OTOY Forums

Regarding the caustics -- yes, the black shadows cancel out the caustics. The black shadows are composed in with the normal layer mode which is alpha blending. Mathematically, alpha blending works like this: alpha * shadows_pass + (1 - alpha) * background. This means that indeed where the shadows are fully black (alpha = 1), the background with the reflections on top of it is (1 - 1) or invisible.

You can change the opacity of the shadows or you can swap the order of the reflection and shadows layer, likely you will have to make the reflections more transparant in this case because they will appear to strong. Like I said, backplate compositing is more of an art than a science so you have to tweak until it looks good subjectively.

cheers,
Thomas

goes well with single objects simple, but how can I do when I want to isolate from a surface an object that consists of 20, 30 or 100 items and more? it takes a lot of time and patience to connect all the nodes, there is no way of being able to consider it a single object?
have you considered this?
When I export from C4D what should I do?

augdalmo wrote:goes well with single objects simple, but how can I do when I want to isolate from a surface an object that consists of 20, 30 or 100 items and more? it takes a lot of time and patience to connect all the nodes, there is no way of being able to consider it a single object?
have you considered this?
When I export from C4D what should I do?

Hi,

If you assign an object layer map to your geometry node(s) but don't connect an object layer to each pin of the object layer map, the objects associated with the empty pins will all be on layer 1. What you then can do is only connect an object layer map to the pin of the object you want to isolate. For example in the picture, everything is in layer 1 while the floor is in layer 2:



cheers,
Thomas

linvanchene wrote:

augdalmo wrote:goes well with single objects simple, but how can I do when I want to isolate from a surface an object that consists of 20, 30 or 100 items and more? it takes a lot of time and patience to connect all the nodes, there is no way of being able to consider it a single object?
have you considered this?
When I export from C4D what should I do?

In lack of other users adding their opinions:

My suggestion would be to add an additional pane / tab dedicated to the task of assigning object layer nodes to specific render layers.

The object & render layer tab displays a list of all the object layer nodes and render layers in the scene.

The user can use combinations of CTRL and SHIFT clicks to create a selection of Object Layer nodes.
Right clicking, drag and drop or a pulldown menu is used to assign the select Object Layer nodes to a specific Render Layer.

- - -

Update/Edit:
Removed some information in order not to go further offtopic.

I do agree that even though the standalone NGE has options to group / ungroup nodes it becomes increasingly difficult to navigate large scenes and work with them after 100+ nodes are present in the scene. Nevertheless I do realize that maybe this is more of an issue how plugins are collecting the nodes in .ocs files.

Maybe here the solution would be that plugins try to create separate groups for each object in the scene by default. That way imported .ocs scenes would be more manageable after import.

How ever that may be I believe the NGE should not be the only interface for tasks that only need a fraction of all the nodes available in the scene.

Additonal tabs that focus on specific tasks like assigning Object Layer Nodes to Render Layers would seem very helpful.

Hi,

Indeed the standalone isn't that great when you're working with large projects. In this case it's lacking the ability to "edit" multiple object layer nodes at the same time. Assigning a render layer ids to multiple object layer nodes is only half the problem. There's also adding the object layer maps itself and connecting the object layer nodes.

We're thinking about a solution. The plugins can (and should) make the project as complete as possible. But it's called OctaneRender standalone so it shouldn't rely on plugins to set up the project because some customers only use the standalone. Your proposed solution definitely makes sense for assignment of render layer ids and is something that can be done. But since it doesn't solve all the problems we're still exploring if we can do better.

cheers,
Thomas

Thomas, thank you so far for filling in my gaps.
I hoped that other users could jump in but I guess for now you seem stuck with me asking questions.

- - -

My three main questions about EXR are:

1) Is the default exported .exr file in 8 bit or 32 bit?
2) Do I need to select transparency or as alpha channel when importing in photoshop?
3) Does the EXR file support layers?

With the underlying question beeing:
Why do I just get a plain black image file when opening the EXR in photoshop?

- - -

obsolete parts edited and removed by user

Hi,

I'll try to answer your questions.

EXR is always 32-bit or 16-bit per channel. It's a file format for high-dynamic range images developed by ILM. You can get more information here. EXR is the best suited for a linear workflow as described here.

Yes, EXR supports multi-layer files. So you can export all your passes into a single file.

To use multi-layer EXR files in Photoshop you will need a plugin. Have a look here.

We had a look at exporting .psd files from Octane but there are only libraries available that support older formats and writing our own library would eat away a lot of our time.

cheers,
Thomas

stratified wrote:

augdalmo wrote:goes well with single objects simple, but how can I do when I want to isolate from a surface an object that consists of 20, 30 or 100 items and more? it takes a lot of time and patience to connect all the nodes, there is no way of being able to consider it a single object?
have you considered this?
When I export from C4D what should I do?

Hi,

If you assign an object layer map to your geometry node(s) but don't connect an object layer to each pin of the object layer map, the objects associated with the empty pins will all be on layer 1. What you then can do is only connect an object layer map to the pin of the object you want to isolate. For example in the picture, everything is in layer 1 while the floor is in layer 2:

implicit_assignment.png

cheers,
Thomas

many thanks for the fast replay

Manu

linvanchene wrote:Thank you again for your quick reply.

- - -

I am not quite done experimenting with this but everything points me to the one question:

Would it in theory be possible to have Octane combine the "black shadow" and "reflection passes" in one pass (layer mode normal)? ¨

At least that is the combination I would be most interested in.
For some cases it might as well be interesting to have
"Shadows and reflection"
or
"Colored Shadows and reflection"

each combination in one pass.

- - -

Update/Edit:
The underlying question I have is:

Are the choosen layer blend modes set in stone or can OctaneRender calculate combinations of those passes with a resulting layer blend mode of "normal"?

Black shadows (normal)
Reflections (linear dodge - add)
Colored shadows (multiply)
Shadows (mutliply)

Another example:

Would it be possible to render out

Beauty + Black Shadows + Reflections

in one pass with layer mode normal?

- - -

Just to be clear for some cases you definitely want to have shadows and reflections on separate layers.

But the longer I test with this the more I get the impression that for compositing it might actually be easier for some cases to have black shadows, reflections and maybe even beauty combined in one pass with the layer mode normal (IF that would be possible).

How do others think about this?

Hi,

No, Octane cannot combine those passes in one single pass. I don't fully understand the original question but let me try to answer the underlying question

The blend modes are fixed but we didn't choose them arbitrarily. For example, consider a black shadow cast on a plane. You can think of this shadow as a ratio defining how much of the underlying plane is actually visible (Or the other way around, how much light is actually falling on the plane). More formally: shadow(at point x) = light(reaching point x) / total_light(send towards point x), a ratio of 0 means fully shadowed, a ratio of 1 means fully lit. For black shadows (and matte material), we write this value directly in the alpha channel, alpha 1 means shadowed and alpha 0 means not shadowed. If you then do alpha blending (normal mode) you get alpha * foreground + (1 - alpha) * background. Since the black shadows is only an alpha channel foreground is 0 and the formula reduces to (1 - alpha) * background. And that is why normal mode is used. Does it make any sense?

Some render engines (Mental ray for example), render shadows as the differences between render without shadows - normal render. In that case, the shadows need to be subtracted. We choose not to do this because it requires to render the scene twice (with and without shadows). I also don't know if this works good in compositing on a backplate for anything but black shadows.

cheers,
Thomas