PDA

View Full Version : Requests for Improvements to GI Rendering



Yog
07-22-2003, 12:19 PM
As Lee Stranaham has indicated there are no significant changes to the renderer in LW8, but also indicated that it might be a priority for future releases, I thought I would post my reasons for moving to 3DS MAX for my GI rendering, as it might give Newtek an idea in which to take future versions.
BTW – Remember when as Lightwave users we used to laugh at MAX’s renderer ?

Well mainly just the GI renders, but there is a good reason for this.
I mainly use GI rendering for my architectural work, where it has probably more impact than in any other field of CG. Unfortunately for architectural work Lightwave’s GI rendering isn’t very good and is definitely too slow when compared to competitors products.

The renderer of choice for me is Vray from The Chaos Group which runs on 3DS MAX (although I also own and use Final Render, also for MAX). Here are some of the reasons why I use Vray and not Lightwave for GI renders.

Splotch/speckle free renders at low settings when using HDRI illumination only.

Various GI smoothing algorisms that not only work far better than Lightwave’s one trick pony (shading noise reduction), but also doesn’t smooth out the bump mapping as LW does.

Here’s one that doesn’t seem too important at first, but is massively useful in use. The ability to see GI sample distribution during render calculation. Add to this the ability to concentrate samples based on colour, distance, or polygon-normal thresholds (that you set) and you have the ability to concentrate samples mainly where they are needed, massively speeding up renders. As LW gives no controls for sample skewing I can only assume that it takes a blanket approach to sample distribution.

As well as the monte-carlo (view dependant) GI rendering system like LW, Vray also offers view independent photon mapping, which can be used for first and second light bounces, first light bounce only, or second light bounce only.

The render engine is smart enough that on proceeding frames of an animation it only calculates for the areas of objects that weren’t visible on proceeding frames, building up a more complete solution as it goes.
Think rendering every 20th frame, then frames 2-19, 21-29, 31-39, etc render with no additional GI calculation. Works best on camera only animation, but this is ideal for architectural visualisation.

The ability to store GI solutions out to disk, meaning they can be used by all machines in a render farm.

The ability to add to GI solutions already saved out to disk.

Because Vray renders in passes (you determine how many passes), you can assess overall lighting conditions very quickly without full renders, i.e. Pass 1 is very blocky but very fast and gives a good lighting assessment, pass 2 is less blocky, slightly slower and gives a bit better idea of lighting, etc. Think of Worley’s G2 preview window and how that refreshes in stages. Don’t like the way the feel of the overall lighting is going, no problem, cancel mid render, tweak and re-render.

GI calculation is independent of anti-aliasing and motion blur, again greatly speeding up renders.

A texture / lighting baker that adds no discernable time to a render. Think minutes, not days.

True pixel displacement. OK, not a GI function, but pretty cool to be able to create light casting/reflecting grass on a single flat plane with only a fractal noise texture.

Sky portals. They are like a kind of area light you place in window openings, but instead of casting their own light (they can if you want), they just intensify the ambient light (inc HDR environmental textures) that enters the room. Maybe not technically correct, but they do help to better simulate what the human eye sees.

Whereas additional light bounces in LW seem to take exponentially longer for each bounce, Vray’s extra light bounces seem to take a fraction of the time that the initial light bounce takes. Ten light bounces ? no problem, you’ll hardly notice. You can also set the number of samples cast from proceeding light bounces to a different number cast from the first light bounce, which helps a tremendous amount.
Again I wouldn’t like to say how technically accurate it is, but it looks great.

And finally it’s blindingly fast.

Vray does have a brute force GI calculation mode, and at a glance I would say it is comparable in speed to LW’s. This leads me to believe that the massive speed increases in Vray are a result of being able to tailor the calculation to the scene, the various smoothing modes (that don’t loose important detail), and the way the solution is stored and added to.
All my modelling is done in Lightwave, and for non-GI renders all my rendering is still done in Lightwave (G2 rules and there is no equivalent for MAX), but for GI Lightwave is still way too slow and the settings need to be cranked right up to avoid splotches in certain situations (HDR illumination, internal scenes).

It would be great if Lightwave would incorporate some of the things I have mentioned, but it would be an even bigger step if Lightwave would finally allow 3rd party renderers to be used. Chaos Group are developing a version of Vray for Maya, and I would love a LW compatible version so I could go back to doing all my 3D work in one program.

jr_sunshine
07-22-2003, 01:47 PM
I currently have little use for all the GI features you mention. I expect that as my 3D work increases, I will have need for quality GI rendering. I will take all of your arguments at face.

The one thing not mentioned in all of the threads I have read recently about what's in LW 8 and what's not in LW 8 is the price. You mention 3DS MAX and VRay and Mental ray. If you need three apps to get what you want, what did it cost you to get it?

Compare that cost to cost of entry for LW. For lontime users, the renderer is probably old hat and certainly needs a boost. But for the new LW convert, the price of entry is great and the best for the money. Or am I wrong?

v/r,

sailor
07-22-2003, 03:49 PM
Maya 5 Complete is 1999 $ and includes Mental ray plus Maya's native renderer.....i think price is not an argument anymore :)
for 400 $ extra u can have Nurbs, Hierrachical subdivisions including edge weighting, history and rigid dynamics ...u shouldnt have mentioned price i'm afraid :D

organzero
07-22-2003, 04:16 PM
Sailor where're you getting Unlimited for $2400? The price is $6999 on the alias site...

Unless you mean MAX, but I wasn't aware MAX comes with Mental Ray?

TyVole
07-22-2003, 04:17 PM
I agree, the price argument is getting very lame.

A cheap solution would be for Newtek to create interfaces to third-party renderers. That way Yog and everyone else could have the best of both worlds. Newtek could even sell Mental Ray (or whatever) licenses to those who need them.

Being a complete solution, LightWave can't possibly offer solutions comparable to software that is focused on specific aspects of 3d such as rendering or CA. The best it can do is offer ways to interface with these packages for those who need the extra power.

Even Blender offers such 3rd-party rendering interfaces.

TyVole
07-22-2003, 04:19 PM
Originally posted by organzero
Sailor where're you getting Unlimited for $2400? The price is $6999 on the alias site...

Unless you mean MAX, but I wasn't aware MAX comes with Mental Ray?

Sailor was clearly refering to Maya Complete, which is $1,999.

trick
07-22-2003, 04:19 PM
If LW's renderer is becoming antique, why is there no 3rd party renderer. From what I know talking to 3rd party developers the SDK does not provide the tools to develop a good renderer.

So how about a better, more open SDK ??

Currently, people that do not need GI, certainly do not need a full fledged 3D application, or they are antique too. It wont take long and GI is a very, very basic feature. It will be the artist with specific skills that can take an application to extreme heights: so it's to the appliction to provide him ALL, not just SOME, certainly not ONLY THE BASIC, but ALL the tools he needs. I like LW not only for it's tools but also for it's creative environment. There are ALWAYS possibilities I can artistically create with LW, to satisfy my customers, even if MAX, XSI, C4D, MAYA or whatever offers the golden eggs. The only thing now is....I'm losing confidence...and that is a bad thing. I don't have a personal relationship with Newtek. I more have a relationship with my customers or even the LW community. Newtek just needs to provide the toolset...and the sooner the better...

Yog
07-22-2003, 04:20 PM
Jr_Sunshine.
Yes you are correct, for the price LW is a real bargin, I said as much in another thread not more than a couple hours ago.
And despite not having any significant updates to the two parts of Lightwave that I use most (Modeler and the renderer), I'm overjoyed that I pre-ordered my upgrade to LW8 because I was already thinking of getting DXF+ anyway, so now I get it for a third of the price, with a free upgrade to LW8 as well.

And I'll be the first to agree that 3DS MAX is rediculusly over priced for what it is. I really don't like the marketing attitude of Discreet (initial cost, upgrade costs, loss of upgrade ability, subscription memberships, etc). Although I should point out that ver.6 of MAX, due out any time, will come with a full version of Mental Ray.

But these are not the real costs involved in software.
The real costs come if I want to use GI lighting in an architectural walkthrough, I can either let it render for several days (or even weeks) with Lightwave, or I can let it render overnight with MAX+Vray.
It doesn't take many such jobs for the second software package to pay for it's self, especially when you consider the time cost of test renders.

In fact for me 3DS MAX is JUST a pass through application for me to render Lightwave models in Vray, because over the course of a year it is a far cheeper way of working for me.

TyVole
07-22-2003, 04:23 PM
I don't think they need to improve the SDK to create interfaces to third-party renderers. In fact, there are already some RIB exporters.

sailor
07-22-2003, 04:27 PM
Organzero :

yes my sentence wasnt clear i'm afraid....Maya Complete is 1999$

(i meant 400$ more than LW)

for that price u have Mental ray, Maya native render, Nurbs, hierarchical subdivs edge weighting, history, soft/rigid dynamics and more....so price isnt the argument again :)

trick
07-22-2003, 04:28 PM
Originally posted by TyVole
I don't think they need to improve the SDK to create interfaces to third-party renderers. In fact, there are already some RIB exporters.

O yes, you can put LW in a RIB pipeline, but that is CERTAINLY not the same as a 3rd party rendered that complies with everything you do inside LW (and ALL plugins) !!!!!

GruvSyco
07-22-2003, 04:32 PM
For what its worth, Newtek when Newtek had posted job openings for programmers a while back, they did mention something about wanting experience with Renderman or RIB. If I had to make a guess from that, Newtek are probably looking pretty seriously at having external rendering capabilities.

jr_sunshine
07-22-2003, 05:09 PM
Ok guys. I'm not going to go crazy on the cost argument, but you all are simply casting it aside like it doesn't matter. I am asuming that you guys never have a problem paying the typical upgrade of about $400...right? It's just $400 dollars.

Look, my point is that LW is a great value. Every tool has its big uglies about it. Is Maya perfect? Is MAX perfect? XSI? Software disagreements border on religious battles. I hate After Effects, but say that on a forum and you'll get the entire FX community on you.

LW ... it works ...it fits the budget ....it has continuing value.

I like it.

:)

bobakabob
07-22-2003, 07:34 PM
Well said, JR! The bottom line is that price is a crucial factor for many users and certainly here in the UK, LW has built a solid reputation for superb quality and value... for pros, semi pros, students and serious hobbyists.

Maya would never had lowered its ridiculously inflated price were it not for Newtek's forward thinking, audacious and canny marketing strategy - a delicate balancing act in the face of fierce competition. In the UK Maya Complete is 50% more expensive than LW... sorry, but that's a lot of dough, especially for an (in)Complete app.

Yep, Maya's features are superb, and LW obviously has its competition cut out, but for many users (pros included) buying into a Maya, XSI or Max upgrade path is not an easy option. I've also yet to see models and animations on the galleries of competitors' sites that surpasses work produced here.

extrabyte
07-22-2003, 09:11 PM
Has anyone else seen this?

Specifications for TURTLE® demonstration version at SIGGRAPH 2003
- LIQUIDLIGHT® Raytracing & Global Illumination with caustics, color bleeding, Final Gather and Skylight
- Support for Maya shading networks
- Support for Maya light sources
- Soft shadows (all lights can be area lights)
- Integrated with the Maya GUI
- Display/render layers
- Multipixel filters
- 3D Motion Blur
- Animation and batch rendering support
- Depth-of-field
- Support for Subdivision surfaces
- Preview Mode - Interactive visualization with shading, reflections/refractions and Global Illumination
- Texture Baking with support for direct lighting and Global Illumination effects as well as shadows. Supports baking to textures and vertex colors.
- Windows NT/2000/XP version for Maya 4.5 and 5
- Linux version for Maya 4.5 and 5


- Additional Global Illumination features, such as subsurface scattering, are under development.
- Production features such as HDRI, Image Based Lighting etc will be added in coming versions.
- A MacOS X test version of TURTLE® for Maya® 4.5 and 5 will become available in Q3.


Yikes!

grannysmith
07-22-2003, 10:47 PM
Originally posted by Yog
I mainly use GI rendering for my architectural work

So you only render interior scenes?
Because it is inefficient to use GI with outdoor scenes.
Having said that, what is your approach to rendering environmental phenomena, including volumetric cloud effects (you know, moody cumulus clouds sporting an atmospheric crepuscular ray effect, overshadowing a city block or neighbourhood?)



Unfortunately for architectural work Lightwave’s GI rendering isn’t very good and is definitely too slow when compared to competitors products.

Sky portals and indoor scenes, sure, GI and VRay.
That's interior design. But where is the architecture?

Again, have you been rendering external architectural environments with GI ?? If so, I wouldn't want to be the client paying for your rendering time.



.

novadesigns
07-22-2003, 11:37 PM
Yog, I'm not familiar with your work and this isn't specifically adressed to you, but I think using GI for anything yet is too render expensive. No matter what software you use. If you really know how to light well you don't need GI. In many cases I can virtually duplicate the look of GI and render my shots in 1/1000th the time. I see a lot of people using GI because of this perception that its an easy way to get decent looking ambient lighting--because learning to light is not easy people look for ways around that. But in most cases GI looks fake just like any other plugin used poorly!

And every GI solution has its flaws... if its not grainy shadows, its hugely innaccurate color, blown out highlights or killer rendertimes. No one solution has ever made me so impressed that I'd be willing to jump ship from LW in order to use it... aside from the fact that I work in a production environment that makes GI completely unfeasible. :)

Any trick or technique if used as is tends to look flat and boring. Even HDRI does too if it isn't combined with a good lighting technique. Once you get over the wow factor of the "next cool new thing" and you start seeing every other model rendered with radiosity on a blank gray backdrop you start to see its flaws and of course the rendertimes make you say to yourself, "why do they bother?"

It reminds me of when everyone in the graphic design field started throwing drop shadows under everything :p ;) :D

That's how GI is now.

Neil_Campbell
07-23-2003, 01:16 AM
I think using GI for anything yet is too render expensive.
I disagree - it may be too expensive in LW, but other applications are faster

If you really know how to light well you don't need GI.
Maybe, it really depends on the scene - there are times when lighting a scene the old fashioned way is the best way, and there are times when that simply won't cut it because to get a realistic lighting solution might require literally hundreds of lights and the time taken to set those up is enormous, and then with so many lights the renderer becomes so slow you should have stuck with GI in the first place :)

In many cases I can virtually duplicate the look of GI and render my shots in 1/1000th the time.
It's great that you're so skilled at lighting - I'm not - but I seriously doubt that you could get a GI-quality shot of a complex architectural scene in 1/1000th of the time

I see a lot of people using GI because of this perception that its an easy way to get decent looking ambient lighting
Agreed - and in many cases it is a good way of getting decent ambient lighting

because learning to light is not easy people look for ways around that.
That's true of many things. GI is no different from anything else in that respect

But in most cases GI looks fake just like any other plugin used poorly!
Agreed, if used poorly / inappropriately, anything will look cliched or crap, including GI. But that doesn't mean GI doesn't have it's uses.


And every GI solution has its flaws... if its not grainy shadows, its hugely innaccurate color, blown out highlights or killer rendertimes.
I think there's a bit of hyperbole in there - it's quite possible to get stunningly clean GI solutions, with accurate colour bleeding and decent highlight control, without having to wait forever for the solution - just not with LW at the moment

No one solution has ever made me so impressed that I'd be willing to jump ship from LW in order to use it...
Good for you - keep on 'wavin :)

aside from the fact that I work in a production environment that makes GI completely unfeasible.
I work in a production environment also, and for my line of work GI is absolutely essential.

Once you get over the wow factor of the "next cool new thing" and you start seeing every other model rendered with radiosity on a blank gray backdrop you start to see its flaws and of course the rendertimes make you say to yourself, "why do they bother?"
Agreed, but that's not what many of us do - we use GI as a means of lighting highly complex still scenes that require photorealism, not 'Arnold style' greyscale renders


And drop shadows are still useful for lots of things - if used discreetly and appropriately :D

trick
07-23-2003, 01:22 AM
Originally posted by novadesigns
...that makes GI completely unfeasible...

grainy shadows: only with stochastic solutions, NOT with progressive radiosity solutions which store lighting information in the vertices

hugely inaccurate color: just a user misinterpretation on the physics behind an algorythm: most radiosity or GI is based on the HSV principle and colorbleeding happens on the percentage of saturation. The difference between the color you see and the color that is bounced is physically too complex to make ONE model that solves both phenomenons. GI is not just a one button solution: you still have to experiment by trial and error. If people put a blue background behind a scene just because the sky is blue, you get completely false lighting: the sky is blue because of that what you see, it radiates a spectrum that's more towards white !!!!

blown out highlights: well, I refuse to go into that !!!!

killer rendertimes: I used Lightscape (even with LW), MAX, C4D, Brazil, Finalrender, VRay, Mental Ray: if used well you can reach acceptible results on PAL (720x576) frames, with millions+ polygon interior AND exterior models, on todays SINGLE processor (3 GHz) machines (NO swapping) UNDER 5 MINUTES !!! (in some cases there still is the time to calculate a solution, but in case of animations this is relatively small !!!)

flat and boring: only when you use the diffuse lighting that comes from GI or Rad: you still have to get your highlights from something: either lights casting speculars or HDRI !!!

The only problem I still see with many solutions is when animating elements after a solution has been calculated: but if you're a creative thinker you surely find ways !!!!

GOOD LUCK

Yog
07-23-2003, 03:28 AM
GrannySmith
Although most of my work is architectural in nature (interior and exterior scenes), it’s not exclusively so.

I’m not sure if you are saying that designing interior spaces and having an eye for the quality of light that would reach it is exclusively the role of Interior Designers, and not the role of an Architect, but if that is what you are saying then I would have to disagree with you, and so would most architects I know. Where I come from architecture is not exclusively the design of the exterior shell.

You also seem to be under the impression that rendering external scenes with GI is more computer intensive than rendering internal scenes with GI. In fact the opposite is true by several degrees of magnitude, this is even true of GI rendering in LW. I have found that exterior scenes with several hundred thousand polygons will generally render a lot faster than simple internal scenes with just a few thousand polygons.


Nova Designs
No worries, we’re all talking general terms right ? :)
Yep, I’m fairly familiar with faking interior lighting, I’ve been doing architectural visualisation part time for about 10 years, started doing wire frame models in Autocad so that more traditional visualises could use it as a base for their paintings.

It is horses for courses. If the client just wants one or two stills, then I will often just go with more traditional CG lighting techniques (low intensity spots and points with falloff, or directional lights without shadow casting), but if the client wants more, then I will often shift to GI rendering to save both me and the client time.

If on an internal image the client wants to pay for the extra quality that “can” be achieved with GI rendering, then it is a small thing with incremental addition to GI solutions to put on a 6 second, PAL Res, 90 degree camera pan of that room on overnight (150 frames, 10 light bounces, overnight on a single CPU = no problem). The upshot is that it is very little extra work for me, and the client gets a very cheep (or free if I want to impress) short animation, which always goes down well.
Do this for a few rooms in a building, and with incremental add to the GI solution you are well on the way to a walkthrough animation with not much increase.

But you are right, that if I were using GI rendering in LW, then the situations I have described would not be cost effective, and certainly none of my clients would be prepared to pay what it would actually cost.

quatermain
07-23-2003, 10:09 AM
I for one agree with yog. This needs to be addressed; if not, LW will lose what little ground they have architecture visualization specialists.

I also find that LW's radiosity is not suited for interior, multibounce situations. Not only are the rendertimes a killer, but it also doesn't look so good. Maybe it is my inexperience with it, but i have yet to see killer interior architecture images made with LW's radiosity engine (visualization purposes), although I have seen images where radiosity was used effectively as an occlusion pass on top of a raytrace light setup - which is twice the amount of work. But I do agree that while LW is a good value, usefulness rules in the end.

grannysmith
07-24-2003, 06:35 AM
I have found that exterior scenes with several hundred thousand polygons will generally render a lot faster than simple internal scenes with just a few thousand polygons

No, that wasn't my point.
Rendering environmental phenomena associated with landscape and its architecture is NOT the same as rendering a euclidean building object in the sterile light of global illumination.

Interiors are controlled illumination exercises within cartesian space. Photogenic light boxes. Lets not get too excited about the aesthetics of luminaires and global illumination.

Back to my original question:


what is your approach to rendering environmental phenomena, including volumetric cloud effects (you know, moody cumulus clouds sporting an atmospheric crepuscular ray effect, overshadowing a city block or neighbourhood?)

In other words, are you promoting VRay's GI in the aesthetic gaze of the architectural object - luminaires and all?

And don't quote architecture. When has an architect ever managed to render realistic urban forest canopy with subsurface scattering? Their desire to polygonise (or 'poly-agonize') nature is incurable. Applying GI to a streetscape with trees is lamentable. Hope you worked out the appropriate BRDF for the leaves. Your advocacy of the GI pipeline for natural environmental phenomena is misleading. You are advocating the wrong solution (GI) in relation to naturally dynamic environmental phenomena - rain, clouds, fluids, atmospherics. And your critique of LightWave consists entirely of one illumination solution. LightWave has a more intelligent approach to natural illumination than your critique suggests.

In terms of the photogenic lightbox interior scene, or the aesthetic architectural object sterilised in euclidean space by the unheimlich radiance of global illumination, yes you could be correct, other renderers produce a noumenally quick slick GI solution. But it is phenomenally dead nature.




.

grannysmith
07-24-2003, 06:46 AM
Originally posted by grannysmith
appropriate BRDF for the leaves

BDRF, the keyboard lies; and a link for anal retentives who render trees as architectural parsley garnishing buildings-as-objects.

a comparison of forest bdrf simulated by the forest reflectance (http://www.aai.ee/~nilson/alps.html)






.

jr_sunshine
07-24-2003, 06:55 AM
LOL.

This thread is so beyond me....

My brain hurts.

anieves
07-24-2003, 07:46 AM
you know, Max renderer still makes me laugh!

It makes Maxers laugh too and that's why they have to get Brazil, Vray etc. after spending 3k on a package you spend another 3k on renderers alone. I bet most Maxers use "shareware" versions of those renderers too, Yikes.

Lightwolf
07-24-2003, 08:06 AM
Actually VRay is only $800 including ten network nodes, which is pretty good considering the features and speed.
May be NT should outsource some programming to b programming to Bulgaria or Russia :D
I'd rather have NT sell the renderer as a third party add on to other packages, and focus on further development and a limit of, let's say, 5 network licenses per LW package, than a standstill in render development.
Cheers,
Mike

anieves
07-24-2003, 08:52 AM
Originally posted by Lightwolf
Actually VRay is only $800 including ten network nodes, which is pretty good considering the features and speed.
May be NT should outsource some programming to b programming to Bulgaria or Russia :D
I'd rather have NT sell the renderer as a third party add on to other packages, and focus on further development and a limit of, let's say, 5 network licenses per LW package, than a standstill in render development.
Cheers,
Mike

agree, my point is that we, lightwavers have always been able to create kick *** renders without the need of a third party renderer while Maxers have always had to look outside to get good looking renders that didn't look like cheap plastic.

That said I agree that NT should implement some sort of suport for 3rd party renderers, competition and flexibility is always welcomed.

Lightwolf
07-24-2003, 08:57 AM
anieves:
Actually, I meant it the other way around. Open up the LW renderer for other apps to use it. Let it hook into Maya, Max, whatever, make it RIB compatible. Another source of revenue that can be dedicated just to the development of the renderer (well, o.k., and to the hooks for the other apps).
Then again, currently it doesn't compete feature wise in that area (the third party render market), so it would require a lot of work.
Then again, If LW would use RIB natively, we could go out and use a RenderDrive :D
Cheers,
Mike

Exper
07-24-2003, 09:29 AM
If NT will manage well they could do both of them!

Stranahan said (http://vbulletin.newtek.com/showthread.php?s=&threadid=8044&perpage=15&pagenumber=5):
"One major goal is to greatly expand the SDK for LightWave. The current one has too many roadblocks and there's never been any good reason for that. Expect mnay changes as time goes on. We're not going to delay the release of 8.0 for it, but will expand it as time progresses and new releases come out."

They're working good and in the right way now... so we can hope for better future!

Meantime... a complete/integrated RIB exporter should please everyone who's really displeased about LW's renderer! ;)

Bye.

anieves
07-24-2003, 09:32 AM
Originally posted by Lightwolf
anieves:
Actually, I meant it the other way around. Open up the LW renderer for other apps to use it. Let it hook into Maya, Max, whatever, make it RIB compatible. Another source of revenue that can be dedicated just to the development of the renderer (well, o.k., and to the hooks for the other apps).
Then again, currently it doesn't compete feature wise in that area (the third party render market), so it would require a lot of work.
Then again, If LW would use RIB natively, we could go out and use a RenderDrive :D
Cheers,
Mike

oooh, got ya:) that would be pretty cool, but then my boss would make me use 3dsmax more... yuck

Lightwolf
07-24-2003, 09:33 AM
Exper:
I'm really looking forward to the new SDK, I've got plenty of ideas...
What I'd love to see in the distant future, would be a RenderMan complaint shading language within the surfaces panel, that can create its own gui and can get exported to RIB compliant renderers.... :)
Cheers,
Mike

Lightwolf
07-24-2003, 09:35 AM
Originally posted by anieves
oooh, got ya:) that would be pretty cool, but then my boss would make me use 3dsmax more... yuck
LOL :p But at least you'd get to work with the LW renderer :)
Hm, would you prefer Modeler as a Max plugin? ;)
Cheers,
Mike

Exper
07-24-2003, 09:44 AM
Originally posted by Lightwolf
What I'd love to see in the distant future, would be a RenderMan complaint shading language within the surfaces panel, that can create its own gui and can get exported to RIB compliant renderers.... :)
Well... exporting the whole Layout into a RIB compliant renderer shuold be a lot better... and naturally for free! :D

Bye.

Lightwolf
07-24-2003, 09:47 AM
Exper:
That's the point. But you could prototype, or even use the shaders natively from within LW :)

Exper
07-24-2003, 09:54 AM
Lightwolf:
Maybe...
"LightWave 9" should be "RenderLightManWave 1"! :D

Lightwolf
07-24-2003, 09:57 AM
Exper:
I'll be happy enough if they have decent Texture filtering in 8 ;)
...and now: we're really OT :D

Exper
07-24-2003, 10:27 AM
Some constructive things (not in a specific order):

- Faster renderer (with more in-deep settings)
- Better AA / DOF / MotionBlur
- Micropolygon renderer
- Programable shaders with a simple shading language
- Better network rendering
- Bump-maps would distort the Shadows
- Reflection & Refraction Surface based exclusion list (similar to lights)
- AutoReflection & AutoRefraction ON/OFF and Surface based
- More accurate and light sensitive Caustics
- Volumetric Caustics
- Photometric lighting
- Circular Spots (parallel and conical)
- Rectangular Spots (parallel and conical)
- Single-sided Area Light
- Deep Shadows
- Perspective Shadows
- Better Soft-Shadows

Bye.

Lightwolf
07-24-2003, 10:32 AM
May I add?

- advanced Texture filtering
- expanded SDK (more shader/procedural control, start / stop render from the SDK, SDK based light types).
- distributed network rendering of single frames and previews from within LW (or hooks for that within the SDK).
- automatic reflection / refraction mapping
- saving and loading of reflection/refraction/shadow maps
- improved filtering of image / refraction / projection maps.

..for a start ;)
Cheers.

papou
07-24-2003, 12:32 PM
May i add too?

¹· Multithreaded Surface Baker or better : network rendering Surface baker.
²· Faster AA.

cavalos
07-24-2003, 02:47 PM
Yeah, all of the above BUT VERY WELL IMPLEMENTED.

Best
Christian

cavalos
07-24-2003, 02:57 PM
Something else:

* Dispersion
* Absorption
* Sub-Surface Scattering

Hervé
07-25-2003, 02:11 AM
hey Papou, I am really with you on the first one.....:)

Lightwolf
07-25-2003, 02:14 AM
I forgot a couple:
- better multithreading across the render, including multithreading shadow map generation and scene pre-processing (transforming polygons, optimizing objects)
- better scene optimizations for raytracing (AFAIK LW still only uses a bounding box scheme).
- user defineable sample degradation for multi-bounce monte carlo radiosity (sounds cool, doesn't it :) ) Example:
Rays per Evaluation set to 4x12, Bounces to 4, degradation to 2x:
1st Bounce: 4x12
2nd Bounce: 3x9
3rd Bounce: 2x6
4th Bounce: 1x3

Rays per Evaluation set to 6x18, Bounces to 3, degradation to 4x:
1st Bounce: 6x18
2nd Bounce: 4x12
3rd Bounce: 2x6

Elmar Moelzer
07-25-2003, 02:26 AM
I can second most of the requests and want to add the following:
1. Alternative AA- methods from within the SDK (not just as imagefilters)
2. The ability to replace entire parts of the renderer from within the SDK simillar to what Cebas does with 3ds MAX and final render (they use parts of the MAX- renderer and go in where they need to).
3. Support for custom texture- projection from within the SDK (not as a procedural)
4. Lightshaders
CU
Elmar

cresshead
07-25-2003, 07:45 AM
lightwave's renderer in future could benefit from the introduction of a bucket render system whereby the renderer only needs to look at a user specified "bucket" of say 32 x 32 pixels at any one time and it can use multiple buckets..one for each cpu..

this also aids memory or scene complexity managment too.

brazil, mental ray, final renderstage 0 and vray do this already
and final render stage 1 [now shipping] extendes this to a network render where by you can put all your cpu's on a network to render on each image...so you may have 20cpu's working on a single frame at any one time...

and 20 x 3ghz makes for a fast render that's around 60,000mhz of render calcs...

steve g

Lightwolf
07-25-2003, 07:58 AM
cresshead:
I'd love to see that too.
I was actually thinking of how far one could approach that goal from within the current SDK (rendering distrbuted buckets across a network), it could probably be done actually (with a wicked combination of network aware pixel filters...).
BTW, www.realsoft.fi does bucket rendering across a network too, so does mental ray (actually mr has had that feature for ages...).
Cheers,
Mike