Light Cove light leaks

[Sawyer]

Working on a new project last week and I ran into some trouble. The image called for a light trough which is a really thin member a 1x5 @ 6' high 3" from the wall. This runs the length of all of the walls and is the main light source for the hall. The light is behind the 1x5 and under it is a base so the light shoots UP and no light should bleed down.

Using Irr map and lc even at high settings I had light leaking down. In the preview of the LC I could see the light leak

I ran down the issue from old posts here I think I found a solution. I clicked sample visability and that almost solved the issue. I ended up with some horrible noise on the reflections. I ended up switching to irr map/ QMC and no leaks but render times went really high. 10 hours at 2500 pixels. I will post some images in a few.

[Sawyer]

Under this light in the cover there is a base that connects to the wall there should not be light below.



I am not familiar enough with QMC as a second bounce. Why is this so splotchy?



What is up with the reflections here in the wood trim?

1000 samples for LC and -4,0 min, max. I am pretty sure these only appeared when I clicked Show Sample vis.

I am sorry that I don't have all of the settings I used for this. I was modifiying this file in an incredible hurry to have it render. Some of these are images I threw out but I don't know what was going on. I have not had this problem before.

[vlado]

It depends on whether the light leak was caused by the light cache or the irradiance map.

a) if is was caused by the light cache: try reducing the "interpolation samples" of the light cache (the default is 10), try 5 or even turning off filtering for the light cache completely (set filter type to None). If this does not help, you can try reducing the light cache sample size and/or switching to "World" scale. If this is unacceptable, in the V-Ray 1.5 Object Settings dialog there is a parameter "GI sample id" or something of the sort (can't remember the exact name); make sure the geometries on the two sides of the light leak have different GI IDs.

b) if this is caused by the irradiance map: try increasing the irradiance map max rate to 0.0, reduce the number of interpolation samples, or if all else fails - use QMC GI.

Best regards,
Vlado

[DaForce]

what are you using for the lights? Vraylights or Vraylightmtl?

[Sawyer]

Thanks Vlado - I will try those. I did change the lc units to world and at .01". Maybe it just needed to be smaller.

Daforce - In these images it was a vray light. I also used a ltmat and I think I had the same problem but the geometry was slightly different not a 1x member but a crown molding.

[Sawyer]

I am updating this file. I may have the original light leaks taken care of. I am not sure what is going on here:



I am using irr map -3, 0 and lc 500 samples world. filtering set to none (this also occured with default filtering)

[pengo]

some questions to understand better wich could be the cause
- are u using solids or surface (foil) as geometry for walls and others?
- wich units are u using ?
- is there some raytracing materials in the scene?
- is the light source behind glasses (of windows orr anyway glasses)?
- are u using high first diffuce bounce multiplier and anyway different from 1 and 1 rispectively in irrad map parameters?
- are u using vraylightmaterials to light the scene with wich settings if yes or vraymtlwrapper with high generate GI value?


no more questions for now

[vlado]

Sawyer, you just need more light cache subdivs and maybe a little larger samples.

Best regards,
Vlado

[mcvltd]

I've had similar problems to your light leaks in the past. I also spent hours fiddling with the settings, but found that it was more a geometry issue.

For the light trough make sure its modelled as a solid with some good thickness and also make sure it intersects the wall (good idea to extend it past the wall by 25mm or so).

This allows the irr map some area to go from directly lit (above the trough), to completely unlit (where the solid trough intersects the wall), to indirectly lit (beneath the trough). Any inaccuracies in the irr map usually get hidden in the 'completely unlit' area.

I also get similar light leaks to your ceiling - for me it usually happens when I've got the background set to a light colour and the verticies of my walls/ceilings don't quite match up (I model in rhino, so have to convert my nurbs to a mesh which, if you haven't joined the suraces together, will produce verticies that dont match up). The only way arround this is to use vray lights in your windows and set the background colour to black.

Hope this helps a little.

Rob.

[Sawyer]

Sawyer, you just need more light cache subdivs and maybe a little larger samples.

Best regards,
Vlado

Thanks Vlado and thanks to everyone else. Yes it was lightcache and I think there was something geometry related as well. This was from ADT. I rebuilt everything the problem was that even with the 3gig switch and turing off frame buffer if I went too high on hte lc settings I ran out of ram. I rebuilt almost everything in the scene and I have it re-rendering now. Before when I used irr map/qmc it took over 24 hours to render at 2k pixels wide now it is almost half that. This was a rush job that I didn't spend the time to rebuild the first time around. Now they need a change and didn't need it immediatly so I rebuilt it. 12 hours still seems long but I guess its because its a not well lit?