PDA

View Full Version : The most difficult surface to simulate...



IMI
05-15-2007, 04:46 PM
The most difficult surface to simulate...

I say it's glass (with skin being a very close second).

<rant>

I couldn't add up all the time I've spent messing about in LW over the past few years trying to get good, passable glass. There have been a few times I thought I had it pretty good - you know, after staring at my screen for an hour or so, continually tweaking in the surface editor, moving the light, changing the refraction and reflection, caustics, double-sided, non-double-sided....bump and let's not even get into how the teeniest bit off the specularity is can turn your glass into plastic...

And then when I finally think it looks good - these mental images of a million glass visual memories distorted by what my brain is telling me looks "right" on my screen, I get up and go into the kitchen...

And there, on the countertop, left to irtself all day from the morning, 12 hours previous, is a *real* glass which at one point had orange juice in it. The sunlight through the window is causing real caustics as it passes through not only the glass, but the ever-so-thin film of dried orange juice. And it has smears on it; fingerprints; imperfections within it. Looking at it at just the right angle, there is a faint hint of the spectrum of light.

In spite of the fact that it needs to be washed, it really is a beautiful thing.

I return to my 'puter room and look at the glass on my screen, and it most definitely is *not* a beautiful thing; it's 'flaws' are overdone, it looks like it was never even touched by anyone, let alone used for anything, and there is no spectrum... it's just.not.right.

Why must glass torment me so?!?! I know it can be done - I've seen it done, and done exceedingly well!
</rant>

I suspect this is why there are so few glass monsters and spaceships in CG. :D

Matt
05-15-2007, 06:07 PM
Skin is definately a holy grail of CG, because as humans we are so familiar with it and can detect the slightest differences from reality. In fact any surface with sub-surface properties, wax, porcelain to name a few others are difficult to acheive.

Out of the manmade materials, a good carbon fibre surface is pretty difficult, and I've never managed to achieve a decent spark texture finish on metal (think of the specular finish on a silver iPod here) anti-aliasing always seems to smooth it out if you're not careful. Glass is reasonably simple I would say, as is shiney chrome. The trick with both of these is what the material reflects.

Bytehawk
05-15-2007, 06:08 PM
naaaah, just slap dielectric on there :)

hardest for me : industrial metals. (aluminium tanks, rusty frosted steel,...)

mattclary
05-15-2007, 08:10 PM
Glass is pretty easy IMO. I definitely think it would have to be skin.

Andyjaggy
05-15-2007, 10:30 PM
Car paint. Especially the metallic paint where you can see the little flecks of sparkles in it.

Matt
05-16-2007, 03:40 AM
Although I seen example node flows, a car paint material shader would be awesome, especially if it was faaaaast! How about it NT?

Captain Obvious
05-16-2007, 03:46 AM
Glass is a piece of cake. Then again, I never render glass where I would need caustics, so... Heck, most of the time I don't even use refractions on it, and it still looks great.

The worst materials to simulate are the ones with small details, like sparkles or small bumps or something, when viewed from a distance. The client goes "this surface is supposed to be sparkly!" and I go "it is, but since we're seeing it from half a kilometer away, the sparkles kind of blend together" and the client goes "make it sparkly!"

Gah. :(

IMI
05-16-2007, 04:09 AM
Well I must be missing something then, because glass has always given me trouble.

Dielectric, I haven't messed with much, as I'm just beginning to come to grips with the node editor.
But I'm not talking about refraction, reflection and specularity - I was meaning more about the fact that your average glass that's seen any kind of use at all is littered with imperfections: smudges, tiny cracks, maybe air bubbles within.
Rarely in real life do you see perfectly clear, clean glass.

Here's another question though: If you had a prism and wanted to do a render where the light shining through it was casting a spectrum on a white surface, how would you go about it?

Iain
05-16-2007, 04:27 AM
When I was tinkering with 9.2 beta, I tried to get a prism to disperse light properly using a luminous polgon with the dielectric material but I just couldn't get anywhere.

There's a render in the Maxwell manual which shows this setup but I don't know if it's fudged (probably not).

For me, realistic fabrics are difficult.
Glass, as the Cap'n said, is easy if you're not going for those close up caustic overload shots VRay spits out.

Captain Obvious
05-16-2007, 04:38 AM
But I'm not talking about refraction, reflection and specularity - I was meaning more about the fact that your average glass that's seen any kind of use at all is littered with imperfections: smudges, tiny cracks, maybe air bubbles within.
Rarely in real life do you see perfectly clear, clean glass.
Ahh, yes, that is quite hard, isn't it? Bubbles are easy enough (just model them), but getting realistic-looking smudges, cracks and scratches can be quite the challenge. I normally deal with glass from a hundred meters away on the side of a big building, so it's completely different.

Matt
05-16-2007, 06:22 AM
The worst materials to simulate are the ones with small details, like sparkles or small bumps or something, when viewed from a distance. The client goes "this surface is supposed to be sparkly!" and I go "it is, but since we're seeing it from half a kilometer away, the sparkles kind of blend together" and the client goes "make it sparkly!" Gah. :(

Copy cat, that's what I said! ;)

It's AA that wipes it out!

Gary Wales
05-16-2007, 07:10 AM
Exactly what happens with my brushed steel surfaces. Nightmare trying to light it to make it look realistic and the AA wipes out any effect unless in extreme close up.

Weepul
05-16-2007, 07:51 AM
Glass isn't that hard - you can even get 90% of the way to beaten glass with some good grunge maps.

Skin's pretty hard. Skin for full heads/bodies is even harder because it's so dependent on convincing textures, which will generally be very model-specific. I haven't even tried - organics aren't my strong suit. At least, not yet. ;D

Car paint is pretty easy to do for distant shots. For close-ups it gets harder, but I think it's in the realm of LW's abilities. Just tonight, having seen this thread, I made a pretty decent metal flake surface, using IWF2 Nodal. (Download here (http://homepage.mac.com/weepul/metalflake_surface_IFW2.zip).)

Now, you know what's hard? Volumetric anisotropic surfaces like tiger's eye (http://en.wikipedia.org/wiki/Tiger's_eye), or various decorative plastics I've seen. I've even seen it in certain gel substances like hand soaps. I'd like to see those done in LW! ;)

Weepul
05-16-2007, 08:07 AM
Accursed 5 minute edit limit: slightly updated the metal flake material to include specularity.

Andyjaggy
05-16-2007, 08:19 AM
Looks like a cool material. I just recently discovered the IFW textures. Really cool stuff. I like the kitty :)

colkai
05-16-2007, 08:28 AM
purty :D - I got me IFW2 Nodal so have to play with this - thanks Weepul. :)

Weepul
05-16-2007, 08:31 AM
You're quite welcome, but beware...it's definitely not the most intuitive node tree! ;D I clustered most of the control nodes near the end surface input node thingy - gradient for changing color based on incidence (make both keys the same to have solid color), diffuse value (in the Lambert Shader node), metallic coat reflectance, and the flake size.

riderx
05-17-2007, 10:27 PM
the most difficult surface for me to make is...the surface that was on the rock monster on the LW 9 modeler book... i still dont get that kinda surface even i tried using APS.. maybe im just too newbie here cause im using LW just for 4 month now..lol :D

anyone got any good trick to do that kinda rock monster surface without modifying the poly? (i meant.. just using surface edittor n node stuff)

Puguglybonehead
05-18-2007, 12:34 AM
Human skin has to be the toughest. Curly maple wood grain (like on the top of Jimmy Page's Les Paul guitar) seems pretty impossible to do. It's sort of a 3D illusion in itself, that changes its look with light incidence. Photo textures only look OK (sort of) in stills. Never seen curly maple done convincingly in 3D.

colkai
05-18-2007, 02:31 AM
Yup. for wood like that, the only answer is to resort to good texture maps.

I do love a nice bit of woodgrain on a guitar. :)

F1Racer
05-18-2007, 02:40 AM
Talking of car paint, what about pearlescent white ? Is that acheivable ?

IMI
05-18-2007, 03:28 AM
Yup. for wood like that, the only answer is to resort to good texture maps.

I do love a nice bit of woodgrain on a guitar. :)

Lightwave's good old standard surface editor can do some decent wood grain, by layering procedurals and then using the Surface Baker shader to make an actual texture map for further painting on or editing, and quicker rendering.
With a good deal of tweaking of the layers of the procedurals, you can get some really nice effects for the textures, spec, diffuse and bump channels.
I've made a few wood textures this way which I thought turned out pretty well. (haven't had much luck with this method and glass though)
In any event, it's a helluva lot of fun making textures that way. :D

colkai
05-18-2007, 03:55 AM
Talking of car paint, what about pearlescent white ? Is that acheivable ?

Dave Jerrard did a lovely pearl surface in his 5.6 Applied book, don't see any reason it couldn't be revised to work via nodes in 9.2. Note, this is 5.6 object and uses 5.6 shaders

mattclary
05-18-2007, 05:17 AM
This is the surface that I am proudest of, pretty similar to car paint.


http://www.newtek.com/forums/attachment.php?attachmentid=13802&d=1091990151

colkai
05-18-2007, 05:48 AM
Yup, no two ways about it, dats purdy. :)

F1Racer
05-18-2007, 05:59 AM
Dave Jerrard did a lovely pearl surface in his 5.6 Applied book, don't see any reason it couldn't be revised to work via nodes in 9.2. Note, this is 5.6 object and uses 5.6 shaders

Oh, I have that book. I must have miseed that one. I`ll skim through that tonight and check it out. Thanks.

riderx
05-18-2007, 06:05 AM
http://www.youtube.com/watch?v=c4ji3xrdczA this is what i think the closes that i can do for mimicking the rock monster surface (LW9 modeler book cover)

any comment or helpful thoughts?

colkai
05-18-2007, 07:06 AM
Oh, I have that book. I must have miseed that one. I`ll skim through that tonight and check it out. Thanks.
The whole 'gems' section is worth a second look. :)

Exception
05-18-2007, 07:39 AM
I have a car paint metal flake node setup here for you guys:
http://www.except.nl/lightwave/examplescenes.htm

I was impressed way back with Dave's anisotropic cafe table top material with all the little circles. I found that one a real challenge.

flakester
05-18-2007, 08:27 AM
Howdy all.

One of the hardest materials I've tinkered with in the past is realistic (or as close to as is possible) MILK! <-doesn't help that it's normally contained in glass vessels!

Very tough. :stumped:

I'll wait for our present project to finish, then I can start learning nodes in 9.2 a proper, then maybe I can get me some good milk.

flakester.

Bytehawk
05-18-2007, 08:40 AM
milk : dielectric with some adjusting of the thickness

you could control the absorption of dielectric with a custom node based on thickness

flakester
05-18-2007, 08:55 AM
Superb! Cheers for the advice.

I will give that a go when I have a chance to :D

flakester.

Thomas M.
05-18-2007, 09:22 AM
So far fish is pretty difficult to texture. A lot of things are going on on this little animal's skin...

Reflection, pearlescent stuff, SSS, translucency, etc...

Glass? Ha, ha, ha, ha, ....

Thomas M.
05-18-2007, 09:32 AM
Dielectric doesn't have any SSS. That's what makes milk special. Go for a mixture of Dielectric and Sigma, as Sigma alone doesn't have any real transparency.

flakester
05-18-2007, 10:53 AM
Cool :thumbsup:

May be a week or so before I get to try these techniques, but consider them noted!!

flakester.

Elmar Moelzer
05-18-2007, 12:05 PM
The surface that I had the most troubles with was asphalt for a road.
It was for a car headlight simulator (for rapid prototyping) that we did based on LW, so the surface had to be simulated and had to behave realistically when car headlights fell on it. The resulting renderings were then evaluated, etc.
Now you might say thats easy, but it really is not. The asphalt, especially when it is not fresh is rough and it almost has reflective properties to it. So light is actually reflected in all directions, even back into the direction of the virtual driver- camera (and the headlights).
The whole thing had to work from all sorts of camera angles too. Had to do the same thing for a wett road as well...
This was done with LightWave 7.0. I think it would be a lot easier these days, with the Nodal surfacing etc.
Was a lot of fun to do though.
CU
Elmar

Riff_Masteroff
05-18-2007, 11:17 PM
Yes, concrete has always been a difficult for me. On a large expanse, it has huge variations. Most ppl ignore real life concrete, therefore an approximation is easy to get in lw. But I have not been able to create a surface that is, to me, believable.

Riff Masteroff

jin choung
05-18-2007, 11:35 PM
the vajayjay?

(personally, i think it would have been funnier if i wrote "the ******?" but it asterisked me out.... doh!)

jin

jin choung
05-18-2007, 11:40 PM
sonofa... now i realize i'll never be able to properly set-up dazzling joke punchlines like, "hello, it's called a speculum!"... a pity.

jin

IMI
05-19-2007, 12:45 AM
Yeah, life's a ***** that way. :D

Dave Jerrard
05-22-2007, 09:32 PM
Sponge-like surfaces and retro-reflective materials, like roadsigns.

Sponge, including bread, cake, foam, etc. is not exactly an easy thing to do well, unless you have a lot of patience for building all those little bubbles, or would like to wait for some HV solution to render.


Retro-reflective surfaces like road signs, road markings, reflective tape, and reflectors are tricky as well. Reflectors can be done if you want to model them, building lots of tiny cube corners into a surface. Reflective tape on the other hand is another matter. These materials reflect light most strongly back in the direction from which it came with little change in intensity due to incidence angle.


He Who Hasn't Found A Convenient Way Of Doing These Yet.

Steamthrower
05-22-2007, 09:49 PM
Believable concrete/cement floors have always been difficult for me to achieve. I've found that though image maps provide the most realistic look, they're hard to use on large surfaces. Therefore I usually use procedurals or nodes.

jameswillmott
05-22-2007, 10:46 PM
Yes, concrete has always been a difficult for me. On a large expanse, it has huge variations. Most ppl ignore real life concrete, therefore an approximation is easy to get in lw. But I have not been able to create a surface that is, to me, believable.

Riff Masteroff

Problem with concrete is it never one monolithic mass. It's either cast in segments, cast with joints, or it just cracks like crazy all over... add to that salts leaching out, stains around the cracks and joints... not simple...

HanJobSoSlow
05-22-2007, 11:06 PM
I find nipples are quite hard

voriax
05-22-2007, 11:22 PM
I find nipples are quite hard

Depends on the temperature..

colkai
05-23-2007, 02:52 AM
Ba-dum Tish! :D

starbase1
05-23-2007, 04:47 AM
I think that good glass is not that difficult - but GREAT glass is extremely difficult. The new 9.2 shaders look to be a massive step forwards though, and sometime soon I intend to spend a lot of time trying to get truly great glass.

There's a degree to which this is also true with real glass. If you go to a gallery where they sell works of art in glass, the best pieces have a very special quality which I find very difficult to describe. Or in a different direction I have seen some astonishingly beautiful Romanian glassware with a smoky quality to it that it amazing.

While I understand what people say about human skin, I think if you apply the same level of examination to glass, it's every bit as difficult to get right.

I suggest trying to model several different types of good quality glass in the same shot if you don't beleive me, so that you can see the differences!

Nick

pooby
05-23-2007, 07:08 AM
I'm sure in terms of accuracy the difficulty may be comparable, but the fact is that people don't have a large proportion of their brain dedicated to recognising glass as they do other humans.
Most viewers would just accept glass as glass as long as it's pretty good. Whereas pretty good skin would merely come across as a brave attempt that hasn't quite worked.
I do see your point though, but any 'special' surface is difficult.. There are plastics that are hard to achieve too, even though plastic in general is straightforward.

Stooch
05-23-2007, 09:26 AM
the membrane from the M-Theory.

Elmar Moelzer
05-23-2007, 10:48 AM
Yeah Dave, had troubles with reflectors and roadsigns in that project as well.
I cant quite remember anymore how I did them (its been a few years since then), but I think I got them rather well, if memory serves me.
I also had to get them to glow whenever they were hit by a light, but then the customer decided, he did not want the glow on them, oh well.
CU
Elmar

starbase1
05-23-2007, 11:59 AM
Retro-reflective surfaces like road signs, road markings, reflective tape, and reflectors are tricky as well. Reflectors can be done if you want to model them, building lots of tiny cube corners into a surface. Reflective tape on the other hand is another matter. These materials reflect light most strongly back in the direction from which it came with little change in intensity due to incidence angle.


He Who Hasn't Found A Convenient Way Of Doing These Yet.

Or the Moon is another example...

Fortunately the incidence angle of planets doesn't change very fast in normal circumstances!

But surely you can use diffuse sharpness combined with a a gradient on the diffuse keyed to light incidence?

Nick

Stooch
05-24-2007, 02:00 AM
and throw a bit of corona or bloom to really make it bling.

*Pete*
05-24-2007, 02:08 AM
Snow...has to be one of the most difficult surfaces, atleast for me.

@NiM8R
10-23-2008, 11:55 AM
Does anyone remember a plugin from about 10 years back called "G-Thick"? I seem to recall someone named Graham offering it up for free. I used it with LW v5.5 - and it WORKED GREAT! - Marcus

colkai
10-24-2008, 02:20 AM
There are plenty of 'thickner' plugins out there, try flay.com and search the modeller plugins. :)

Maxx
10-24-2008, 05:34 AM
Does anyone remember a plugin from about 10 years back called "G-Thick"? I seem to recall someone named Graham offering it up for free. I used it with LW v5.5 - and it WORKED GREAT! - Marcus
I have no idea if it still works with 9.5, but Graham's GThick2 is here (http://www.happy-digital.com/freebies/gthick.html). Although with the Node editor in LW now, I'm not sure you'd need it - it can be replaced with the native SSS and Dielectric shaders.