Lighting model interior with clipping plane

[Farid M. Noriega]

I would like to ask how to I can light my model interior after adding a clipping plane.

Should I add a light or is there a method to introduce light in the seccioned interior.

Thank for all

[Ros. Riha]

clipping pane means, that lighting of the model behaves as if clipped parts were still there, you only see through. So you need to add light fixtures to your interior in order to get some light in there.

At this moment, I am working on the same thing. See attached. There are 76 lights in the model, divided into 8 channels (I would love to have more of them channels available).

Issues an questions so forth:

• editing channels in the nxt image editor means playing in the range of 0.0001-0.0005 ... either my model is wrong or I am doing some mistake, but once I turn down the natural light, everything starts behaving overly sensitive... (I am happy to send the nxtimage, but it is 46MB. I can send the mdels too)
• Without lights, the interior seemed too dark - on walls, there is no material defined - should I use some small reflection on them? If no material is defined, what material is used? What are its characteristics?
• Does "Studio Brightness" affect pathtracer, or not?
• How far channels capabilities go? Can I expect to get both day and night images from the sae rendering or should I change the lighting setup somehow?
• What are reliable photometric lights? Where can I find them? I have plenty of IES files on my computer, but they behaved strange...

[Roy Hirshkowitz]

Ros.-- good job answering the question.  However, try not to introduce other topics into a simple thread.  It makes things like searching more difficult.  Start a new discussion instead.

Something's probably wrong with this model-- downloading the nXtImage now-- maybe I can spot it. 

The default material is ~0.7 grayscale, which is about 50% matte reflective surface. No specularity.  No-- don't use specular reflection on the walls unless you want specular reflection on the walls (glossy paint, etc.)  Studio brightness does affect the path tracer-- don't use it for architectural applications. Yes-- you should be able to easily do both day and night images simultaneously in the same rendering-- it's my default way of working now.  Most manufacturers have accurate .ies data for their luminaires-- look around or ask them.

 

 

[Roy Hirshkowitz]

So I took your drawing and set all the lighting channels back to 1.0 where they belong.  I then set the daylight channel (channel 0) to 0.0.  I had to reduce the brightness to -0.37 -- not unusual for night-time exteriors.  (Even though you may consider this model to be a night-time interior, it more closely resembles a night-time exterior to the tone mapper.)  Your luminance values are a little low-- but not unreasonable.  Here's what I got:

 

[Ros. Riha]

thanks for your advice - it did not occur to me to do it this simple using brightness...

Low luminance values: As I am not used to use brightness slider (I thought it should be kept at 1), I did decrease power of light sources - maybe too much.

Night-time exterior: You are probably right, on images such as this I would like to show both interior and exterior

 

I am still struggling with Sky channel - even during the night there is some light from the sky... as you can see on the printscreen below, even if I set sky channel to the lowest value available, there is too much light - or do I misunderstand something again?

Using settings as these I would expect to get nighttime rendering instead this late afternoon one...

[Roy Hirshkowitz]

If you're shooting for a moonlit interior you have a difficult assignment on your hands.  Use the brightness slider if you must.

 

Here's a more rational example.  I set the sky to 0.001 of it's daytime value-- not out of the question for night.  All the other Channels are set to 1.f and the Brightness is reduce to -.37 (as before) to account for scotopic light levels.

 

 

[Ros. Riha]

this looks closer to what we want to get... in the meantime I arrived to the similar result using brighness closer to 0 and light levels different from 1 - which way is better in terms of learning real facts from the simulation given by nXt? For example: would it be possible to state some recommended brightness levels for different tasks (nighttime, daytime... from dusk till dawn:-)

Yes, I already noticed that moonlit interior is not as easy as I orriginally thought... would it not be better to use some nighttime HDRI for such a specific task? Could you recommend something properly calibrated? Could you even add some other HDRIs to the default installation package? Like "nighttime", "dawn", "dusk"?

[Roy Hirshkowitz]

It's always better to use realistic lighting values and modify the exposure setting.  That message has been consistent throughout and hasn't changed.  You may end up with a good picture doing the reverse, but the data will be garbage.

 

You don't need to alter the brightness much for anything but exterior night scenes (which is what I consider your model to be.)  In those cases I would start with -0.35.

 

I haven't really looked around at night time hdrs-- but they'd be pretty easy to calibrate provided they capture a decent dynamic range.  If you find a good one you can pass it to me and I'll let you know what scale factor to use.  I didn't mean to say that it's hard to combine night background luminance and artificial lighting-- that's very easy.  It is difficult to do an interior scene lit only by moonlight-- although probably quite possible.

 

 

 

[Peter Milner]

Yes, it's possible.

[Roy Hirshkowitz]

Nice.  I would probably add a little more desaturation to mimic what happens during night vision.  Sky still looks a little too blue to me.

[akinlolu olugboji]

What is the setup for this scene? Lighting Scheme, Tone generator term etc

[Peter Milner]

Normal daylight setting with cloudiness=0.6 and sky brightness=0.5.

I used a dark blue as a background colour (could use black).

I turned down the brightness, contrast and saturation and finally added a bit more blue in Photoshop.

[Roy Hirshkowitz]

It might be possible to do it with accurate data as well.  Moonlight is technically about 500,000 times weaker than sunlight (0.000002) .  The problem is that you may run into some threshold illumination levels with nXt-- I believe that 0.001 cd/m^2 is as low as you can go.  Luminance in direct moonlit areas with reflective surfaces should still be in the range of 0.04 cd /m^2-- that's ok-- it's everything else I would worry about.