IDL again...

jergenoman_b33d4209f8

Ok, here is my newbie question of the day. :-)

I have attached three small test renders. They all share the same basic setup:

- One shader mixer spotlight in the ceiling with a spread angle of 170 degrees). Raytraced shadows with intensity 1 and samples set to 4.

- An IDL camera with idl intensity set to 0.5, final gather on, and photon count set to 500000.

The first image has shadow blur set to 0%, while the second one has 3% blur. The blurry shadows gives a grainy effect (graininess IMO can be seen in all 3DLight renders) and will need a higher "Samples" settings, but why is the blurry one significantly darker than the first image? What am i supposed to do to compensate for that? Simply raise the light intensity?

The third image has no objects shadowing the wall corner. The light is positioned in the ceiling as before and approximately at the same distance from both walls so that it should illuminate both walls equally. As you can see it doesn't. :long: The walls have the same very basic shader applied. Simply a diffuse color and a bit of specular. The tiles are a bit more complicated but not much. Does anyone have an idea of what causes this? And why isn't this showing on the tiles?

Edit: Should have said that the spot is facing vertical down to the floor

/Jörgen

jergenoman_b33d4209f8

Another reflection regarding IDL in Studio compared to Poser/Firefly IDL:

Does anyone know how 3DLight IDL is implemented? Is done with forward or a backward raytracing? Firefly uses forward raytracing to pre-calculate IDL before the actual render takes place with backward raytracing. This is really the way to do it because IDL calculations are actually the opposite to rendering: When rendering you follow rays from the pixels in the resulting image into the scene to see what object/color correspond to the pixel. When calculating IDL you follow rays from the lights into the scene, bouncing around and illuminate everything. It is fundamentally impossible to combine them both in a single pass unless you have unlimited computing power.

I wonder because the IDL interface in Studio is very sparse. You can set the number of photons in the photon mapper, but can i control the distribution of them? I feel that 3DLight spread them out evenly and wastes most of them on big flat areas. The photons should go where they are needed!! Into corners, and onto complex geometries and displacements. There isn't even a control for the number of bounces...

Jörgen

This is a screenshot of the IDL render pass in firefly. Note the adaptive distribution of the rays.

jergenoman_b33d4209f8

Ok, i found this page with much info. :-)

http://www.3delight.com/en/uploads/docs/3delight/3delight_40.html

Not all of this seems to be supported from withing Studio though.

Szark

I think the main problem with helping you with this is A; lack of documentation and therefore not many folks that frequent these forums are not yet familairy with the Shader Mixer. With my machine I have never been able to test IDL via Shader Mixer at all.

jergenoman_b33d4209f8

You are right, and this is a major problem!

DAZ needs to realize that the lack of documentation is seriously holding Studio back. This is the main reason that has kept me from switching from Poser for many years even though i was ready to kill myself many times due to its millions of drawbacks, not to mention the instability. I recently had an incident with Poser that made me uninstall it and i hope to never use it again. I have constantly kept an eye on the development of Studio and i have had most major releases installed. If there would have been a proper documentation i would have changed software years ago.

I can't spend days searching YouTube, the archives here (witch is a complete chaos worthy only of a Poser archive...) etc. I simply don't have the time! I have a full time job and a family. If i want the specs of a shader node in Poser i just open the reference manual PDF and hit CTRL-F and thats how it should be. I could live with an online documentation but then it shouldn't look like this: (example from the "Indirect Diffuse" brick)

Parameters
Outputs
• Indirect Diffuse : Color - Description…
• Occlusion : Float - Description…
• Avg Environmental Direction : Vector - Description…

Inputs
• Vector : Normal - Description…
• P : Point - Description…
• Strength : Float - Description…
• Axis : Vector - Description…
• Max Distance : Description…
• Cone Angle : Description…
• Samples : Description…
• Bias : Description…

What is the point of even having this page online? I don't get it!

Please DAZ you are scaring away the intermediate and advanced users you need to make Studio to really take off!!

Jörgen

Szark

We all feel pretty much the same when it comes to a Daz Studio maunal and I cannot say anything to help or to change the situation. Also the DAZ3D staff don't read the forums so much so it might be better to vioce your concerns directly to them in a support ticket.

I am not too sure if this will hep any http://docs.daz3d.com/doku.php/public/software/dazstudio/4/referenceguide/interface/panes/shader_mixer/start

jergenoman_b33d4209f8

I'll buy a RSL book and start from scratch. ;-P

ReDave

Number of Max raytrace bounces has an effect on ID, in particular odd number versus even. I had never tried blurred shadows, but thinking on it, that should make everything darker.
I don't know how it's implemented, but I expect it's just backward raytracing. It isn't adaptive unless it is coded so, that's why it's slower than Poser's. There are point-cloud scripts on the web for 3Delight (one, with just AO supported is available in "Scripted Render" instead of 3Delight as render engine in the advanced render settings), but I don't know how to convert them for DAZ Studio. (http://www.jorgepimentel.com/?p=90)

jergenoman_b33d4209f8

I wonder if "Min depth" has something to do with the IDL bounces... According to the documentation it does "Description..." ;-)

Blurred shadows are supposed to emulate that the light source isn't a point but has an area. It isn't supposed to do anything but giving a transition between shadowed and non shadowed areas. See attached image. For outdoor scenes this is very important since objects that are far away from its shadow cannot have sharp shadows or it looks very bad. About 2% blur is good for the sun. A fire is another example of a light source that needs blurred shadows.

Reading the 3DLight documentation here: http://www.3delight.com/en/uploads/docs/3delight/3delight_40.html gives me too the impression that it is a backward raytracing emulation of IDL and not a proper one. :long: It seems to me that the primary purpose for this photon mapping implementation isn't illumination at all. Quote:

"The main applications for photon maps are:
Render caustics ...
Render color bleeding using "final gathering" ...
Render fast global illumination previews ..."

I dont think 3DLight can do what i want... :long:

millighost

ohman said:

Ok, here is my newbie question of the day. :-)

I have attached three small test renders. They all share the same basic setup:

- One shader mixer spotlight in the ceiling with a spread angle of 170 degrees). Raytraced shadows with intensity 1 and samples set to 4.

- An IDL camera with idl intensity set to 0.5, final gather on, and photon count set to 500000.

The first image has shadow blur set to 0%, while the second one has 3% blur. The blurry shadows gives a grainy effect (graininess IMO can be seen in all 3DLight renders) and will need a higher "Samples" settings, but why is the blurry one significantly darker than the first image? What am i supposed to do to compensate for that? Simply raise the light intensity?

Hard to tell, but if you use displacement, surfaces tend to get darker because of self shadowing, when shadows are involved. Even more so when rendering without gamma correction, where each averaging operation (like blur) darkens the brightness. But if you want to track it down, i would suggest to set the pixel samples, pixel filter width and -height to 1, which makes things easier to see.

The third image has no objects shadowing the wall corner. The light is positioned in the ceiling as before and approximately at the same distance from both walls so that it should illuminate both walls equally. As you can see it doesn't. :long: The walls have the same very basic shader applied. Simply a diffuse color and a bit of specular. The tiles are a bit more complicated but not much. Does anyone have an idea of what causes this? And why isn't this showing on the tiles?

Edit: Should have said that the spot is facing vertical down to the floor

/Jörgen

I would guess that it is the specular on the wall that makes the back wall appear brighter; it seems to be the only thing in your setup that is view dependent.

Another reflection regarding IDL in Studio compared to Poser/Firefly IDL:

Does anyone know how 3DLight IDL is implemented? Is done with forward or a backward raytracing?

Yes. It uses photon maps if you enable them, otherwise pure raytracing.

Firefly uses forward raytracing to pre-calculate IDL before the actual render takes place with backward raytracing.

Never heard of that, do you have a reference? The red dots that are displayed by poser are the positions where the irradiance is calculated when it fills the irradiance cache. I am not sure if 3delight does irradiance caching, i believe not, which makes it slower, but on the other hand avoids those splotchy corners you often get with irradiance caching. And you can always manually bake lighting as a last resort.
I think what you get when using your setup without photon mapping (i.e. only the indirectdiffuse, but with photons set to 0) is similar to what firefly does (without the IC, of course).

millighost

ohman said:

...For outdoor scenes this is very important since objects that are far away from its shadow cannot have sharp shadows or it looks very bad. About 2% blur is good for the sun....

Given that the sun is 150mio km away and has a diameter of only 1.4mio km, an even better value would be 0.5%. (unless by "outdoor scenes" you mean scenes on Mercury, in which case a value of 1 to 1.5% could be used).

jergenoman_b33d4209f8

millighost said:

Hard to tell, but if you use displacement, surfaces tend to get darker because of self shadowing, when shadows are involved. Even more so when rendering without gamma correction, where each averaging operation (like blur) darkens the brightness. But if you want to track it down, i would suggest to set the pixel samples, pixel filter width and -height to 1, which makes things easier to see.

There is no displacement going on apart from the genesis skin shader. Good idea to turn off filtering! I will try that.

I would guess that it is the specular on the wall that makes the back wall appear brighter; it seems to be the only thing in your setup that is view dependent.

That was my first thought but the effect remains from all angles, but slightly different in strength.

Never heard of that, do you have a reference? The red dots that are displayed by poser are the positions where the irradiance is calculated when it fills the irradiance cache. I am not sure if 3delight does irradiance caching, i believe not, which makes it slower, but on the other hand avoids those splotchy corners you often get with irradiance caching. And you can always manually bake lighting as a last resort.
I think what you get when using your setup without photon mapping (i.e. only the indirectdiffuse, but with photons set to 0) is similar to what firefly does (without the IC, of course).

I might have been to fast to claim that Firefly uses forward raytracing... I read it somewhere but when i go back and read it seems like it is another backward estimating algorithm. However it is a good one, and you can easily get rid of splotchy corners if you tweak the settings.

I still don't understand how 3DLight IDL is meant to be used. When i set the IDL intensity to 1.0 i get a result very different from what Firefly does. Seems to me like the focus is more on color than on shadows. Are we supposed to use it in conjunction with ambient occlusion? I don't like AO at all.