Lighting in archviz

[zooropa]

Hi, I am trying to achieve my first render interior.
I posted in another topic related to this, but I think it got lost there.

Iam trying to illuminate an interior that has only one opening in the larger side of the "rectangle" (glass window).
I can post the scene if needed).

Have a few questions regarding how to illuminate the space.

I tried with an HDR, but seems it is not effective in terms of intensity. So I decided to add a rectangle plane and assign an emissive material to boost the lighting. Still it is not strong enough to mimic the picture I am trying to pair with.




I can keep cranking the emissive values, but then everything burns close to the window area. It is true that in the pic above...also happens that, guess is physics.

Scene so far:


Do you have any recommendation of how to illuminate the interior closer to the real picture?
I am using interior mode with  16 ray bounces.

I also could add a different area light behind my camera, but then I will need to hide it when I change the camera angle.
Which will not allow me to have an similar lighting with different cameras.

Summing up, I am wondering If I can mimic the light from the picture only with one HDR ? The emmisive material helps a lot, but then I have the burnt area close to windows and also I am not sure If I can erase that surface in postprofuction in order to add a background layer with an environment image.

[mattjgerard]

I'd say that the lighting values are actually pretty similar between the 2. One tip that someone put out there a while ago was to turn both images greyscale and look at the values there. I did that for yours, and similar areas of the images have similar 0-255 grayscale values, so as far as light distribution, I think you are mostly there. The outside will almost always be photoshop work, so that whole side of the building will be cut out and replaced, and the burnt areas around the interior of the glass will look natural.

Now, one thing you might try is rendering in 32bit PSD. Going to be a huge file, but there will be much more flexibility in post as far as pulling back some of that detail in the blown out areas that might be possible. My suggestion would actually to do the opposite, to render it "underexposed" and then in photoshop you can tweak the levels and curves to blow out the areas that you want, and the shadow detail should remain. If you render it with areas blow out to 100%white,  there's no telling if that area will be recoverable in post. This is my experience with high bit depth video, is that slightly underexposed is better than over, since the codec can handle the detail in the shadows better than loosing information in blown out areas. That knowledge just comes with trial and error as to which is better.

Keep working on your model, get the geometry in place and locked down, then tweak lighting more. If you have a rows and rows of tables with reflective tops, that will change your lighting as well, and so on.

Good luck!

[zooropa]

I'd say that the lighting values are actually pretty similar between the 2. One tip that someone put out there a while ago was to turn both images greyscale and look at the values there. I did that for yours, and similar areas of the images have similar 0-255 grayscale values, so as far as light distribution, I think you are mostly there. The outside will almost always be photoshop work, so that whole side of the building will be cut out and replaced, and the burnt areas around the interior of the glass will look natural.

Now, one thing you might try is rendering in 32bit PSD. Going to be a huge file, but there will be much more flexibility in post as far as pulling back some of that detail in the blown out areas that might be possible. My suggestion would actually to do the opposite, to render it "underexposed" and then in photoshop you can tweak the levels and curves to blow out the areas that you want, and the shadow detail should remain. If you render it with areas blow out to 100%white,  there's no telling if that area will be recoverable in post. This is my experience with high bit depth video, is that slightly underexposed is better than over, since the codec can handle the detail in the shadows better than loosing information in blown out areas. That knowledge just comes with trial and error as to which is better.

Keep working on your model, get the geometry in place and locked down, then tweak lighting more. If you have a rows and rows of tables with reflective tops, that will change your lighting as well, and so on.

Good luck!

Thanks man! I very much appreciate your feedback. I did not know the greyscale value tip! It might be also the tint of light in the picture also differs from my render. It seems the real pic has a colder light (blueish).
I am developing the model in to se how it behaves.

[mattjgerard]

I think it was either Will or Dries that made the greyscale tip. But yeah, I use that all the time now.

I would agree as well, the color temps and are a bit different between the 2, so that will influence how it looks. The bluer lighting will usually appear to be brighter than the warmer lighting, perceptually. The actual values may be the same, but its the color that changes how its perceived in brightness. Warmer headlights of the same lumen measured output will usually be perceived as not as bright as the cooler white headlights of the same measured output.