IBL + IDL .. Yay or nay?

yay

I did a version with IBL off but didn't keep it for comparison - I don't think there was a huge difference.

IBL works with IDL, but it works differently than without IDL.

When IDL is enabled, the colors in the IBL are projected on a gigantic virtual sphere that surrounds your scene. If your subject is inside a close room, the light from the IBL is blocked. Without IDL, the light would reach everywhere, passing right through all surfaces.

If you are just doing a figure and it is not enclosed, the IBL light will reach it, but it won't pass through arms and legs. Thus, the shadows produced are correct.

Light-based AO is inferior to this and is ignored when IDL + IBL is used. Even if you turn AO on for the light, the renderer ignores that and uses the superior rules afforded by IDL.

Oh I forgot to mention. If you use my EnvSphere, there is no point in using an IBL. The virtual IBL sphere is bigger (infinite) and so the light will come from the EnvSphere, which completely encloses your scene.

With IDL, the lighting from the EnvSphere is automatically correct and matches the scenery you load into it. Further, in the enclosed room above, the EnvSphere only provides light that can be seen through the windows. The room blocks the rest of it.

Quote - Oh I forgot to mention. If you use my EnvSphere, there is no point in using an IBL. The virtual IBL sphere is bigger (infinite) and so the light will come from the EnvSphere, which completely encloses your scene.

BB, thanks for clearing that one up... I think you've told us this several times over the last few days but it's only just sunk in with me.

Quote - > Quote - Oh I forgot to mention. If you use my EnvSphere, there is no point in using an IBL. The virtual IBL sphere is bigger (infinite) and so the light will come from the EnvSphere, which completely encloses your scene.

BB, thanks for clearing that one up... I think you've told us this several times over the last few days but it's only just sunk in with me.

Yea this just clicked in my head :)

Baggins ftw :)

I love BagginsBills posts!

What would we do without you, BB.
You truly have patience.
cheerio
lululee

Well, without me, you'd probably have the old library GUI and nobody would be complaining about small thumbnails. ;-)

    BB,
As a vendor, I have made a lot of mistakes when trying to push the envelope. It is devastating.
I used to show horses. When I got thrown off the horse or lost a competition my mother always instructed me to get back on the horse saying I could only quit when I got it right or won. I became a champion horsewoman because of it.
    You are doing something very complex. Something no one else would take on. So, it is not perfect the first time. You are still a champion and will push the envelope for all of us. You will bring out a fantastic new library soon. It will be way better than the old one because now you know want users really want. In 3 months all of this will be forgotten.
Steady on my friend, you will overcome.
cheerio
lululee

Except, it may be useful to do just that in some circumstances. These are crops from a couple of renders, same setup for each (car, figure, env sphere), the only difference being the use of IBL with AO set at 2.0 on the second image. That seemed to have improved the shadows on the face.

That there is a difference is shown on the right. If this is a bug, it could be a useful one!

The image is self-explanatory: identical setup, the only difference being Ray-Traced shadows vs. Depth-Mapped shadows.

I just ran the same test with a different figure in the scene - this one had hair with a specular map - and got even more of a difference: ray-traced 21:45, depth-mapped 7:14.

EDIT: is it just me, or does the colour of Miki's suit look slightly different?

the skin looks alittle tiny bit more saturated in the 2nd one, but maybe its just me.

Quote - > Quote - If you use my EnvSphere, there is no point in using an IBL.

Except, it may be useful to do just that in some circumstances. These are crops from a couple of renders, same setup for each (car, figure, env sphere), the only difference being the use of IBL with AO set at 2.0 on the second image. That seemed to have improved the shadows on the face.

That there is a difference is shown on the right. If this is a bug, it could be a useful one!

Huh. You're right. I did not do a detailed test until now. Despite my own advice to the contrary, I trusted my eyes, and what I thought I saw was that IBL had no impact when the ESphere was present.

So I just did an experiment. I set up an IBL with only green in it. I set up an EnvSphere with only red in it. I rendered some white props.

With only the IBL, everything is a pure shade of green.
With only the EnvSphere, everything is a pure shade of red.
With both, it looks pure red, but there is a tiny bit of green in it!!!!!!

This is a bug. Don't count on it in the future.

This is the IBL render.

Make the EnvSphere somewhat transparent. This lets the IBL light leak through. Here my EnvSphere is 50% transparent. Looking directly at it, it is still red, because it is self-lit and pays no attention to the IBL. But the props are picking up half the light from the EnvSphere and half the light from the IBL. 50% red + 50% green = dark yellow.

 Bookmark.

There's an important difference between how an IBL light can behave with indirect lighting enabled, and how an environment sphere (or other geometry) that emits Ambient will behave: the IBL can have shadows turned off.  Ambient emitted from a material will always be occluded by polygons that are not transparent; IBL with shadows off will always pass through it - AND bounce.

Cspear, according to Stewer, when indirect lighting is enabled, light-based AO is skipped, so you might as well not bother with it.

PJ - my EnvSphere has shadows turned off and my IBl has no shadows enabled of any kind. Why are you contradicting the evidence I demonstrated? Not being an ass. I'm wondering if you've tried it a different way than I did.

Ummm, isn't there a contradiction here?

One says that IBL with IDL will be occluded.
One says that IBL with IDL will not be occluded.

What I thought I had learnt is that IBL with IDL does not work as IBL at all, but as ambient light emitted from a virtual all-enclosing sphere. So it will be occluded.

(edit) oops, cross-post, never mind!

A flattened sphere hovering over the ground.

The only light is an IBL.

On the IBL, shadows are disabled. AO is disabled.

On the sphere, Cast Shadows is disabled.

Yet there is occlusion under it. I don't see what else there is to say about this. The only possible conclusion is that IDL looks to the first opaque surface and that's where bounced light comes from. If it never reaches a surface, then bounced light comes from the IBL, "mounted" on an infinite sphere enclosing the scene.

By the way, I just demonstrated another bug. Who sees it?

Shadows enabled?

Quote - Compare this with the earlier one. There is a thin box under the flattened sphere. What is different?

There's a shadow on the bottom half of the sphere in the second image.

Quote - > Quote - Compare this with the earlier one. There is a thin box under the flattened sphere. What is different?

There's a shadow on the bottom half of the sphere in the second image.

Right!

It would appear that when IDL sends out a ray from the bottom of a sphere, it passes right through the Ground without realizing it, and picks up light from the IBL, when it should not.

When I put a flat box there, it saw the box, same as it should have seen the ground.

Poser 8 Ground seems very buggy. I have found more than one situation where Poser seemed unaware that the ground exists.

For example. Load a couple props. Go into the material room. Click on one prop - its material gets selected into the material room display. Click on another prop - that one gets selected.

Now click on the ground. It selects the Background material instead of the Ground material.

It acts like the Ground isn't there.

Uli told me the Ground is not real geometry - it is procedural. It seems that not all the parts of Poser are procedural-geometry aware.

They should have got rid of that Ground and just used a prop for real. These problems wouldn't exist.

On my Poser 8, the GROUND material comes up with Shadow_Catch_Only enabled by default. According to the manual, this means it is transparent to light, but catches shadows produced by other objects. This sounds roughly consistent with what you're seeing to me. That would mean it's not a bug but a feature, I think.

Whichway

Have you forgotten I'm a Poser God? I have my own shadow catcher that is far superior - look in my free stuff.

First thing I did (FIRST THING) was make the ground normal. Clearly it's not a shadow catcher because you can see it in my render.

It picks up light from the ground, but doesn't count for occlusion.

Sorry, (dang, I'll get one of them right one of these days.) My only defense is that I don't think you said - at least I didn't read - anywhere nearby that you'd normalized the ground, but your comment about it being visible is certainly valid. Your devoted acolyte quails before your august presence.

Whichway

That doesn't work either.

I'm switching my default scene to a scaled box. That's the only thing that works right.

I wonder if walls made of single-sided objects have a problem, too.

Because they do indeed behave differently.  See the linked file.

http://cid-b233dcaeefa9709c.skydrive.live.com/self.aspx/Public/Poser8%20samples/IBL%20vs%20Ambient.zip

This is with an IBL, shadows turned off:

and for reference, another "room" box is included, "Room Outer", with the normals facing out.  With both the IBL and the environment sphere, it occludes all incoming light (which imo shows again that including normal facing in GI calculations is not a good idea, it shouldn't matter).

Edit: to clarify, it doesn't work like either of us thought it did; with an IBL that has shadows turned off, light passes in through the backfacing normals; with the environment sphere, it is occluded.  With outward-facing normals, both are occluded.

Wait, I don't even know what you're talking about anymore.

IBL with shadows off? You mean shadows, like we use on a directional light, raytraced or depth-mapped shadows? You're not talking about AO? Why are we talking about that now? On an IBL, the "Shadows" property is identical to an infinite light, and is actually nonsense and has always been that way.

Also, you last said "light passes in through the backfacing normals".

Well, that appears to depend on the size of the polygons or whether we're inside or SOMETHING, I don't know which.

I'm getting confused now with my current tests. Small one-sided squares don't seem to block IBL at all, but my environment sphere blocks IBL even with normals in or out.
 

I used Bagginsbill's Environment Sphere with a panoramic photo from Flickr. The shaders/textures and morph are of my own design. Comments are welcome! 

On the subject of occlusion, I've experienced similar issues.  When I built my Photo Studio using planes as walls, there was almost no shading at the corners where the walls met.  When I switched to thinned-down cubes, the missing occlusion magically appeared!

IS:

Are you using GC shaders with IDL? I haven't tested extensively, but I think you really can't. The data that GC shaders reutrn is for how to make that luminance on a screen. For light measurement purposes, those numbers are way too high. That would cause over-exposure for sure. That's why linear rendering is so important.

With Poser 8 and IDL, dump GC shaders altogether. Use linear output only from shaders. Make sure all diffuse calculations throw less light than arrives (Diffuse_Value < 1). Use HSV Exponential Tone Mapping at 2.2 for the final adjustment (approximately) to sRGB color space.

Quote - I'm getting confused now with my current tests. Small one-sided squares don't seem to block IBL at all, but my environment sphere blocks IBL even with normals in or out.

Dunno if this is relevant but Poser 7 used to have all sorts of problems with small props.  Firefly used to produce some very odd shadows (square shadows from spheres for e.g)

When I say small, I don't mean inches. I mean a 10 foot wall 5 feet away versus a 1000 foot wall 500 feet away.

OK I've found something else completely strange.

PJ - you said based on which way a normal faces, it may or may not let IBL light through under IDL.

How does this grab you?

Load a subject prop. Set up a constant light gray IBL, no shadows no AO. Render. The prop looks uniformly lit.

Load a one-sided square. (EDIT: MAKE IT BLACK) Scale it up 500% (EDIT: I said 5000% first time - I meant 500%) and position it beside the prop so you can see the prop, but this wall can block a significant angle of the environment. Render again. No change to the appearance of the prop.

Now duplicate the wall, so there are two of them. Render again. The wall now blocks IBL and the surface facing the wall are darker.

What is your theory?

Variations:

Turn either or both walls around, so the normals face towards the prop. Same thing happens. One alone does not block IBL, two do.

I had all shadows and AO on lights, props, everything turned off.