driving me nuts - light leak

[mskin]

this is driving me insane.

 

the geometry is solid. i created a solid, ran the stl mod to check geometry. cpied and scaled it smaller... about 1'4" and subtracted the inner from the outer. anotherwords, the geometry is at least 1'4 all the way around.

 

why is the sun leaking in so badly?

[Macer]

Have you got GI on? If so, try a very small photon radius (less than a couple of hundred mm). I've had problems with large or default photon radius causing similar leakage issues, so thats where I'd start.

[SandmanNinja]

Yes, when your sampling radius is too big it can sample from the other side of the wall.

Having said that, if you leave the min sampling radius unticked, 3ds will use the size it thinks is best.

 

Not sure how you made your walls. I get the wall polylines from autocad and use File Link Manager (with an AutoCAD parameter that I made) to import them into 3ds. I then sweep or loft the wall shape to the wall path.

 

You can select the various objects, go to vert selection mode, Control A (to select all), then hit weld. I always open the dialog box for weld, as it gives you a vert count before and after.

[mskin]

here it is with "maximum sampling radius" set to 10" . seems better but needs work. i will keep playing with this.

[mskin]

i am so badly struggling with this and have blown most of my day. attached are renders with various sample radii on th GI. all other settings are default.

[mattclinch]

how many photons are you actually firing into the scene?

[mskin]

whatever the default was... 500?

 

i like this one, i used a 30' sampling radius and 20000 photons, but i don't know if this is good practice.

[SandmanNinja]

I generally use 20,000 photons to test and fire 150,000 to 250,000 in my final scenes. A thirty foot sampling radius??

 

If you have all your photon 'hits' spread far apart because you didn't fire many photons into your scene, grouping 250 'hits' might be a metre in area. If you have enough photons in your scene (and therefore, plenty of photon hits) you shouldn't need to tell it not to group over a certain value. i.e. don't tick the max photon sampling size. I always untick the box when I'm through testing.

 

The larger the sampling radius, the less detail we get.

 

Here is what I generally use:

GLOBAL ILLUMINATION

[x] Enable

Maximum Num Photons per Sample: 20,000

[x] Maximum Sampling Radius: 250mm

[x] Merge Nearby Photons (saves memory): 0.0mm

[x] Optimize for Final Gather (Slower GI)

Volumes:

Maximum Num Photons per Sample: 1

Trace Depth:

Max Depth: 30

Max Reflections: 30

Max Refractions: 30

Light Properties:

Average GI Photons per Light: 250,000

Decay: 2.0

Geometry Properties:

[ ] All Objects Generate and Receive GI and Caustics

 

I sent you a PM - if you'd like to email me the file, I'll light it for you and explain why I chose the settings that I did.

[Justin Hunt]

"Maximum Num Photons per Sample: 1"

 

why so low, your basically saying only have 1 photon per sample, surely this will give you very little detail. Typically I use between 500 and 1500 for really nice detail.

 

 

jhv

[SandmanNinja]

for volumetric lighting, which wouldn't be used in this particular scene.

[artarc]

Try tune your light start/end attenuation correctly, adjust the start decay position too. You can also choose the object that should or not receive your light. In this way light will not leak to unwanted areas.

Feel better now?

 

Victor

[Justin Hunt]

Got it, The volumetric setting often confuses people. I once watch a tutorial where are significant portion of it the volumetric setting was being adjusted, but no volumtrics where being calculated??????

 

jhv

[SandmanNinja]

yeah - it won't let me zero that field out, so I just turn 'em as low as I can.

[ronyrain]

hi i am new at this but my exp says sometimes changing the light source or the direction of light can solve your prob.

[Cesar R]

I generally use 20,000 photons to test and fire 150,000 to 250,000 in my final scenes. A thirty foot sampling radius??

 

If you have all your photon 'hits' spread far apart because you didn't fire many photons into your scene, grouping 250 'hits' might be a metre in area. If you have enough photons in your scene (and therefore, plenty of photon hits) you shouldn't need to tell it not to group over a certain value. i.e. don't tick the max photon sampling size. I always untick the box when I'm through testing.

 

The larger the sampling radius, the less detail we get.

 

Here is what I generally use:

GLOBAL ILLUMINATION

[x] Enable

Maximum Num Photons per Sample: 20,000

[x] Maximum Sampling Radius: 250mm

[x] Merge Nearby Photons (saves memory): 0.0mm

[x] Optimize for Final Gather (Slower GI)

Volumes:

Maximum Num Photons per Sample: 1

Trace Depth:

Max Depth: 30

Max Reflections: 30

Max Refractions: 30

Light Properties:

Average GI Photons per Light: 250,000

Decay: 2.0

Geometry Properties:

[ ] All Objects Generate and Receive GI and Caustics

 

I sent you a PM - if you'd like to email me the file, I'll light it for you and explain why I chose the settings that I did.

 

I am assuming you turn the FG on top of these setting? I tried you 250,000 photons and 20,000 samples on a small kitchen scene and without FG it doesnt look that great. Then I wonder if you are going to turn on FG then you can lower the sampling cant you?