Q1: B
Q2: A
The box room appears to have evidence of construction in those two images.
The overall image results are very good and as a thumbnail it’s very hard to tell which is real and which isn’t.
B & A
Please please please tell us
That is so cool! The ultimate reference for the cycles developers. I’m impressed how close it is. Now - add something with SSS and some smoke and they can see how well they are doing 
Q1: B
Q2: A
Very good job, it requiers carrefull examination to spot the differences between cycles and reality…
At first, thank you so much for a lot of answering and comments!
The long-waited-for answers are… B is real for Q1, and A is real for Q2!
As I thought, most of all people have great eyes!
Then, where to look is below…
Q1 B is real
Some people seem to notice the imperfection of the cornell box.
The box is made of alminum plates and frames, and has small gaps at the corners.
Besides, the area light is included as LED array in image B! That’s my mistake.
The area light is made of LED array (can adjust intensity) and translucent acrylic plate. All of stuff is DIY.
Q2 A is real
Taking the result of Q1, I made some fakes in Q2.
I kept the LEDs from acrylic to diffuse, and made a gap and lint.
That’s a little spiteful, but many guys have great insight!
Finally, I think Cycles’ one tends to be low contrast image, and fresnel factor is important.
Of course, I used fresnel nodes to make shading better, but not enough.
Thank you everyone! 
Thank you for this, LeoMiyashita.
I had noticed because the box, not because the render. I am pleasantly surprised how accurate and realistic is Cycles.
This is remarkably well executed. You are good at building cornell boxes, even better at replicating photos in blender. Won’t even try to guess… that’s a first for me.
We’re still waiting.
The answers were given 3 posts before you. (Post #25)
Thanks Leo
Great work. Really realistic renders.
Interesting idea! I recognized the renders especially becouse of the residual noise on the top corners compared to the clarity of the real ones; another little thing it’s that seems like in the real one there is a bit of glare/aberration at the borders of the light source on top.
I think both B are made with camera or this is a trap question and all 4 images are rendered inside cycles 
There’s a new question now - Q3: Who has read the WHOLE thread?
I did but as far as we could guess the initial statement that 2 of the images are made with the camera could be false.
I did but as far as we could guess the initial statement that 2 of the images are made with the camera could be false.[/QUOTE]
The answers are in post #25, which is titled: The answers
Incredible work and attention to detail, both with cycles and the physical construction!
Great work! You can also tell which one is fake from the noise on the images. Even with 10000 samples there’s still some noise visible
Thanks very much posting the answers Leo!!!
That is four fabulous images that you produced.
Now…for the prizes…there HAS to be prizes…what do we get for winning?? Let me guess!!! A free copy of the next version of Blender!!! AM I right, am I right??? Oh!!! Please let me be right!!!
What an awesome idea! Maybe it would be worth building one on my own. And very impressive to see what Cycles is capable of. Actually I am also impressed by that fact that you have the Utah teapot sitting around at home