Hello guys,

I have a scene with a lot of particles in it. It is been lit by three spot lights. The camera is on top of the particles.

When i render my scene, i get the following output. Can someone please tell me whats the problem and how to solve it.?

Hello guys, Can someone help me solve this issue..? Please…

Hi,
The problem could be the value of the cone angle parameter in the spot lights.

also turn each light on and off to check which light is causing the problem.

If the problem resists with all lights, it might be something with either your shader or perhaps attributes that are coming from the particles that are causing that strange shading (perhaps N).

Try perhaps moving the obj node of the particles around, just to see if this changes anything.

also turn each light on and off to check which light is causing the problem.

I have already tried turning on and off the lights and replacing the spot lights with directional lights. Nothing helped me really.

perhaps attributes that are coming from the particles that are causing that strange shading (perhaps N).

And i just noticed that my particles doesnt have an attribute called normal. But the emitter that emits it has normals and i have mentioned the particles to inherit the attributes in the source pop under Attributes tab. Wont that inherit the normals from the emitter.? If not how to go about assigning normals to these particles.?

Thanks a lot in advance.

try assigning a constant shader to it.
See if it is the lighting model of the shader (sprites?) that is causing it.

Try to make it so the attributes that are going to the shader are clean/non-existent.
You can delete the N attribute or assign it to (1,0,0) or a value that is facing the camera.

It almost looks like there is another object in the scene blocking light.

If you position a default sphere at the same position, does the shading of the sphere behave strangely as well?

If you position a default sphere at the same position, does the shading of the sphere behave strangely as well?

i tried with sphere and also another set of particles. Sphere rendered out well but the particles dint. The same prob existed with particles.

And i try assigning a constant shader and then normals and will let u know. Thanks a lot peter.

try assigning a constant shader to it.

The problem was not there when i applied a constant shader to my particles.

You can delete the N attribute or assign it to (1,0,0) or a value that is facing the camera.

I assigned the normals to (0,1,0) since the camera is on top of the particles but the problem is still there.

so it sounds like something in the shader is going wrong. Perhaps a division by zero, or negative values. It really does look like a radial effect. Is the center of that spherical dint the origin?

Try simplifying your shader and piping the different lighting calculations in the output color. That way you can narrow the problem further.

I don't know the scale of your scene, but could it be a camera clipping plane problem?

Perhaps your camera is slicing the closest particles away. Whereas the shadow is probably a shadowmap and is calculated for all the particles.

Peter, I have attached a simplified scene file in my first post. It would be very helpful if you could see it and tell me is that the camera's clipping plane that causing this issue..

The easy answer is to simply turn off the object option “Render as Points”.

You will need to append a Point SOP and enable then increase the Scale parameter like:
$PSCALE * 10

Instead of disks you now have primitive spheres at each point and it kinda looks like an issue with the geometry normals.

Export the normal plane in a Mantra ROP (which you should always use btw). Just inspect those normals in mplay using the white and black point options as mplay sometimes auto-adjusts to the full range: -58 to +69. That seems kinda high even when the lighting models normalize these before computing illumination…


I'd still submit in to support as a bug to see what they say.

Another way is to change micropolygon rendering to raytracing in the Mantra ROP's render engine and leave the render as points option. Normals are built properly then.

Thanks a lot Jeff. Turning on raytracing in the mantra rop did the trick.

And can someone tell me how to export the normal plane to the mantra rop.?

I think i found out how to export the normal plane. I did just by adding an extra image plane in the mantra rop and mentioning the VEX variable as N..?
Am i right.?

just something to be aware of: when you turn on raytracing you will not gain the benefits from a procedural (like clusterThis) anymore in terms of memory efficiency - as the geometry is now kept in memory rather than thrown out when a bucket is finished with it. But not sure if you are using it for this.

No Peter, im not using clusterThis because i couldnt find a dso that was compiled for houdini10 and win64. If someone in this forum has one and ready to share, it will be of great help.

Raavenish
I think i found out how to export the normal plane. I did just by adding an extra image plane in the mantra rop and mentioning the VEX variable as N..?
Am i right.?

Yes that is correct. I usually export a lot of shader layers when I am debugging a shader.
Have a look at some of the default Materials. In Houdini the default convention is to color all the export Parameter VOPs Green. You can export any part of your shader tree VOP network in to an export plane. It's very accessible too.


The issue is submitted as a bug wrt micropolys and the normals.