrog
05-14-2009, 12:46 PM
Great tips. Why would using luminous polygons be better than say, an area or sphere light?

#1 benefit is that they show in reflections

#2 they're often faster for some reason

JeffrySG
05-14-2009, 01:37 PM
I was also thinking that if you wanted to do a interesting comparison between LW and Maxwell you should pick a scene that would not exist in reality. Maybe a cool spaceship fighter / battle scene with explosions. Let's see how Maxwell renders that? Perfectly photo-real probably will not work out too well in that instance.

toby
05-14-2009, 09:19 PM
Find otocon, do everything he says

Filmdesigner
05-14-2009, 10:13 PM
After going through a collection of former projects it has been agreed to select one of the attached scenes. The wider image is a pretty good place to start as the render settings are actually extremely basic. The second was a dog of an image and suffers from my lack of patience (and lack of modeling and texturing ability).

I would agree with the comments about using another scene other than an interior as LW does a great job elsewhere, but as my main money creation facility is Archivis I will try this interior. I have no trouble with exteriors for the obvious reasons, but interiors are like a thousand paper-cuts.

I will post settings shortly.

Phil

biliousfrog
05-15-2009, 03:00 AM
I don't think that there's too much wrong with those renders lighting wise, most of the things I'd pick up on are regarding the models and surfaces.

jasonwestmas
05-15-2009, 07:58 AM
#2 they're often faster for some reason

But only faster when using GI? Luminous polys only light stuff up if GI is used.

adamredwoods
05-15-2009, 11:43 AM
But only faster when using GI? Luminous polys only light stuff up if GI is used.

I tried to use a luminous polygon, but it wasn't powerful enough to light a room, even at 10000%, and 400%GI. I was treating as if it an area light, so I'm wondering if there's a better way to implement it?

The alternative I used was textured background (easier to position than imageworld). That worked and made the room nice and bright, but with a lot of light leaks (different problem).

jasonwestmas
05-15-2009, 12:13 PM
I tried to use a luminous polygon, but it wasn't powerful enough to light a room, even at 10000%, and 400%GI. I was treating as if it an area light, so I'm wondering if there's a better way to implement it?

The alternative I used was textured background (easier to position than imageworld). That worked and made the room nice and bright, but with a lot of light leaks (different problem).

That's a good thing to experiment with. I tried using only a lumigon sphere to light a head with final gather GI. I got some really creepy but realistic results. It was creepy because of how life-like the lighting condition was. I think I'll try that again. Of course I wouldn't use that for animation but it makes for a really cool still.

toby
05-15-2009, 12:34 PM
Lumigons (good term!) are faster than area lights because you're using interpolated gi instead of 'brute force' samples, where you always have noise without really high settings. But with lumigons you'll have to worry about flickering.

Keep in mind that lumigons are not casting light, you're relying on your subject to cast enough rays to hit the lumigon and pick up luminance data from it. So if it's a small lumigon it will be missed by most rays, and it won't add light to your scene, or you'll need to raise your rpe higher, slowing things down.

I did a worst-case test when 9.5 beta came out and posted it, I'll try to dig it up.

adamredwoods
05-15-2009, 01:19 PM
Keep in mind that lumigons are not casting light, you're relying on your subject to cast enough rays to hit the lumigon and pick up luminance data from it. So if it's a small lumigon it will be missed by most rays, and it won't add light to your scene, or you'll need to raise your rpe higher, slowing things down.

I don't see how that would work for a room with a window, then. Is there something I'm missing like using "fake" poly's in the room itself? The problem with that is that you can't hide the polygons from rendering, b/c then the light will not show (no rays).

It may be faster, but it doesn't seem so practical.

toby
05-15-2009, 01:49 PM
I don't see how that would work for a room with a window, then. Is there something I'm missing like using "fake" poly's in the room itself?I don't follow :confused:


The problem with that is that you can't hide the polygons from rendering, b/c then the light will not show (no rays).

It may be faster, but it doesn't seem so practical.
I think leaving Seen by radiosity On takes care of that.

toby
05-16-2009, 12:12 AM
I tried to use a luminous polygon, but it wasn't powerful enough to light a room, even at 10000%, and 400%GI. I was treating as if it an area light, so I'm wondering if there's a better way to implement it?

Here it is
http://www.newtek.com/forums/showthread.php?t=83349&
I had the lumigon at 3000% & gi 100%, but it was bigger than a door (more rays hit it so you get more light from it)

biliousfrog
05-19-2009, 06:13 AM
Here it is
http://www.newtek.com/forums/showthread.php?t=83349&
I had the lumigon at 3000% & gi 100%, but it was bigger than a door (more rays hit it so you get more light from it)

It won't allow me to access that for some reason toby.

You can hide the polygons using unseen by camera, I often use them in windows for interiors so that you get realistic reflections and light bleed...for lights too if they need to be picked up in reflections.

The three attached images use only 'lumigons'...the interior might have used DPont's infinite light for the sun but everything else is luminous polys. The interior is from an animation, I needed to disable directional rays (GI) to stop the lumigons from showing through the glass.

...lumigons is a great term BTW

SAHiN
05-19-2009, 07:39 AM
Does Maxwell render better than this ? I never played with maxwell render.
But this was rendered on LW out of its package..No other render plugins.

http://www.newtek.com/forums/showthread.php?t=98794

jasonwestmas
05-19-2009, 07:42 AM
Very cool Biliousfrog. Did you use luminous planes or some other kind of geometry to light the scenes?

SAHiN
05-19-2009, 07:56 AM
Does Maxwell render better than this ? I never played with maxwell render.
But this was rendered on LW out of its package..No other render plugins.

Sorry for repeated post ( couldnt target the image on first one I guess )

biliousfrog
05-19-2009, 08:03 AM
Does Maxwell render better than this ? I never played with maxwell render.
But this was rendered on LW out of its package..No other render plugins.

http://www.newtek.com/forums/showthread.php?t=98794

'Better' is down to preference but it is much more accurate for photo-realism. All surfaces are based on physical properties, the lighting is physically accurate and the camera acts just like a real camera.

Lightwave is more reliant on the artist creating something realistic with the supplied tools. I guess it's a little like comparing a photograph with a painting, neither is better but they each have an edge over the other for different purposes...in Lightwave's case it is speed, Maxwell's is accuracy.


Very cool Biliousfrog. Did you use luminous planes or some other kind of geometry to light the scenes?

For studio renders I use strips or panels just like in a real studio, for the interior I used discs for the down lighters and duplicated the glass panels for the window illumination.

adamredwoods
05-19-2009, 07:55 PM
Lumigons seem to work ok.... the trick is to make a curved panel. I used four polys in this exercise, set in a concave arrangement. This vastly increases the radiosity samples versus a single flat polygon.

lumigon lumination 100%, light blue background
GI 500%, 2 bounces, 700RPE, 20% multiplier

I've also attached the Textured environment render without the lumigons. THe render time was significantly higher, but I think it was due to the AA on the image through the glass.

allabulle
05-20-2009, 05:38 AM
Also, sometimes it's useful to use a simplified version of your scene and import it into Blender. Then, using the free Indigo renderer to simulate the light mood you desire (it's an unbiased render like Maxwell or Fryrender) you get an impression on what to do later using plain LightWave tools, taking the renders from Blender/Indigo as a reference on tones, light, and radiance. Knowing the main light sources of your scene and having a more or less real light distribution before you play with LightWave settings it's a real time saver on most occasions and, also, makes it easy to plan what to do in post and why depending on the schedule, the CPU's available, etc.

toby
05-21-2009, 10:21 PM
It won't allow me to access that for some reason toby.

blast, that's ooold 9.5 beta forum, guess it's still restricted.
here's a re-up

Iain
05-23-2009, 07:49 AM
I've done some tests trying to get Maxwell like accuracy from LW. The easiest way is to run a test in Maxwell and copy it in LW but failing that, I'd say get the surfacing 100% and you're almost there.

Lighting can be tweaked to a very high level of realism, especially using Area Lights with Inv Dist^2 falloff or luminous polygons for fill.

allabulle
05-23-2009, 09:28 AM
I've done some tests trying to get Maxwell like accuracy from LW. The easiest way is to run a test in Maxwell and copy it in LW but failing that, I'd say get the surfacing 100% and you're almost there.

Lighting can be tweaked to a very high level of realism, especially using Area Lights with Inv Dist^2 falloff or luminous polygons for fill.

Yep, I've come to the same conclusion: it's the surfacing. But to know what to expect from "real" light behavior it's quite useful to test it with an unbiased renderer. A cheap option is Blender with Luxrender, which are free (and not the Indigo renderer as I stated in my last post: it seems that now it's a commercial renderer. Sorry if I confused anyone with my mistake.)

Iain
05-23-2009, 09:47 AM
I think it's also worth mentioning that I've seen hundreds of Vray/FPrime/LW/Kray/Mental Ray (etc) renders that look better than most Maxwell and Fry renders due to artistic and technical ability.

Good Maxwell and Fryrender images just always have something extra though, something hard to pin down like subtle vignetting or DOF.
All of which are easily achieved in LW-it's just that Maxwell does it by default and usually to perfection.
If the eye doesn't pick it up right away, it usually means it's right.

allabulle
05-23-2009, 01:16 PM
I think it's also worth mentioning that I've seen hundreds of Vray/FPrime/LW/Kray/Mental Ray (etc) renders that look better than most Maxwell and Fry renders due to artistic and technical ability.

Good Maxwell and Fryrender images just always have something extra though, something hard to pin down like subtle vignetting or DOF.
All of which are easily achieved in LW-it's just that Maxwell does it by default and usually to perfection.
If the eye doesn't pick it up right away, it usually means it's right.

Yes, I completely agree. You explained it very clearly. Maybe Geradstrada could add something here: using a linear workflow helps, toning the colors well adds most of that "something" that makes images look real, or photorealistic.

I'm trying to implement such workflow, now. Learning first how gamma, color, light and grading work properly.

Filmdesigner
05-25-2009, 04:34 AM
Well. I've tried to get an interior worth viewing, but have given-up. The parameters are far too varied to get any kind of result I am happy with.

I've given up and have now sourced some contractors using competing software to do the work for me. That's bad news but I can't waste time trying to make Lightwave do a great job.

After this exploration and frustration I am seriously considering purchasing Maxwell and dropping Lightwave as my interior renderer. This is how frustrating the process was.

I agree that Lightwave is excellent for the "arty" type of artwork and texture-heavy renderings, but the subtlety I require is too difficult.

Oh well.

allabulle
05-25-2009, 06:08 AM
Well. I've tried to get an interior worth viewing, but have given-up. The parameters are far too varied to get any kind of result I am happy with.

I've given up and have now sourced some contractors using competing software to do the work for me. That's bad news but I can't waste time trying to make Lightwave do a great job.

After this exploration and frustration I am seriously considering purchasing Maxwell and dropping Lightwave as my interior renderer. This is how frustrating the process was.

I agree that Lightwave is excellent for the "arty" type of artwork and texture-heavy renderings, but the subtlety I require is too difficult.

Oh well.

Some people say LightWave works well with Maxwell. Adding LWCAD helps too. Maybe you could consider that scenario as well if quitting LightWave is something you don't desire.

Panikos
05-25-2009, 08:22 AM
Personally I dont care about other apps because images are subject to user's skills.
I am more focused in LW output. I want better and faster renderer in LW/FPrime.
I want flickerless GI in LW and I want volumetric lights in FPrime.

Can I have these ?

jameswillmott
05-25-2009, 09:02 AM
I want flickerless GI in LW...

Can I have these ?

9.6 has this now...

Panikos
05-25-2009, 09:21 AM
what about flickerless GI with displaced objects, light changes, surface changes ? :)

glebe digital
05-25-2009, 09:32 AM
After this exploration and frustration I am seriously considering purchasing Maxwell and dropping Lightwave as my interior renderer. This is how frustrating the process was.

I agree that Lightwave is excellent for the "arty" type of artwork and texture-heavy renderings, but the subtlety I require is too difficult.

Oh well.

I use both Maxwell & LW's internal engine for rendering, both are excellent in their own way & yes, the Maxwell plug works very well within LW.

However, if you want to produce interiors with Maxwell [within realistic commercial timeframes] you will need some serious hardware & multiple nodes, or good budgets that allow for renderfarm outsourcing........otherwise, you'll be asking for some major headaches & uncomfortable client conversations.

Maxwell is really superb for certain types of shot, product renders for example.......but interiors [with all that light bouncin' about forever] do take a LONG time to get clean..........

jasonwestmas
05-25-2009, 10:13 AM
what about flickerless GI with displaced objects, light changes, surface changes ? :)

Nope, we don't have that. You want to animate lights and surface changes? For what?

Panikos
05-25-2009, 11:09 AM
If we remove animation from LW, because GI doest like it, we will end up in a Stills generator.

Iain
05-25-2009, 12:07 PM
There is actually quite an easy way to get Maxwell-like interiors from LW:

Use Image World with a HDRI for your backdrop, add a 'sun' light using Sunspot, and turn on Monte Carlo radiosity with 4 bounces.
(You may need an Area light or two in the window openings.)

It might actually take longer than Maxwell to render but it'll look great.

toby
05-25-2009, 12:46 PM
There is actually quite an easy way to get Maxwell-like interiors from LW:

Use Image World with a HDRI for your backdrop, add a 'sun' light using Sunspot, and turn on Monte Carlo radiosity with 4 bounces.
(You may need an Area light or two in the window openings.)

It might actually take longer than Maxwell to render but it'll look great.
Do you mean use un-interpolated radiosity?

Some fantastic outdoor hdr's btw:
http://www.openfootage.net/?cat=15

toby
05-25-2009, 01:01 PM
Here's Otocon's gallery, 100% lightwave if I recall correctly
http://otacon.kvaalen.com/
( don't you just hate his guts?? :grumpy: )

virtualcomposer
05-25-2009, 01:27 PM
You know I hate to break it to you; but the true talent lies in the artist, not the software.
Your competition will only show the strengths and weaknesses of your skills, not the 3d renderer.

I agree. There are obvious setbacks like Lightwave vs Bryce or a really cheap LCD screen vs a really well made one with attention to colour accuracy. Lightwave, in my opinion is really the talent of the artist. It's extremely comparable to Maya. Not sure about Hudini since that's a 3D animation program unto itself. LOL My boss, who has been in the graphics design industry for over 20 years and could not tell between the real picture and what I did in Lightwave. It's all about textures for the most part but also full control over the environment like LW has. I have allot to learn and will always be learning but have been fully impressed with LW in the real world and the quality standards. One major thing that makes me look good at work is the render time. WOW WOW WOW They are used to long render times in the past but when a highly detailed render is finished in three hours with full radiosity and 200 antialiasing with other stuff, they are floored. LW just may get me a raise. LOL :)

glebe digital
05-25-2009, 01:39 PM
Here's Otocon's gallery, 100% lightwave if I recall correctly
http://otacon.kvaalen.com/
( don't you just hate his guts?? :grumpy: )

Nah.....plenty of Maxwells in there.

toby
05-25-2009, 01:46 PM
Nah.....plenty of Maxwells in there.
You know this for fact? Please explain -

glebe digital
05-25-2009, 02:18 PM
Just check the tags on the images Toby, in otacon's own words.........plus I remember some of them from way back, and the best were on the Next Limit site for a long time.

toby
05-25-2009, 02:20 PM
Nah.....plenty of Maxwells in there.

Example
http://www.newtek.com/forums/showthread.php?t=86241
and
http://www.renderosity.com/mod/gallery/index.php?image_id=1750511&member

glebe digital
05-25-2009, 02:32 PM
What's your point?
You stated they were 100% LW......I'm just correcting that, no need to get defensive and all 'thread mental' on my arse.......

All his work is great, I'm not disputing. :)

The darts, knife, compass and pocket watch.......all Maxwell, that's all I'm saying.....just calm down dude. :D

toby
05-25-2009, 02:34 PM
Just check the tags on the images Toby, in otacon's own words.........plus I remember some of them from way back, and the best were on the Next Limit site for a long time.
Ok, I didn't look at every image, a lot have no tags. But you can see he does do impressive stuff with lw -

And I'm only partially in agreement that "it's the artist not the tool"... time and patience play a factor in the finished product, and better software / renderers can vastly improve that.

toby
05-25-2009, 02:36 PM
What's your point?
You stated they were 100% LW......I'm just correcting that, no need to get defensive and all 'thread mental' on my arse.......

All his work is great, I'm not disputing. :)

The darts, knife, compass and pocket watch.......all Maxwell, that's all I'm saying.....just calm down dude. :D
:confused:
what makes you think I'm freaking out?

My point is that lw is not incapable of doing maxwell-looking renders, that's all.

glebe digital
05-25-2009, 02:39 PM
it's me.....soz......it's late, my deadline's approaching......appologies, I didn't mean to be curt with you big man. :) I'll go grab a cuppa and lower my pulse a bit....

archijam
05-25-2009, 02:44 PM
Well. I've tried to get an interior worth viewing, but have given-up. The parameters are far too varied to get any kind of result I am happy with...
Oh well.

So you gave it, what, a week and a half? (At least since you started this thread)

And that's it?

toby
05-25-2009, 02:55 PM
it's me.....soz......it's late, my deadline's approaching......appologies, I didn't mean to be curt with you big man. :) I'll go grab a cuppa and lower my pulse a bit....
Ok, no probz!


So you gave it, what, a week and a half? (At least since you started this thread)

And that's it?
Well that could be 60hrs just spent on render settings for one image...

<wishful thinking> What would be great is if there was a library of calibrated surface presets...</wishful thinking>
A linear color workflow could help a lot too - gerardstrata has one but omg it's hard to understand and implement.

JeffrySG
05-26-2009, 09:18 AM
Do you mean use un-interpolated radiosity?

Some fantastic outdoor hdr's btw:
http://www.openfootage.net/?cat=15

Cheers on that link!

:thumbsup::thumbsup:

Filmdesigner
06-12-2009, 11:08 PM
Thanks to all for reading this thread. My reasons for dropping this quest are simple; If I can open an application, dress the scene and hit render and get "god-like" results, why would I waste time trying to endlessly tweak settings to get the same result?

I charge by the hour. I can't justify spending more than a few hours getting a render right. If Lightwave can't offer simple presets and set-ups in future releases I will be moving to another renderer for my interiors. That said, I love Lightwave's exterior renders and will keep it for these alone and the film work I do.

I have seen some truly excellent renders done in LW and these examples keep me trying, however, I noted on the stats at the bottom of the page of one truly world class render that the set-up took months of adjusting.

I can't afford this.

Thanks again.

Cageman
06-13-2009, 01:32 AM
If we remove animation from LW, because GI doest like it, we will end up in a Stills generator.

What kind of surface changes do you want to change? Color should work just fine, even after the GI has been cached.

Regarding flicker, you really need to use the non-interpolated modes for that if you are having deformations on animated objects/characters. If the objects are only translated, rotated or scaled, Interpolated MC should be quite fine. For everything else, Non-interpolated Montecarlo all the way.

As a comparsion, in MR, final gather is a flickerfest for sure, so this isn't just an LW-thing.

Iain
06-13-2009, 04:10 AM
If Lightwave can't offer simple presets and set-ups in future releases I will be moving to another renderer for my interiors. That said, I love Lightwave's exterior renders and will keep it for these alone and the film work I do.


I can rig a realistic light set up in about 5 minutes. After an hour or so of tweaking, it's usually there.

Now Maxwell does cut out that hour and adds ultimate realism but the render times for an interior can be (literally) days at decent resolution.

In short, it's good to have the option to use both. You can't reasonably demand that Newtek implement the abilities of the most advanced renderer into a medium range animation app. So, with the profit from one project, you just buy that renderer. Simple :hey:

IMI
06-13-2009, 05:27 AM
I can rig a realistic light set up in about 5 minutes. After an hour or so of tweaking, it's usually there.



And he CAN, too.
Makes me sick just thinking about it. ;)

Iain
06-13-2009, 05:49 AM
Ack it looks a bit like I was boasting.

I just meant that once you have a light rig that works for interiors, you can easily set it up and adapt it to any room.

Exception
06-13-2009, 06:14 AM
I agree with Iain on this.
It's grea to have quick setup time, however you trade it off with:

- slow rendering (starts to matter for animations)
- little creative control

In LW, you can make it look like anything, even if it's totally unrealistic. In Maxwell, you are much more limited. This matters depending on what you're trying to achieve. For some it's irrelevant, for some it's paramount.

Mr Rid
06-13-2009, 03:25 PM
I find that most artists overthink photoreal lighting- area lights and radiosity (never needed more than 2 bounces and 36 rays). I find HDRIs overrated. It's the surfacing part that most artists dont understand. Mainly, you have to incorporate fresnel grads on all your surfaces. But I also would not have boasted LW or Kray over Maxwell or Vray. Although Maxwell sounds impractically slow for much more than stills.

Everyone should understand fresnel-
http://www.box.net/shared/static/6hauabokvz.pdf

There was a good thread here somewhere awhile back where people were trying to duplicate lighting in Vray and Maxwell renders using a standard scene... there was a name for it- the blank room with some spheres and a teapot or something.

Mr Rid
06-13-2009, 03:41 PM
I...
Everyone should understand fresnel-
http://www.box.net/shared/static/6hauabokvz.pdf

P.S. LW's incidence angle values are backwards and you have to think inverted. Would love to know who the hell is responsible for that major screwup.

toby
06-13-2009, 05:18 PM
I find that most artists overthink photoreal lighting- area lights and radiosity (never needed more than 2 bounces and 36 rays).
Agreed, sortof. I think doing arch interiors with 2 bounces or less just doesn't look right. But just about everything else, yea.


Mainly, you have to incorporate fresnel grads on all your surfaces.
*Very* interesting. You mean every channel of every surface (within reason)? I'll have to try that. Do you use the 'energy-conserving' method with that too?

Thanks for the link.

Mr Rid
06-13-2009, 06:02 PM
Agreed, sortof. I think doing arch interiors with 2 bounces or less just doesn't look right. But just about everything else, yea.


*Very* interesting. You mean every channel of every surface (within reason)? I'll have to try that. Do you use the 'energy-conserving' method with that too?

Thanks for the link.

When going for photoreal, I apply fresnel grads in every channel used. Obviously if your surface has no luminosity, reflection or translucency then you dont need to apply textures there. In a hurry, I may throw on fast fresnel and check the appropriate channels. But usually I want control over the grad.

You have to guess at fresnel grads in LW. On the Matrix sequels they used a super expensive reflectomoter to acquire real world incidence values.

brdf-
http://www.box.net/shared/static/7dz0t26mas.pdf

anisotropy-
http://www.box.net/shared/static/ybfdl7nf1l.pdf

grn
06-14-2009, 12:42 AM
I can't get the occlusion detailed in 9.6. As a relative example, there is no occlusion under filmDesigner's stereos in the render that he posted.
http://www.newtek.com/forums/attachment.php?attachmentid=73407&d=1242360525

The Monte Carlo has been useless to me since 9.5. It seems very different compared to ealier versions of LW. Gotta bake game objects in 9.31.

Cageman
06-14-2009, 12:29 PM
I find that most artists overthink photoreal lighting- area lights and radiosity (never needed more than 2 bounces and 36 rays). I find HDRIs overrated. It's the surfacing part that most artists dont understand. Mainly, you have to incorporate fresnel grads on all your surfaces.

Everyone should understand fresnel-
http://www.box.net/shared/static/6hauabokvz.pdf

HDRIs overrated? Uhm... ok... ? I mean...that is a very controversial statement from someone who is working in the VFX-field. Having an HDRI with the correct exposure and gamma, will make it alot easier to apply a linear color workflow, and that, imho, makes ALOT of difference for photorealism and integration. :)

It would be interresting to hear more about why you think HDRIs are overrated, especially within the VFX-field.

Anyhow... what about material shaders in node-editor? They do the fresnel effect by default, if I'm not misstaken (such as conductor, dialectric etc)?

jasonwestmas
06-14-2009, 02:48 PM
HDRIs overrated? Uhm... ok... ? I mean...that is a very controversial statement from someone who is working in the VFX-field.

I'm sure using HDRI makes a difference similar to painting on Oil Ground Linen vs. Acrylic Gesso Canvas. :D

Cageman
06-14-2009, 04:25 PM
I'm sure using HDRI makes a difference similar to painting on Oil Ground Linen vs. Acrylic Gesso Canvas. :D

HDRIs are very usefull for reflections as well, not just lighting, I might add... so, unless you want to manually build and place reflective objects to replicate what was on set, one can use a single HDRI instead.

:)

toby
06-14-2009, 04:29 PM
HDRIs are very usefull for reflections as well, not just lighting, I might add... so, unless you want to manually build and place reflective objects to replicate what was on set, one can use a single HDRI instead.

:)
Absolutely, using a 24bit image for reflections should be a *crime*!

Mr Rid
06-14-2009, 04:40 PM
HDRIs overrated? Uhm... ok... ? I mean...that is a very controversial statement from someone who is working in the VFX-field. Having an HDRI with the correct exposure and gamma, will make it alot easier to apply a linear color workflow, and that, imho, makes ALOT of difference for photorealism and integration. :)

It would be interresting to hear more about why you think HDRIs are overrated, especially within the VFX-field.

Anyhow... what about material shaders in node-editor? They do the fresnel effect by default, if I'm not misstaken (such as conductor, dialectric etc)?

I was wondering when someone was going to say something. Thanks for remaining calm when I make such crazy statements (I have many).

I have learned to do things in the most straightforward way possible which oddly enough turns out to be a rather 'controversial' attitude. Some places want to complicate the hell out of pipelines and re-invent the wheel on every show where it just isnt necessary.

I understand the advantage of HDRIs, but you just dont need them or complicated color workflows for realism. If it looks right, then it is right. Am so used to working with plates with no set or camera info and where no one bothered to shoot corresponding HDRIs. I just stitch an enviro image from frames (maybe from surrounding footage to get a piece of sky and ground) and map it in a sphere I can scale to control the radisoity falloff and it works just dandy. When I play with HDRIs they just dont make a big difference in what I get without them, except with reflections.

I see CG coming out of high end pipelines like say Hulk 2 (used HDRIs) and it still looks CG-ish, or about like what could be rendered without them. I hope to show some recent examples when we get clearance.

Here's an example of an early lighting test I did on LXG for plate integration, rendered in LW 7.5 with no GI, no EXRs, no multipasses. Comped in Fusion.
74397 Only one person, the director, was able to spot
what was CG.

Cageman
06-14-2009, 05:22 PM
I was wondering when someone was going to say something. Thanks for remaining calm when I make such crazy statements (I have many).

No worries... When people say such things, they usually either go silent or forget about the thread. I was quite sure you wouldn't go that route! :)



I have learned to do things in the most straightforward way possible which oddly enough turns out to be a rather 'controversial' attitude. Some places want to complicate the hell out of pipelines and re-invent the wheel on every show where it just isnt necessary.

Aye... I can actually relate to that, since I, for some "strange reasons" use LW at work and, obviously with the help of MAGIC, can get the job done with such an inferior app. ;) (sorry... but sometimes that is how it feels like not being techsavy in Maya).



I understand the advantage of HDRIs, but you just dont need them or complicated color workflows for realism. If it looks right, then it is right.

Aye.. I was introduced to LCS workflow some time back... I think, maybe late last year, and I too, have been able to make CG integrate to backplates without it (without HDRIs as well for that matter). But, on the other hand, I can also see the benefit from streamlining the LCS workflow so that it becomes an integrated part of the pipeline. Once the template has been set, it should be something that is easily transfered from project to project, and very little has to be done by the artists manually. Granted, the template may take some time to setup.

Regarding HDRIs, ever since I've had the chance to use them, I wouldn't want to trade them with moving back to something else. But granted, as you say, sometimes (especially in the beginning of HDRI history), one didn't have access to such imagery and there were few who could shoot them. So, in a sense, the oldschool ways of solving such problems is something that few are aware of these days?



Am so used to working with plates with no set or camera info and where no one bothered to shoot corresponding HDRIs. I just stitch an enviro image from frames (maybe from surrounding footage to get a piece of sky and ground) and map it in a sphere I can scale to control the radisoity falloff and it works just dandy.

Hmm...that sounds pretty neat.. never thought about using such a sphere for GI. Good tip indeed! :)



When I play with HDRIs they just dont make a big difference in what I get without them, except with reflections.

...And...that you don't have to puzzle together your own imagebased solution. ;)


I see CG coming out of high end pipelines like say Hulk 2 (used HDRIs) and it still looks CG-ish, or about like what could be rendered without them. I hope to show some recent examples when we get clearance.

Looking forward to that!

Regarding lighting, I've also found that many people today are so confortable with GI, that they end up doing sloppy lighting / no lighting at all...instead they add a keylight and then rely on GI to do the rest.

While that is all nice and dandy, it also adds a certain type of complexity to scenes and the process of rendering those out. But since these artists clearly lacks knowledge of basic lighting (i.e the time when GI was still too expensive to use), they have a hard time comming up with solutions where their GI-solutions clearly doesn't work / adds alot of problems. Many people are becomming increasingly unaware of the concept of problemsolving (this goes for many aspects of production, I might add), or that the most complex solution is the way to do it, since it is...well...complex.

Is that something you have seen as well?




Here's an example of an early lighting test I did on LXG for plate integration, rendered in LW 7.5 with no GI, no EXRs, no multipasses. Comped in Fusion.
74397 Only one person, the director, was able to spot
what was CG.

That is a very neat example and is a perfect example of good, traditional lighting techniques. I bet that this one rendered damn fast as well? :)

btw... are you asking us what is CG in this one? :)

jasonwestmas
06-14-2009, 06:34 PM
mhmm, thanks for all this info guys, it's like getting a free subscription to a 3Dmag. Very fun and interesting. That was a good comp there Mr.Rid, very seamless. I saw the video too.

Mr Rid
06-14-2009, 07:31 PM
So, in a sense, the oldschool ways of solving such problems is something that few are aware of these days?

Yeah, the 'old school' I refer to is simply about doing things the most efficient way, saving considerable time and money. My old tricks came about from a necessity to strictly optimize on very low budget schedules that still apply with more advanced pipelines. Most artists learn sloppy habits that are considered standard method. I see incredible amounts of time and resources wasted at big houses. Theres a reason why I rarely have to work more than 40 hours a week or come in on weekends, while friends at Sony are going 12 hours a day 7 days week.



Hmm...that sounds pretty neat.. never thought about using such a sphere for GI. Good tip indeed! :)

...And...that you don't have to puzzle together your own imagebased solution. ;)

I find it essential to project the BG on geometry to control falloff distances, and adjust parts of the BG image separately (like sky from ground, or hallway from windows, etc) with the luminosity/diffuse values, rather than just dropping an image in the backdrop at one value for the whole environment. I have special primitives already set up and usually can just replace one image for each plate. It only takes few minutes to stitch something if necessary since it can be indistinct.



btw... are you asking us what is CG in this one? :)

The test image is all a miniature except for the zeppelin closest to camera. Unfortunately, worse lighting went into the final shots in the film. I was rather frustrated on that project as I had no communication with the compositors, and they did not use CG elements the way I intended. But in the sequence everything was exploding and no one really cared. I also rendered flaming chunks ripping thru the first zeppelin's fabric (burning outward from the holes) and crashing on the floor in spark explosions that were never incorporated for some unknown reason.

GraphXs
06-14-2009, 11:52 PM
Mr Rid, great information and wonderful example. How do you achive the soft shadows, did ya use a bunch of Area lights? Spots/Shadow mapping? I do agree that using GI just seems to add more time to rendering, and yes it is faster now but does require alot of tweaking to get it right. The image you posted was wonderful, any chance of ever breaking down how you achieve that quality?

When you use a sphere with a world/plate image maps on to them do you use background GI or just diffuse/ref/spec passes?

Is that sphere 100% self-illuminated?

What types of lights do you use for non-GI solutions? Distant? Area? Spot?

Thanks!:thumbsup:

glebe digital
06-15-2009, 02:25 AM
It's the surfacing part that most artists dont understand. Mainly, you have to incorporate fresnel grads on all your surfaces. But I also would not have boasted LW or Kray over Maxwell or Vray. Although Maxwell sounds impractically slow for much more than stills.

Everyone should understand fresnel-
http://www.box.net/shared/static/6hauabokvz.pdf


+1
When using Maxwell you create most surfaces based on purely physical principles, once these are correct you generally get excellent results.
This process has radically changed how I surface in LW as a result, and correct fresnel-curves on everything is a good example.

clagman
06-15-2009, 09:25 AM
As much as I hate to say it I think I'm one of the lazy ones ;o) These days computing power is relatively cheap so I find myself using more of the more advanced options for the sake of cutting down on labor expenses (i.e. using the surfacing nodes and such).

gerardstrada
06-16-2009, 05:59 PM
A linear color workflow could help a lot too - gerardstrata has one but omg it's hard to understand and implement.

Hard to understand, maybe - probably I don't express myself very well - but it's really simple to implement, mostly the linear workflow at gamma level. In simple terms:

1. Linearize your input images (FPGamma or SG_CCFilter or SG_CCNode)
2. Pick all your colors with SG_CCPicker
3. Gamma encode the output (FPGamma or SG_CCFilter or SG_CCNode)

That's all :) You'll have a properly LCS workflow at gamma level (monitor color space) which works fine for web and TV (mainly PAL).

For print or film, a LCS workflow at gamut level is more desirable.

Remember, these workflows are used to make our work simpler, with better results in less time because we don't need so much tweaking in lighting and shading. It's not to complicate things. The inverse linear workflow and the multipass linear workflow are even simpler because we don't need to change the way in which we already work (though they are not physically accurate). In this regard, the linear workflow plugins of Kray look like a really smart solution.


I find it essential to project the BG on geometry to control falloff distances, and adjust parts of the BG image separately (like sky from ground, or hallway from windows, etc) with the luminosity/diffuse values, rather than just dropping an image in the backdrop at one value for the whole environment. I have special primitives already set up and usually can just replace one image for each plate. It only takes few minutes to stitch something if necessary since it can be indistinct.

Don't know if you know it, but if I understand well, this technique that you are using for lighting reconstruction (probably for years) is called Spatial IBL. A technique published, more or less, recently :) The only difference is the usage of HDRIs instead of LDRIs.



Gerardo

toby
06-16-2009, 07:16 PM
Hard to understand, maybe - probably I don't express myself very well - but it's really simple to implement, mostly the linear workflow at gamma level. In simple terms:

1. Linearize your input images (FPGamma or SG_CCFilter or SG_CCNode)
2. Pick all your colors with SG_CCPicker
3. Gamma encode the output (FPGamma or SG_CCFilter or SG_CCNode)

I will try to test that out this week!
And it's really only 2 steps if you're compositing anyway, right?

Mr Rid
06-16-2009, 07:44 PM
....

That's all :) You'll have a properly LCS workflow at gamma level (monitor color space) which works fine for web and TV (mainly PAL).

Gerardo

Oh no. :foreheads Here I was trying to de-complicate. Dont do it Toby! :)

Ironing out the color space conundrum ( http://www.theasc.com/magazine/jan05/conundrum/page1.html AC article) between all the different 3D and 2D apps, on different platforms, on different monitors, with different LUTS, and between different facilities is far, far from simple. This topic always goes on indefinitely. I went around and around this color space mess in LW (using the CCfilter) for months with many informed heads and finally scrapped it for a much simpler pipeline that works perfectly fine.

Mr Rid
06-16-2009, 08:34 PM
Mr Rid, great information and wonderful example. How do you achive the soft shadows, did ya use a bunch of Area lights? Spots/Shadow mapping? I do agree that using GI just seems to add more time to rendering, and yes it is faster now but does require alot of tweaking to get it right. The image you posted was wonderful, any chance of ever breaking down how you achieve that quality?

When you use a sphere with a world/plate image maps on to them do you use background GI or just diffuse/ref/spec passes?

Is that sphere 100% self-illuminated?

What types of lights do you use for non-GI solutions? Distant? Area? Spot?

Thanks!:thumbsup:

I mostly use areas, Q- 3. Most scenes can be lit with 2 or 3 lights at most. For the LXG shots, I used spinny light rigs that saved considerable time in 7.5, while mimicking an area light.

The GI sphere or other geometry is usually 100% lum. I may use diffuse values instead when a CG subject is close to a GI mesh and should cast a shadow on that particular area, so the radiosity bounce is appropriately darkened. Like for instance where a CG tree intersects with the ground in a plate. I may increase the lum value for other effects like recreating a bounce card from set.

In LW 7.5 most people had tried radiosity and went, 'Aah, takes too long!' But I recall that I had to put a CG creature in the middle of a plate of a hallway. So I loaded a card and positioned it for the floor of the hallway, then I kept cloning and positioning the card in place of the walls, furniture, etc. I discovered that if I front projection mapped the plate onto these cards the radiosity lighting integrated the CG perfectly with the right temperature and falloff and all that, and the radiosity traced very fast since the luminous environment does not receive a bounce and only casts light on the subject.

I may use shadow maps with a high map size (5k) but only for low intensity fill or rim lights where there is low detail on the meshes or they are not seen too closeup. Shadow maps are not accurate enough for fine detail or realistic shadows.

A big mistake I see often is people using fill lights with no shadows on. I dont know what they are thinking. Nothing in reality casts light without a corresponding shadow. Another goofy thing is when someone bakes an occlusion pass into geomety and also renders it with radiosity. Or they combine occlusion and radiosity passes in the comp. This will only double up the occlusion unrealistically.

toby
06-16-2009, 09:29 PM
Oh no. :foreheads Here I was trying to de-complicate. Dont do it Toby! :)
Don't worry I'm very pragmatic about doing extra work ( read -> lazy as hell ) if it doesn't show a definite improvement in the image I won't go to any more trouble. And there's no compatibility issues at my home. But mostly I want to know how to do it and what to expect.

gerardstrada
06-16-2009, 09:52 PM
I will try to test that out this week!
And it's really only 2 steps if you're compositing anyway, right?

Yes. It's only two steps if we save our output images in some floating point format.



Oh no. Here I was trying to de-complicate. Dont do it Toby!

Ironing out the color space conundrum ( http://www.theasc.com/magazine/jan05...rum/page1.html AC article) between all the different 3D and 2D apps, on different platforms, on different monitors, with different LUTS, and between different facilities is far, far from simple. This topic always goes on indefinitely. I went around and around this color space mess in LW (using the CCfilter) for months with many informed heads and finally scrapped it for a much simpler pipeline that works perfectly fine.


It seems you are confusing linear workflow with color management... again :)



Gerardo

Mr Rid
06-16-2009, 10:07 PM
It seems you are confusing linear workflow with color management... again :)

Gerardo

I refuse this topic. :stop: It will just wind up going on and on. I found it is hard enough to discuss in person but just impossible in text. But you might want to put a link to that other color space thread here.

toby
06-16-2009, 10:54 PM
that other color space thread
bam!
http://www.newtek.com/forums/showthread.php?p=809998#

gerardstrada
06-17-2009, 03:57 AM
I refuse this topic. :stop: It will just wind up going on and on. I found it is hard enough to discuss in person but just impossible in text. But you might want to put a link to that other color space thread here.
In fact, it's better to put it in black and white (text). But if the way you have found works in the final result, that's what matters. Maybe you haven't noticed yet but the instinctive way you have found (eyeballing) might have a logic and a scientific background after all, like it seems has happened with your technique for Spatial IBL. I can ensure this by my own experience too. This also had happened to many people with the linear workflow and some other workflows, methods and techniques. At first, people did tend to choose desaturated colors by increasing ambient lighting and tweaking their shaders to compensate the lack of gamma correction. Later, they began to gamma correct the output and adjusting roughly the appropriate colors by sight (this is well-known as the perceptual method). Nowadays we have tools that calculate accurate values for us automatically, and we know that realistic results are easier with a linear workflow applied to the whole pipeline - this is pretty easy for most of people at gamma level. My guess is that, at some point, the same thing is gonna happen with linear workflows at gamut level and color management for CG work.

--- o ---

Don't want to get off-topic but for people interested: a basic linear workflow at gamma level is pretty simple (the 2-3 steps I mentioned before), things begin to get complicated when we need more accuracy and we jump to gamut level (at least 7 steps), since this implies to take into account the color flow - color management (CM) and many people get lost here. And people tend to get lost there because color flow is in dependence of the image devices and systems we use in production. seeing that in every project something changes, every project needs a different scheme for color management. This happens just because we have used a different photo camera, or a different digital camera, or a different video camera, or a different film stock, or a different print, or a different scanner, or a different laboratory, or a different projector, or a different platform, or a different DI system, or a different output medium, etc. However, the principle is the same for any of them.

At this point, SG_CCTools are designed to manage this color flow within Lightwave. There's no other 3D package at this moment that offers this kind of facilities at gamut level. And these are free tools. However its features at gamut level are designed for people whom understand these principles and know what they are doing.

In this context, to simplify things for studio environments, it could be a general CM window that manages CC modules and automates the setup (better if convertions are hidden for CG artists) and a person (a DT or Supervisor) who design the color flow for these modules so that the CG artist just fill them with the indicated data, and it be applied globally. I think the incoming support for EXIF metadata and the already announced Python compatibility, could help a lot for this purpose.


Or they combine occlusion and radiosity passes in the comp. This will only double up the occlusion unrealistically
Yes, the only way that I've found to do this correctly is by separating the color bleed pass from the RO pass in post, or with the so useful DP_Filter Node Editors, and mixing it with a new RO pass (from an IBL rig) or an AO pass.



Gerardo

ingo
06-18-2009, 02:02 AM
.....Another goofy thing is when someone bakes an occlusion pass into geomety and also renders it with radiosity. Or they combine occlusion and radiosity passes in the comp. This will only double up the occlusion unrealistically.

Well normally you should use the occlusion pass as an alpha channel to paint dirt in that areas, means not only darken but also color variations or textures.

Good ol' days of slow raydiosity, i have found an old tutorial about Cameramapping an ambient occlusion pass on my website, just 6 years old (is that old school now ?) http://im-graphics.com/ptuto/p/tuto-e.html