Here is the next version of my interpretation of Tim Nikias' room. This
image is rendered without radiosity (see below).

Changes have been made to:
- finish/diffuse of the background image to make it slightly brighter.
- wall and ceiling textures; blending more the different colors, adding a
crackle pattern and tweaking the normals.
- texture of the crates; different wood textures underlying the paint.
- a stove in the corner; I googled a barrel stove from Alaska that exactly
fitted my needs, and modelled it in Moray and Silo before exporting it to
POV-Ray through PoseRay. I used two image_maps provided by Nicolas Rougier
for his Chain in this ng, that perfectly fitted the stove.

I experimented with photons to see if some caustics would be reflected on
the ceiling, but results were unconclusive, so I left it for the time being.

I still have to do something about the glass in the windows.

And then radiosity:
I noted that Tim used two different render sizes: a smaller one for the
gathering pass, and a larger one for the second pass. I was wondering how
much the two sizes could differ. A small size for gathering radiosity data
is faster, but to what degree can you then use that data in a final larger
render, e.g. 320x240 for the gathering pass and 800x600 for the final pass?
I could not find info about this in the docs. The point is important to me
because I have a slow computer, I have to increase count substantially to
diminish color banding which tend to hide the more subtle wall texture, and
I encountered some memory problems. All in all, the gathering pass takes
several hours on my machine.

Like always, comments welcome!!

Thomas

Post a reply to this message

Attachments:
Download 'windows_room_03.jpg' (86 KB)

Preview of image 'windows_room_03.jpg'