Planetside Software Forums

I wanted to try to get smoke easier, but I can't seem to get the white haze off the card. So; card, and projected white on black greyscale image file. Used the glass shader, but I might have to do some more testing.  You can use one sided or double side (in both glass and card), visible for other rays, etc.
Never mind the strange mist veils and absence of grass  ;)

---

Nice. Can you add an opacity mask ?

I just did some testing, but with or without opacity mask won't make a difference. In a test setup I had no haze on the plane, and a crop test of the same file just now, didn't have it either, strange.
Interesting; when using a (displacable) plane, it's not working properly, because the transparent part is subdivided less (like water, so I'd have to set subdiv levels to 1), and with a card it's fine. Probably they are calculated differently.

I see.

Rocks are superb btw.

Strange, I just can't get rid of the haze. If anyone has a clue...

---

I'd try an imported card for comparison,thinking of some of my experiments.

You may be right, I'll do that.

A nice challenge you've taken on ...great stuff!

First test: same problem.
No, whatever I do, it stays a problem; double sided glass, different index, negative reflection, sun other side, flip normals, flip object, water shader at infinity..... I guess it's not that perfect a method.

There has to be something,because the method works fine usually.
Tried with prop blur and other stuff (with the new glass shader) and
no haze problems.
At the moment I have no idea,though.Hmmmph!

The only way not to see the difference is use the whole width for the plane, but I noticed that I have to set the subdiv settings to 1. So obviously I do something wrong.

A chance to share a stripped down version with basic image maps or whatever you use to have a look Ulco?

Here's a crazy thought the Colour adjust?

Subdiv settings to 1 is required for transparency and reflections as far as I know.
At least when close enough to notice.
So maybe you're not wrong at all.

Hi, Ulco. I've been doing something like this recently for someone with a Card object(Haven't forgotten you, E! Just been busy).
Try inputting 'Default Shader 02' into 'A Smoke' and set the 'Glass Shader' as its child. Now the transparency will only be applied to the smoke area. The Default Shader opacity will completely remove the black area of the Card so, there will be no transparency applied there.

* And reposition the fractal child, of course.

Rock faces look great.

I missed saying, those mountain walls are superb. I would like to see more of this kind of work. Also, your tree models are choice.

Thanks, guys (and good to hear from you, Martin). BUT, it won't work. Opacity doesn't seem to work with a projected image. At least, I can't get it to work, or my brain has been deteriorating lately.
Strangely, in a simple setup, there's no haze, but in the mountain file there is. Maybe it's the clouds that have an influence....
Anyway, here's some file to play with, see what you can make of it. Changed the TIF to a JPG, for posting.

---

Forgot these; when making the plane/card enormous, covering the whole area (so not to have the edge), this is what happend. Strange, huh?

---

There is a solution probably Ulco but i haven't found one yet.
Not sure if it has to do with the using of nodes or a card issue or kind of a bug.

The only thing i can say is if you disable "Ray trace objects" you get different results.

Thanks Kadri. But that really is no option. I'll check it out later. Quite busy now.

Quote from: Dune on January 26, 2015, 09:06:59 AM
Thanks Kadri. But that really is no option...

Not surprised.

I found one with a merge node that kind of works. I will post it here.

Might be a base for further play Ulco.
If you change the "Mix to A" settings value the cloud changes kind of its opacity.
With "Ray trace objects" on.

Still this should be easier to use at as it is.
Not sure if we miss a very basic different way or not.

[attachurl=2]

---

A little rant:

Nodes might be very powerful but i still feel that there should be at least a kind of a supernode for surface layers then small different nodes.
You use a surface layer but then you want to use opacity for example and have to use the default node or you want then transparency and have to use another node etc. mostly.
I think at least  a node that has all the features of the " Surface Layer", Default shader" and "Glass-water shader" (at least) would be very nice and made some things much easier.
And a separate mask input for all the different options (transparency,opacity,slope angle, color etc.) would be awesome.

Ulco why are you using a camera projection here?
And is that essential to you?

Not going to comment on the issue (since I can't offer any thoughts) but I love, love this setup. Only maybe you need to model a more appropriate station/stop. :)

Thanks Kadri, I'll check it out, and see what this does in my official file. Thanks so much!

And, no projection isn't necessary, but it just turned out that way. A UV based plane may work better indeed. I'll revisit the file anyway, because I want the stones higher, but that's a bit hard with the track setup (smoothing can't be applied).

Your setup doesn't work any different I'm afraid, Kadri; it still has the haze! I'll do some more experiments.

---

I think the problem is the Render Subdiv settings, it blurs the rocks behind the glass in that haze. And I can't get the opacity to work only for the cloud with the projection. I'll try a AV based plane.

Here is the easiest way I can think of.
[attach=1]
Let me know if you need something more complicated. ;)

---

I really don't know what you want to achieve Ulco. This is with different subd settings that you already know.

---

J meyer Ulco tries to use a camera projection.With UV it is already easier.

Kadri - Ulco said:

QuoteAnd, no projection isn't necessary, but it just turned out that way. A UV based plane may work better indeed. I'll revisit the file anyway, because I want the stones higher, but that's a bit hard with the track setup (smoothing can't be applied).

And the card object is UVed it's meant for billboards etc after all. :)

Yes i know but he was still playing with a camera projection method until the last post. Anyway :)

Ok,no problem.At least he can choose between different setups now.

I just changed the Mix to A values in the merge node and one last image with a different gamma conversion number in the image map shader settings.
This is with your last file without the surface layer attached.
You might be after a different look maybe i don't know for sure.

---

You're terrific, guys! I have to admit I didn't realize the card was UV'ed. How shortsighted of me  >:( And yes the render subdiv settings make the difference, and cause the haze. I wonder if it takes much longer (in water it does considerably, but then again, this just a flat screen), which I was afraid of and thus didn't want to apply. Happens when I do things in a matter of fact way  ;)

Anyway, you earn my gratitude! For this version I left the smoke out, but I'll do a crop and put it in again. This took almost 4 hrs, due to RT reflection on the loc.

---

Nothing is impossible when all good heads get together. Also like your array of various weed/flowers on either side of the tracks.

Ulco, the black areas in one of your renders are because you're using a broken version. You need to update to this version:

Quote
Build 3.2.03.0 (release)


Fixed a problem with the Reflective Shader and reflectivity in the Default Shader whenever we see a surface from the reverse side. This can happen with many objects, so we recommend that you get this update. The bug was introduced in 3.2.02.0. If you load a file that was saved in 3.2.02.0 you might see a warning about a missing parameter "double-sided surface" in the Reflective Shader, but the warning is harmless.

Matt

I don't know if that is the case, Matt, but I'll check it out. Might have picked the wrong shortcut to open the file. Thanks.

I might have this wrong, but in order to get the (non RayTraced) background in all detail through a glass sheet, you need to up Render Subdivision settings to 1, even if there's no reflection and refraction is 1. Am I correct? And would it be interesting to have it so that you needn't alter subdiv if using these glass settings? Not a lot of users will know how to. 
I did a small experiment, but then the render took twice as long (crop, so it may probably be negligible on a completer render if the sheet isn't very big). So I added a default shader for opacity of the most part of that sheet, and that works pretty good, if the cloud isn't too dark.

Quote from: Dune on January 29, 2015, 02:47:11 AM
I don't know if that is the case, Matt, but I'll check it out. Might have picked the wrong shortcut to open the file. Thanks.

You can check what version a project was saved from by looking at Project Settings (the button at the bottom left of the window).

Quote
I might have this wrong, but in order to get the (non RayTraced) background in all detail through a glass sheet, you need to up Render Subdivision settings to 1, even if there's no reflection and refraction is 1. Am I correct?

Yes, that's how it works at the moment.

Quote
And would it be interesting to have it so that you needn't alter subdiv if using these glass settings? Not a lot of users will know how to. 
I did a small experiment, but then the render took twice as long (crop, so it may probably be negligible on a completer render if the sheet isn't very big). So I added a default shader for opacity of the most part of that sheet, and that works pretty good, if the cloud isn't too dark.

Yeah, theoretically I should be able to bypass the ray trace in certain situations like this, and have the shader either look up the background or automatically set the opacity to 0. But with so many different rendering flags affecting atmosphere, object visible etc. it could become complex to implement, so I'm not sure.

Matt

Thanks, Matt. I was just about to ask whether that could be implemented for some 'real transparency', but you edited just as I was typing.
I'm still having issues with the card and the slight haze (mask is really black, subdiv =1), so I'll keep on testing.
By the way, it was 3.2.03.0.

opacity-glass-smoke-experiment.tgd says it was saved in .02, so I think that explains the weird black areas. I posted about that before I saw your later files, which show that you've updated to .03.

About the slight haziness, are you using a Surface Layer for the smoke? I wonder if the problem is caused by "mask as coverage"; try turning that off. Alternatively, maybe you can just use an Image Map Shader with built-in alpha masking, via the new "Use alpha/transparency for direct blending" parameter.

Matt

I have just done precisely that (without reading your post), but I'm still puzzled. Look at these files; render subdiv is at 1. The top second image is with the simple merge setup, so doublesided glass + UV-defaultshader, merged (1) by screening default.

The black background came from using an imported plane object, but I replicated it in the latest version as well.

---

I tried many things but could only get meh! kinda solutions.
But as i tested it i begun to think that there is a problem with glass node,environment lighting or wherever.
The GI surface detail stage makes it even more apparent in the last phase.

If you look at the image i attached below the 2 images on the left are with the "Enviro Light" node enabled.

The 2 on the right are with the "Enviro light" disabled.

The 2 top images are where your sun is in the original position. The lower images have the sun more from the front.

Without the "Enviro light" node there is no haze!.

---

Good thinking, Watson! I did notice that the difference became more apparent as soon as GISD was ready. So it looks like GI is affecting the glass shader, maybe that's a bug?

We have to wait for Matt :)

Yes, probably just waking up now  ;)

:)

By the way this is the meh kinda solution Ulco.

The edges are a little hard because of the opacity-mask as you see.
The problem is still hiding here in the white parts probably.
I used a very bright light to see better how it looks.
Might be useful in different lighting maybe.

Put those two images in the same folder -you know- for easier use.

---

It looks quite ok. I tried the opacity with a surface shader and inputting default's opacity from the original mask, so the rest as child on the surface shader. But you get hard edges on the smoke. It's tricky, but it would be wonderful if we could have a very basic 'glass' shader unaffected by render subdiv, refraction, reflection or GI, or whatever, just serving as gradient transparency.

Yes some configurations get much more harder edges.
This was the best looking one i could find Ulco.
Might be better ones too of course.

Glass surfacing is still harder then it needs to be.

Ulco - Why do you use this setup? What is the reason you use the default shader and a
         surface layer? I don't understand the purpose.

Don't want to belittle you guys efforts,just being curious.

Quote from: j meyer on January 30, 2015, 01:20:09 PM
...
Don't want to belittle you guys efforts,just being curious.

No harm to ask at all on my part.
I tried your file too in Ulco's last setup for example but it got the haze effect too.
So i had to play with different configurations.

But what you want isn't really glass or how "glass" works in the real world. You're just using the glass shader as a way around the translucency limitation. So what you really want is real non-binary translucency support, correct?

- Oshyan

It might be GISD - could you try disabling that? GISD uses some amount of image post-processing, so it doesn't work the same in transparency rays (or reflections).

Matt

Quote from: Oshyan on January 30, 2015, 07:28:47 PM
But what you want isn't really glass or how "glass" works in the real world...

That wouldn't hurt  :)

Quote from: Oshyan on January 30, 2015, 07:28:47 PM
... You're just using the glass shader as a way around the translucency limitation. So what you really want is real non-binary translucency support, correct?
...

Yes.

Quote from: Matt on January 30, 2015, 07:30:14 PM
It might be GISD - could you try disabling that? GISD uses some amount of image post-processing, so it doesn't work the same in transparency rays (or reflections).
...

Yes looks like that is it Matt.  At least i can't see any haze. Curious how it is on Ulco's side.

Quote from: j meyer on January 30, 2015, 01:20:09 PM
Ulco - Why do you use this setup? What is the reason you use the default shader and a
         surface layer? I don't understand the purpose.
...

After reading your post one more time it looked to me as if you asked maybe for a more specific answer.
I used the default shader (most probably like Ulco) to use the opacity feature to get completely rid of the haze parts.

@Jochen; what setup would you use then to get a smoke column (from mask) without its surroundings? And Oshyan is right, that's what would be quite useful, so you could have distant fires not by making complicated cloud stuff, but just inserting some cards (distant anyway). Preferably even a generic cloud column that could be warped procedurally in world scale to get different shapes.

I'd use the setup shown in my example on page 3.Image map shader set to object's UV
as mask for a surface layer and the glass shader connected to the input of the surf layer.
Using the create transparency feature of the image map shader would be even better,
but I couldn't get it to work.

QuoteBut what you want isn't really glass or how "glass" works in the real world. You're just using the glass shader as a way around the translucency limitation. So what you really want is real non-binary translucency support, correct?

In case you actually mean transparency that would be cool indeed.

Forgot something.

QuoteA......
I used the default shader (most probably like Ulco) to use the opacity feature to get completely rid of the haze parts.

As far as I can say this is doing the same thing twice and thus unnecessary.
Put another way: why simple,when you can do it complicated. ;)
But I thought there might be another reason I'm just too stupid to see.

I also tried glass shader then image map + the transparency settings, but that gives all sorts of errors.

Quote from: j meyer on January 31, 2015, 12:07:19 PM
...
As far as I can say this is doing the same thing twice and thus unnecessary.
Put another way: why simple,when you can do it complicated. ;)
...

The problem is you get a different look with nearly every different combination.
It doesn't look so that it should when you look at the node tree but the renders are different.
And the renders who look the same are not the ones you want.

So i tried many different combinations.
I was going to show some renders and node setups with your setup,mine
and Ulco's and some between but there are so many that i gave up to post them here :)

A not fully matured feature,it seems. ;)

:)
Nobody responded to my rant on page 2 but i stay by it. I want a supersurface node with opacity,transparency,reflectivity etc.
Why should it so hard to mask a surface or use opacity,transparency etc. on a surface?

Quote from: Kadri on January 31, 2015, 05:23:48 PM

:)
Nobody responded to my rant on page 2 but i stay by it. I want a supersurface node with opacity,transparency,reflectivity etc.
Why should it so hard to mask a surface or use opacity,transparency etc. on a surface?

Heh heh, likely because we've all let loose with that rant and are waiting patiently for the Surface shader to come of age...heh.

:)

I don't think we need to rant (I know you didn't mean it that harsh), but ask Matt (and Jo) nicely if something like we need would be possible. I understand it's difficult enough to build software like this and I have great respect, the more since it's getting more complicated with every new feature. And if I see what difficulty I had a few years back getting things together (like a whole city, house by house, exactly nexto to eachother with only box view!), we've gone a huge way.

Of course i don't mean it harsh.  Just want a little more exposure of the things i need and probably others too :)

Didn't respond to your "rant",because I for one don't like the all in one idea,but that's
just me.

Still interested in the haze thing?
Since I never had seen that in my tests with propblur and the like I tried to recreate
Ulco's atmo and light settings etc and used one of my test objects instead.After all
Kadri said he had seen the haze effect with "my" setup and others,too.
So I did,but no haze,no matter what settings were used.Gisd on or off -> no haze.
Defer atmo or whatever -> no haze.After several more tests with my file and Ulco's
latest file it dawned on me finally.At least I hope so. ;)
I couldn't get no haze,because there is no haze.
Had looked for the main portion of the discs to become hazy as in the examples in
this thread.
That didn't happen.The haze effect is merely an optical illusion that fools our eyes.
What happens is the AO effect of the Gisd is blocked behind the transparent object,
simply not rendered and that looks like the object becomes hazy/foggy under certain
conditions.Especially when there are lots of details behind it like in Ulco's example.
In my test file there is just a thin line at the bottom of the wall and thus the disc
stays clear.
Took quite some time to notice and realize that.
[attach=1]
Have a look yourself.Maybe I'm wrong again.

---

You haven't read some of the posts from page 4 probably. Yes it is the GISD.

Quote from: j meyer on February 01, 2015, 10:44:40 AM
I for one don't like the all in one idea,but that's just me....

That is what i sometimes don't understand. You don't like that features should be easier to use?
The separate nodes could be still around for hardcore users.
Why don't you like a node where all the standard things for a surface are in one node united?
Curious.

QuoteYou haven't read some of the posts from page 4 probably. Yes it is the GISD.

I had read Matt's post,but it got me thinking in the wrong direction.
You should have notcied by now that I can't even count to two sometimes. ;D
My point is that Gisd doesn't make it hazy,it just don't get rendered behind
transparent objects,that's all.Sorry for being dumb.

QuoteThat is what i sometimes don't understand. You don't like that features should be easier to use?
The separate nodes could be still around for hardcore users.
Why don't you like a node where all the standard things for a surface are in one node united?
Curious.

To me it doesn't make things easier necessarily.Look at the default shader for example.
It shares some features with the reflective shader and the image map shader,but there
are differences.And those force me to use the other shaders in conjunction often.
On the other hand I have to admit that some features of the default shader are missing
in the reflective shader and I would like to have them there also.Image slots mainly.
The other thing is that I think it's more versatile to set things up the way I want.
At least I imagine that.
But I know that's just me and the majority likes to have all in one for everything.
Shaders,whole applications or whatever.
So usually I keep my mouth shut and try to make the best out of the given.
Hope that explains a wee bit. :)

Quote from: j meyer on February 01, 2015, 12:35:50 PM
...
But I know that's just me and the majority likes to have all in one for everything.
Shaders,whole applications or whatever.
...

Actually we don't know much if the majority wants that.
I suppose it might be so, but when i wrote what i want, it was in a way a probe if others would be say anything in that way.
But i got only silence. So hard to say what others think.

I am in a awkward stage just now that i feel more like a technician the last months then a guy who wants to make art.
I feel the boundary should be narrower then it is now (of course Teragen has come a long way).
Otherwise many 3D software does have the same problems less or more in that aspect.
More tools-nodes are nice but there should be more easier ways to use them together too. 
More nodes doesn't mean better easier use all the time.

Michael's(Thebadger) wish for more inputs for masks for example is another example that there might be other things to think about.

Maybe a thread where we (me actually) don't hijack Ulco's thread :) and say a little more what we want or not might be better.

Sorry Ulco. If i want to say more it will be on a different thread. 

No worries, Kadri.

Some shaders may have been designed for a specific use before others came in and can't just be dumped, I guess, due to backward compatability. If you get to know the different shaders you easily use them beside eachother, no need for an all-in node... for me that is. Masking one by another, like a reflective shader by surface shader or a merged set of shaders, no need for more mask inputs (for me).
Like, just now, I noticed that a Lambert shader followed by a reflective shader takes out the translucency, but when merged (add) it works fine.

Last week I didn't find the time to play/test with the haze-on-card-test file,
so I just try to catch up.Should have said said earlier,but thought it wasn't
necessary.A mistake on my side.Sorry.

Another thing I noticed yesterday:put an imported sphere in front of the cube
in the aforementioned file,connected a glass shader to the sphere insread of
the default shader.Reflectivity set to zero everything else default.
[attach=1]
Remember the render subdiv is set to 1 and it can't be reflected from behind
the camera as  a)reflectivity is set to zero and b) the cube is visible.

Has anybody noticed already and or can confirm,please?

---

I think the blocky sides in the sphere in this setup are because of the frustrum limitation. I gather that's what you mean?

Yes,that's what I mean.
Frustrum limitation,wouldn't that be the reflection from behind the camera/viewer?

Edit:looked up frustrum meanwhile and it seems to be a greater region than I thought.
      So,yes that may be the explanation.Thanks Ulco.

Yes, and it's big because the ball works as a more-than-fisheye lens.

I do agree.

Hope I finally understood your original intention now.
[attachimg=1]

[attachimg=2]

Is that what you had in mind?
Sorry for the white outlining.

---

I got this far as well, but there's still a haze in the parts that are not taken out by the opacity, but it is less noticable, I agree. The white line is probably because of the premultiplying.
This can be done with less nodes; the default for the opacity, then the surface shader on which (as child) the glass and then the image map. I also had trouble making the cloud dark, after using the mask as such, then it's really outlined, and I would like it to 'fade into oblivion'. I might have another look at it...

Getting the cloud darker was no problem,I had problems getting it lighter.
Didn't optimize the setup,easier to work with to me.
And the remaining "haze" (missing AO) within the smoke won't be noticeable
to a viewer that doesn't know about it.It'd be just a grey part of the smoke.
But of course you would know.I'm familiar with that problem,too. :)
Tried it with a smoke object also,but the results weren't better,maybe if done
with a 16bit or 32bit image,but not from the jpg.You could try with mudbox.

I just set up another test. I think this is the best I can get, and I actually don't see a haze here. Thanks for your efforts as well, I think we can leave it at this, don't you?`We wouldn't want to get too picky  ;)

Wholeheartedly agree! ;)

Quote from: Dune on February 02, 2015, 11:35:42 AM
I think the blocky sides in the sphere in this setup are because of the frustrum limitation. I gather that's what you mean?

This region can be changed using the "ray detail region" setting. This is a case where it's useful to choose "360 degree detail (optimal)".

Matt