Problem : Plane wall compuonded from 2 or more faces.Radiosity doesn`t recognize that faces lies in the same plane.So there is evident border between faces,(even if they have same material of course).Playing with radiosity buttons can improve final solution a bit,but anyway, it seem to me like illogical.
Solution ?: Radiosity could take faces(with same material,vertex orientation,etc.) in the same plane like one face. [/b]

show us the model with the problem

I think it is an issue with the mesh

http://www.kalanovic.szm.sk/

these faces need to be connected in some way, are they not?

also, I often force seams by selecting the flat/rounded regions and splitting them from the surrundings in edit mode (y key, then confirm or exit grab mode) so that the entire mesh can be set smoothed, which will reduce the vertex count after the radiosity calculation (though not the face count), and allows better control over where the seams are

Having taken a quick look at your file, I agree with zero3d: you do need to connect the faces in some way.

Only faces who share an edge will appear continuous.

Thanks.I`ll play with it.BUT it would be much better if radiosity will working "independently" from exact
mesh model, and make its own seamless model for counting.There is so much redundant work.

Clovek wrote:

Thanks.I`ll play with it.BUT it would be much better if radiosity will working "independently" from exact mesh model, and make its own seamless model for counting.There is so much redundant work.

radiosity doesn't work that way

(other more caustic response removed)

Of course radiosity doesn`t work that way now.But it could be.Example: Bigger face could be split(just for radiosity patch) according to the smaller adjacent face in the same plane,and without any seams.Where is the problem?

(don`t be sticky lord z3r0_d pls)

Clovek wrote:

Of course radiosity doesn`t work that way now.But it could be.Example: Bigger face could be split(just for radiosity patch) according to the smaller adjacent face in the same plane,and without any seams.Where is the problem? (don`t be sticky lord z3r0_d pls)

because that is difficult to do, and shouldn't be necescary

it isn't difficult to make a mesh without that kind of problem

(sticky lord?)

clovek - you may be right: however it would probably take a complete reprogram of the radiosity algorithmn to fix I think, given the way the solution is stored "per vertex" at the moment....

On the other hand, these issues are pretty easy to avoid most of the time: first of all, make sure you have removed all double vertices! You had 114 in the mesh in the middle of your scene, and removing these gave a pretty good improvement already.

Secondly, if you have doors etc, try modelling these into the mesh instead of making them from separate blocks (this is difficult to explain but easy to do!). But anyway, it took me ~1 minute to fix the section in the middle of your scene with the knife tool - here's my version!:

http://reblended.com/www/leon/Radiosityproblem.blend

Hope this helps a bit anyway!

cheers

Leon

Thak Yuo so much! I`ve got it.

Pity that knife tool havn`t precision CAD option.(....or I didn`t found it...)
(something like: I want to split this face from this point(x,y,z) to that point(x,y,z)).Grid is not useful for everything.(or not so fast actually)
I`ll hope that "CAD-python team" improve that, or add something helpful.