Standard glass not allowing lit interior of wall cabs to render with realism

[GeneDavis]

Rene Rabbit did a helpful video on this.  I watched it to prep for doing it in a kitchen for a client.  I want to show them the option for doing glass doors and inside-cab LED strip lights:  

 

 

So I do the same thing for a pair of wall cabinets, or try to do the same thing, and get not-so-good results.  Both cabinets have material "Glass, standard" for the doors and the single shelf that aligns to the horizontal door muntin.  The cab on my R has four martini glasses, material "glass, standard" from one of Chief's bonus libraries.

 

There is another difference.  I did my light strips from scratch, making them from solids 1/16" thick x 3/8" wide x length.  Rene starts with Chief's rope light (which is also a thin flat strip and not a tube), so I think the light fixtures are similar enough.  Mine are outfitted with spot lights 1" apart.  Rene does 3" spacing which seems to be what Chief's OOB rope light is.

 

My problem is with the door (and maybe shelf) glass material.  Here is the scene in standard camera render.  This and all that follow have the night setting for outdoor lighting.  No backdrop.

 

 

I should explain another detail as re lighting.  The cab on R has martini glasses in lower cab and up on the mid-shelf.  Lighting in-cab in both is the vertical strips as Rene does in his video.  The L cabinet (no glasses inside) has an additional light strip across the top.  Here is the scene using CPU RT rendering.  All lights are set to 1200 lumens, just as Mr R did, but remember his are at 3", mine at 1".  Too much lighting as seen here.

 

 

And too much lighting is generating the bleeding.  I think I can fix that by dimming the strips and adding a little more interior lighting brightness.

 

But it is the RT RT that yields the bad results, and it's RT RT that Rene is using with success.  Look here.  Both cabs have glass in the door panels.

 

 

 

Rene's doors have glass panels and he does not have this issue, and he has less lighting inside.  Why?

 

I dim the lights to 800 and try another RT RT, but change the door glass in the R cab to "opening no material."  Of course I see the lighting, in that one.

 

 

What magic trick is Mr Rabbit doing with his glass?  All I want is something like this.

 

 

 

 

 

 

[Flash691]

I was having the same issue until I watched his video again. That's when I noticed at the very end, he went in to the "Adjust Lights" menu. It's on his toolbar, but you may need to click the light bulb icon and then Adjust Lights on the left side toolbar. Do this while in camera view. CA defaults to 20 lights on which is easily covered in a couple rope lights. Select "Light set" from lighting mode and then all the lights you want illuminated and click done.

 

 

[GeneDavis]

@Flash691 all the lights are on in all the renderings I posted.

 

For my particular model, and the way I want the lighting to appear, CPU ray tracing gives me OK results, while I am unable to achieve any acceptable results using real time RT.

 

Here is a pretty good result, with the doors glazed in a material designated "glass standard."  Rendered using CPU RT, the light strips inside the cabs all have light spacings of 1" (as will the stick-on LED tape that will be used), and each light emits at 25 lumens.  I cut the output of the rope lights (Chief's at bulb spacing 3") under the wall cabs from 200 lumens each to 100, and I think the backsplash and countertop look more realistic with the lower setting.

 

 

With all light settings same, here is what RTRT produces.  Yuk!

 

 

I wanted to study this situation of light sources inside cabinets behind glazed doors, so I built a one room house and a 30" cube, the cube sheeted in 1/8" thick "glass standard" material, raised the cube off the floor 25", and put two light sources inside the cube.  One light is a point, the other a spot.  Both are set to 1200 lumens, the spot has a beamspread ("cutoff angle") of 140 degrees and a falloff ("dropoff") rate of 15 percent.  Here is a scene done with RTRT.

 

 

As is seen, the light from both the spot and the point get through the glass walls of the cube.

 

Here is the scene using a few passes of CPU RT.  I regret not painting all wall and ceiling surfaces with the same material, maybe a matte gray, just a little gloss, but it is pretty obvious the point light is getting through in all six planes of cube to illuminate the room and its surfaces.  It's interesting how the cube form with sheet glass on each face affects the projected light.  Look at the floor.

 

 

So in conclusion, at least for me, it seems that one has to choose to work with whichever rendering technique, RTRT or CPU RT, produces the better realism for your particular model.

 

 

 

 

[Flash691]

If you share one of the plan files, I'll try it on my system to see if the results are the same as yours. 

[GeneDavis]

@Flash691 here you go.  I added a solid about the size of a sheet of paper, inside the glass box, directly in front of the planview right light source spec'd as a spot.  It is on its own layer and can be turned off or on as desired.  My renders posted upthread were done before I added the paper sheet, so turn off that layer to try renders to see if you get what I got.

Light box house.plan

[Flash691]

@GeneDavisYes, I agree with your conclusion to work with whichever method gives you the results you want for your particular model. I'm not a rendering expert like @Renerabbitt by any means. Here's some comparisons rendered on my system. I haven't spent much time testing the different render settings available for CPU_RT or RTRT. I posted images of the settings I used at the end.

 

1. RTRT with only the spotlight on.

 

2. CPU_RT with only spotlight on.

 

3. RTRT with only point source light on.

 

4. CPU_RT with only point source light on.

 

5. RTRT with both lights on.

 

6. CPU_RT with both lights on.

 

The are the settings for RTRT

 

CPU_RT settings:

 

RTRT took about 3-4 seconds to render at 3008X1940px using an RTX4090 GPU.

CPU_RT for 10 passes took about 3-4 minutes at 1500X967px on 13th Gen Intel(R) Core(TM) i9-13900KF   3.00 GHz CPU.

 

[Renerabbitt]

Ha! 
Working through it backwards.
As fun as that probably was to experiment with we are missing some core knowledge here. Chief is using a whitted raytrace for RTRT, single light trace, so whereas the results here are easily manipulated, you have to understand whats happening in a scene in order to get the results you want. Which is to say, these aren't repeatable results. the same light in another scene is going to do something different. This ends up getting into a very complex conversation about rendering, truly. Something I usually offer training on.
Throw in cuastics and refractions and you have a ton of calcs that are happening in all of the surrounding surfaces of a cabinet box that is lit from the inside. Single surface bounce, so based on the pixel density of your screen, the angle of your cabinets, the color and supporting maps of your materials, the index of refraction on your glass, the backdrop you used and its intensity and so much more, you will get different results from scene to scene. 
For instance, if you have overhead lights hitting your glass cabinets, and then lights inside of your glass cabinets, chief kinda sorta has to make up the decision on wether the pix of the material inside of the cabinet is lit from those overhead lights or the lights insisde of the cabinet. Move closer to the cabinet, like real close, and chief might make the decision that they are lit from the interior lights. 
This does get way more complex though, and I wouldn't spend too much time on it until you know what the next version brings

[Ryan-M]

So, this is basically a bug.

 

The way that we apply lighting to surfaces viewed through a glass (or any refractive) material is different than the way we apply lighting to surfaces that are directly visible.  In X13 - X15 there is a fixed limit on the number of lights that we apply to these kinds of surfaces.  This means that some of those lights inside of the cabinet may not actually produce light on the surfaces visible through the cabinet glass.  I can certainly understand that this is confusing and not the ideal behavior.

 

This is unlikely to be problem in newer versions of Chief.

[Flash691]

13 minutes ago, Renerabbitt said:

we are missing some core knowledge here.

Indeed, quite an understatement in my case. 

 

29 minutes ago, Renerabbitt said:

This does get way more complex though, and I wouldn't spend too much time on it until you know what the next version brings

Hopefully there will be a public beta soon.

 

30 minutes ago, Ryan-M said:

In X13 - X15 there is a fixed limit on the number of lights that we apply to these kinds of surfaces

That does explain the issue@GeneDavis is experiencing.