Solaris instance lights and attribute variation

In Solaris we can instance lights onto points.
Make sure your points have  N & Up attributes, like you would with any instanced geo.

The more interesting stuff will be the intensity, exposure, "color" etc.

I know there's a post about this on Matt Estela's CGwiki, but I *think* it applies to an older version of Houdini/Solaris & the instancing method they use there is different - "reference" instead of point instancing. Never sure which is actually better or more efficient!

But how do we access them?? Here's a list of the stuff we can access (taken from the SideFX website)

Primvars

In Solaris/USD, you have access to several light-specific primvars. You can use attributes on the scatter points to drive these primvars. Here’s a list of the available primvars:

• intensity (intensity)

• exposure (exposure)

• width (width)

• height (height)

• color (color)

• radius (radius)

• color temperature (colorTemperature)

• diffuse multiplier (diffuse)

• specular multiplier (specular)

Note that width, height and radius are only available with area lights, while the remaining primvars are available for any kind of light source.

SideFX recommends using Attribute Adjust sops to access these attribs. They're a bit neater and user friendly than wrangles I suppose!

Below is how you use it... plug your points into the Attrib Adjust and type this into the name:

`encode("light:color")`

Don't forget the back ticks - ` `

The same applies to Attribute adjust float, for something like Intensity.. `encode("light:intensity")`

Also.. if you want to do stuff to these attributes in a wrangle (after they've been created), they get given a funky name, but these can be accessed normally

For instance (lol), here i'm multiplying the intensity of the lights with the speed at which they move at in a wrangle.

I'm not sure if the assigned name is the same each time so do check... 

plate camera projection in Karma or Solaris

How to do camera projections in Karma/Solaris.
 
Coordsys is the magic node. Below are the settings.. I want to project a plate image onto /grid using /camera1... so put their names in the boxes shown.
"tex" is an arbitrary name, we'll use it to access the camera space coordinates later, in the material. This is all that NDC stuff.
Just in case the images vanish from this site... Primitive Path should be your grid/projection geo, same again for the Primitives. Coordsys name should be something like "tex" and Target should be your camera path. Everything else can be left as default.

Material X Position is the node that gets you access to the NDC stuff..
We call on "tex" and the ":ndc" gets us to that space. Following that we get a vector, where we only need the X and Y...........which we separate out and combine to make a UV coordinate for the image sequence.


Assign your material to the projection object/grid and then switch to a Karma render view to see the result (VK does NOT display this correctly)!
If it doesn't work...check all your name paths!

render var for aov, karma, solaris

 
This seems to be the least painful way to make an AOV - using a "render var" node just before the Render Settings.

I've named the node itself "pref", so a $OS can pick up the naming.
Under the Standard tab, set the data type and source attribute name. My attrib is called "rest" and it's a vector 3 float. The source is Primvar, which means it comes from the mesh.

Next, on the Husk tab (this might be located under standard, depending on your Houdini version), the name box is what you'll want to call your AOV (here we use $OS to take the node label, pref in my case) and then the format that we are writing out - color3f (rgb).

screen space depth, ndc, camera space, flattening stuff

 

Pinched this from Tomas on the houdini forums.

This is useful for moving stuff in and out of camera space, but maintaining the "2d" flat appearance of an object.

Plug this into a point wrangle- 

string cam = "/obj/cam1";
@P = toNDC(cam, @P); // to ndc space
@P.z *= 0;  // flatten
@P.z -= chf("depth"); // set depth
@P = fromNDC(cam, @P); // if you need bring back in front of cam

i've changed the set depth to a float slider to be a bit more interactive, you could just put a hard value in.

the fromNDC(cam,@P) part is the line that brings your object back in front of your camera...otherwise toNDC just exists somewhere near the origin. 

the @P.z*=0; to flatten is entirely optional... you can use this script to move things around in "camera z" 

rendering with karma/solaris in a more "normal way"

 Just a way to work as if you have a render view, instead of in the viewport.

 In Stage, make a Clone Control Panel tab (where your parameters are)


click the Clones with a arrow button, make a new Local one, select Karma CPU or GPU..
Drag your karma render settings node onto the "Node".
Now you can toggle a render on and off in a Render Gallery window & it will save a snapshot each time (might be good or bad....).