PDA

View Full Version : Standard galvanized garbage can?



jeric_synergy
10-12-2015, 10:04 PM
YES, because I'm lazy...


Anyone got a bog-standard galvanized garbage can with a nice, galvanized surface on their hard drives?

squarewulf
10-13-2015, 11:26 AM
130321

Made one real quick, texture isn't great but passable.

jwiede
10-13-2015, 11:49 AM
I've yet to see a really decent galvanized procedural texture in LW, which I've typically attributed to specular (esp. anisotropy) limitations in LW's current render engine. I've seen photo-based ones that look really close in diffuse, but those never seem to capture that odd, regularly-splotchy crystal-formation-derived patchiness galvanized surfaces have, with each patch having different specular and anisotropy values. I've seen some pretty decent versions done in Substance Designer, but complex specular "behaviors" also don't appear to transfer from Substances->LW that well (yet, IMO).

erikals
10-13-2015, 02:28 PM
there's one 13seconds into this video >
https://www.youtube.com/watch?v=2j02cNjk2B8

download >
http://forums.newtek.com/showthread.php?127028-60-favourite-surface-presets&p=1422153&viewfull=1#post1422153


u owe me http://forums.cgsociety.org/images/smilies/arteest.gif

MonroePoteet
10-13-2015, 04:31 PM
As jwide mentions, it's really the larger, splotchy specularity and diffuse patterns that make galvanized metal look right, IMO. I generated a "galvanized noise" pattern that shows the effect that I expect. I've attached a sample scene and sample objects. Pardon the quick, generic garbage can model!

mTp

jwiede
10-13-2015, 04:59 PM
there's one 13seconds into this video >
https://www.youtube.com/watch?v=2j02cNjk2B8

Sorry, but that preset isn't very close to a realistic galvanized metal surface w.r.t. specular look and unique patch anisotropy (or, in that case, any anisotropy), at least IMO.

Go shine a white LED flashlight on a galvanized metal watering pot, and you'll see how rich and complex the specular look of galvanized metal is in reality. It's much more complex than you seem to realize. Each little "patch" has its own unique combo of specular shading and anisotropic reflection angle, and the patches form in a kind of pseudo-Voronoi-looking pattern common to crystal growth/coverage mechanisms and yields extremely clear delineation between patches. Real galvanized metal has a very interesting and complex appearance, stemming from the crystalline coating(s) on the base metal.

jwiede
10-13-2015, 05:08 PM
As jwide mentions, it's really the larger, splotchy specularity and diffuse patterns that make galvanized metal look right, IMO. I generated a "galvanized noise" pattern that shows the effect that I expect. I've attached a sample scene and sample objects. Pardon the quick, generic garbage can model!

No worries. That's actually pretty close w.r.t. diffuse appearance, though you have a few concave-edqed 2D patch shapes that shouldn't occur, as well as a bit too much variation in scale of patches (has to do with how crystals form & spread in 2D). However, that surface also lacks the specular complexity, slight dispersion reflective effects, etc. that makes galvanized metal interesting (IMO).

Again, LW's support for anisotropy, stacked specular handling, and reflective behaviors just do not seem up to realistically replicating galvanized metal (typically, steel) procedurally.

jeric_synergy
10-13-2015, 05:17 PM
Yowza! WHAT have I started!!!??? :D :hey:

Thanks to the uploaders :bowdown: , and very interesting discussion. --I don't know why people go hiking in the mountains, when we have galvanized trash cans right here in the city. (Seriously, I have little use for nature, ironically, living in what is apparently Paradise for the nature lovers.)(Except, of course, the oxygen and water. That's nice.)

jeric_synergy
10-13-2015, 05:22 PM
130321

Made one real quick, texture isn't great but passable.

Thanks! Here's an interesting thing: I expanded both the LWO and the JPG into the /Objects directory, and LWM picked up the steel.jpg 'naturally'. Was it stored in the Objects directory?

If not, well done Devs.

squarewulf
10-13-2015, 08:54 PM
Thanks! Here's an interesting thing: I expanded both the LWO and the JPG into the /Objects directory, and LWM picked up the steel.jpg 'naturally'. Was it stored in the Objects directory?

If not, well done Devs.

Thanks!

And no, i had it in a proper content directory with separate folders. I believe LW automatically searches the local folder for missing elements before it prompts you that it can't find it.

JoePoe
10-13-2015, 09:07 PM
john.....Hard to know what's going to pass mustard here. "Galvanized" yields a (somewhat) wide variety of examples to choose from. Perhaps an image of your ideal?

Thought I'd just throw an attempt out there. Basic texture is Rman collection Cells displaced with Turb. That drives just about everything.
Blinn on Spec Shading.

130329

jeric_synergy
10-13-2015, 10:12 PM
JoePoe, looks close, but I think those crystals are a bit large. Also, there's a bit where it's QUITE colorful, NW of the handle.

jwiede
10-13-2015, 10:30 PM
john.....Hard to know what's going to pass mustard here. "Galvanized" yields a (somewhat) wide variety of examples to choose from. Perhaps an image of your ideal?

Your example's actually pretty nice, the per-plaque shading to mimic dispersion is a nice touch. Without a moving image, of course, it's difficult to tell if the individual "plaques" reflective properties vary quite the same way they do in real life, but that's a pretty darned good try for Lightwave. Any chance you'll post the responsible node network?

I've seen a couple rendered galvanized surfaces that were basically indistinguishable from reality, even in movement, but not for LW. They involved multiple distinct layers of procedurally-mapped coatings on top of steel base material where each little section had differing properties within each coating layer. Both Vray and Maxwell have ways to express coatings on base materials that make expressing scenarios like galvanized metal surprisingly direct, but even for them there was a significant render cost for multiple, detailed coating layers. I'll try to dig up the link, the surfaces stemmed from discussions on tutorial content covering how to generate layered coating surfaces for Vray.

As for a real-world example, about midway down this page (http://www.stainlesssteelblog.com/2013/05/galvanized-steel-sheet.html) there are a few decent shots of galvanized steel surfaces in close-up, showing some of the more interesting visual properties. The anisotropic reflectivity of the "plaques", IMO one of the more interesting visual properties, doesn't show too well unless you either view with motion (to see how the shifting view angle impacts the light return from the individual glazing "plaques" within the surface), or closely observe how bright light reflections still differ between similar-shaded "plaques".

For as utilitarian a surface as it is, I find it visually quite interesting.

pinkmouse
10-14-2015, 02:39 AM
..For as utilitarian a surface as it is, I find it visually quite interesting.

Indeed. I've made several attempts at a galvanized surface, and never had much luck. Might be time for another play!

pinkmouse
10-14-2015, 04:55 AM
https://www.youtube.com/watch?v=R3wAAXe_7Vs

Okay, very rough first draft. Still needs lots of work...

pinkmouse
10-14-2015, 06:45 AM
Tweaked a little more, still needs more work. I love IFW2! A fern type procedural would be really useful. :)


https://www.youtube.com/watch?v=GXGv65sGvUk&feature=youtu.be

edit: Got to go do real work now.

jwiede
10-14-2015, 12:32 PM
Pinkmouse, nice job overall, definitely getting better. It feels like the base material color is off in diffuse, it's too light-colored to be unpolished steel underneath. It's all a bit too shiny, as well (IMO) -- while a lot of what's "going on" with the detail has to do with reflectivity behavior, the actual amounts of reflectivity involved are pretty low. Your image has plaques reflecting as if polished, a bit too much reflectivity, and the reflectivity is pretty uniform at glancing angles instead of highly direction-specific.

Go actually find and touch a galvanized surface (try a garden store): The plaque's surfaces really are surprisingly rough to the touch (inside each plaque, not just at the boundaries either). They're not polished, and the overall galvanized "effect" is quite matte at distance, only when you get close and/or sharply-lit can you see the plaques have a lot more going on with them.

pinkmouse
10-14-2015, 01:14 PM
John, I'm actually modelling from my waste paper bin sat next to my desk!

However it is lacquered, hence the shiny shiny...:)

The reflection angle is a thing, at the moment I'm cheating by just rotating Ani refection and spec shaders around per plaque, so you get a bit of directionality, but not enough. I need to fake normals per plaque somehow, pondering on how to do it.

Here's another vid I left rendering earlier:


https://www.youtube.com/watch?v=4wQBJC_Kgt8&feature=youtu.be

squarewulf
10-14-2015, 01:43 PM
Rendered with octane
130332

pinkmouse
10-14-2015, 03:08 PM
Rendered with octane

Cheat! :D

squarewulf
10-14-2015, 03:19 PM
Cheat! :D


:thumbsup:

jwiede
10-14-2015, 07:32 PM
Rendered with octane
130332

Looks okay, a bit overly shiny perhaps, but viewpt is too far away to tell very much. Can we please see them a bit closer? Ideally with some camera or can movement involved? Thanks!!

JoePoe
10-17-2015, 08:23 AM
JoePoe, come on dowwwwn! You're the next contestant on The Can Is Right!!!! :D
(make sure to watch big and in HD)


https://www.youtube.com/watch?v=FLdR43xVeVE

DogBoy
10-17-2015, 09:50 AM
that's pretty nice, JoePoe

jwiede
10-18-2015, 07:43 PM
that's pretty nice, JoePoe

Agreed, that's pretty darned close. What's the per-frame render time like for that surface? And can you please show the complexity of the node network? I don't really care as much about the specific node details, as much as seeing the overall scale/complexity of network used, to help gauge render cost.

djwaterman
10-18-2015, 10:17 PM
JoePoe, indeed anyone who thinks they have cracked this nut, if your surface is properly energy conserving, please consider uploading it to www.3dxyz.pro

JoePoe
10-19-2015, 01:26 PM
Hey, glad ya'll liked it it. I'm not in front of LW at the moment but will post it later.

I can say now that the node tree was pretty messy. But I didn't put any effort into streamlining it. It had multiple Cell patterns and multiple procedurals breaking up the Blinn shader (and breaking up the cell patterns :hey:). As a result, the render was kinda expensive. I think it averaged around 4 minutes a frame (3 lowest to 5 highest during the animation). But it is a hero shot , so I didn't pull any punches.
But, fwiw, I've got LW on a fairly anemic 2.5 Ghz Quad i7 laptop. That render was done in 11.6.3. Don't know if 2015 is appreciably faster (or not). My benchmark marbles test came in at 2 hours flat if that helps.

Regarding energy conserving. Uh, how do I check that? I didn't use a Material node. I did use Reflection AND Spec Shading though.... which may take it out of "conservation" range? You will have to tell me.... I do Nodes by the seat of my pants ;D.

jeric_synergy
10-19-2015, 02:12 PM
The way you guys throw terminology around... I'll just keep pushing a broom around in the back of the lecture hall...

Suggests for resources on when to use one shader over another?

pinkmouse
10-20-2015, 03:51 AM
Had a few minutes spare, so did a less "polished" version:


https://www.youtube.com/watch?v=WktHrc2Co4c

Uses IFW2 extensively, but relatively quick to render, under 30sec per frame on average, 768x768 4/16/.01

Joe, could you post your scene so I can try my surface on your can?

Oops: Here's the node editor:

130444

It probably was EC before I used Material Tweaker...:)

pinkmouse
10-20-2015, 04:16 AM
And a full HD close-up. I could possibly turn the Normal bump down a tad.:)

130445

jwiede
10-20-2015, 01:34 PM
And a full HD close-up. I could possibly turn the Normal bump down a tad.:)

Yeah, that's quite rough, a bit extreme. Looks really good otherwise, though. ;)

I did notice one minor thing in the rotating view, at high view incidence angles the Aniso effect appears to fade a bit, and shouldn't. Is there a fresnel node or equivalent "after" the Aniso?

pinkmouse
10-20-2015, 02:26 PM
Nope, exactly as the pic in last but one post. I suspect it's an artifact of the projection limitations of the LW aniso shader.

JoePoe
10-20-2015, 05:11 PM
Well, wait a sec! Do I get to collapse my nodes so everything looks nice and tidy too? :D
(actually.... I don't think it'll help :o)

Crazy time... 130447

Collapsed crazy time.... 130448

Truth be told, I could probably lose about a quarter of it.

Scene etc below.
Rman texture node collection (http://dpont.pagesperso-orange.fr/plugins/Textures.htm) needed from Dpont.

Pinkmouse: scale specific stuff inside. Scale your pail to "my" (it's actually squarewulf's) can.
Also... did you know the the Make Material Node doesn't have any shading associated with it and that your surface will always be flat?
Try Standard instead :thumbsup:. Unless that's whats going on inside the Tweaker???

Oh yeah. Forgot to mention the size of my *cough* 3-5 minute frames. 1080x1500.

pinkmouse
10-21-2015, 02:14 AM
My nodes are always tidy....:D

Don't understand what you mean about the MM node, could you elaborate please?

Thanks for the scene, I'll post up a comparison shortly, should be interesting. And I'll add my nodes for those that have IFW2.

pinkmouse
10-21-2015, 03:35 AM
Hmm, for some reason, your texture was broken:

130454

...(and I do have Rman), so I've just copied the can object and added mine as another layer. Mine is on the right ;)

130455

To open mine, you will need IFW2 Nodal (http://www.shaders.co.uk/ifw2_nodal/about.htm), (Not free, but worth every penny and much more. Buy them!), and db&w Free Tools (http://www.db-w.com/products/dbwtools/docs)

130456

Have fun!

JoePoe
10-21-2015, 06:59 AM
Hmm, for some reason, your texture was broken:

Well that's weird.... and a bit distressing. I wonder why??
The weirdest part is that the simple stuff didn't come in.... invert, multiply etc. The Rman stuff looks like it came in although there's no preview on the Cells :stumped:.

I've saved out the tree as a .nodes file. Try importing that way?

Don't have IFW2. I had a free bunch at one point, but found the interface kinda(very) confusing. Maybe I should look again and try a little harder to make friends :).

pinkmouse
10-21-2015, 07:49 AM
That worked! :)

130460

I'll do an animation overnight, it's about 7min a frame on my old i7 4770 with your area lights and radiosity.

You really should look at Richard's IFW2 nodes , they have revolutionized my texture work, and it's very rare one or more don't find their way into anything I'm surfacing. Was it the nodes you had before? He does shaders and layer type procedurals as well, but I've never used them, even though they are installed.

JoePoe
10-21-2015, 08:09 AM
Cool.... but I think something is still getting lost in translation.
Your render looks much different than mine. 8~

130461

It's probably something in the middle...
I'm sure you've got the final connections right.....

130462

Edit: I think the scale setting for the Rman Cells isn't saving! I've got three in there ranging from .25 to .5 and all three are being displaced differently with other procedurals. Not to mention the other Rman node in there too (shape bomber). I'm gonna set all my Rman stuff to 1 to see if I can replicate your shot.

pinkmouse
10-21-2015, 08:40 AM
Yup:

130463

Do you want to try packaging it all up again, see if that makes a difference? Oh, I'm running 11.6.3, on OSX, what version did you make yours in? Can't think of anything else it could be...

edit: Ahhh! Can you just use constant scalars to set the texture size ?

erikals
10-21-2015, 08:44 AM
the left one needs reflection blur maybe ?

JoePoe
10-21-2015, 08:59 AM
Yeah... at 1 at least the cell size matches yours. See below.

Constant Scalars.... that makes too much sense :D. Will do that. But here's a screen grab just in case (the render is with all at 1, but the numbers shown are what they should be.)
But there's also something not right with the severity of the Blinn (or reflection?). Not sure if I'm up for chasing that down though. It shouldn't be at 100% spec anyway. Maybe just knock it down to 30% or less (judging by your render).

Did mine in 11.6.3 OS X as well :thumbsup:.

Erik: The one on the left needs a whole lotta things ;) There's 3% on there now. More the merrier?

130468

Edit: New .nodes with scalars attached

pinkmouse
10-21-2015, 09:21 AM
Okay, attempt #3:

130470

Sizes set as you suggest, Spec 20%. Diffuse 80% to match. Reflection colours still seem a tad extreme.

edit: just loaded up your new nodes, looks identical to above.

edit: edit: Try saving as a .srf from Surface Editor?

JoePoe
10-21-2015, 10:13 AM
Oh, I meant the Spec inside the Blinn. Spec out in main panel should be 0 (thanks for the reminder).

At least the pattern is back to the right size (crazy how the addition of 4 scalar nodes really seemed to hurt render times).

So here's a render with Blinn spec at 50% regular spec at 0.
Don't know why our results are so different. Other settings must not be getting through. It's an enigma wrapped up in a puzzle :D.

130472

.... and an .srf

jwiede
10-21-2015, 11:29 AM
You really should look at Richard's IFW2 nodes , they have revolutionized my texture work, and it's very rare one or more don't find their way into anything I'm surfacing.

I agree! I consider the IFW2 plugins indispensable, they're among the first third-party plugs I install. JoePoe, you really ought to give them another look.

I'm really hoping the new LW PBR engine includes a more useful and reliable anisotropy shader, like Vray's, or C4D's.