"m@b" <sai### [at] googlemailcom> wrote:
> On 21/01/2021 5:54 am, Mr wrote:
> > "m@b" <sai### [at] googlemailcom> wrote:
> >> On 13/01/2021 4:08 am, Mr wrote:
> >>> "m@b" <sai### [at] googlemailcom> wrote:
> >>>> My first attempt at using radiosity.
> >>>> Comment welcome.
> >>>>
> >>>> m@
> >>>
> >>> The beams through media look very crisp and nice. here are a fex remarks in the
> >>> order that would matter most first:
> >>>
> >>>
> >>> There seems to be too much light from inside to have a balance that would reveal
> >>> radiosity well. Could the inside lights be dimmed ? maybe even remove with
> >>> radiosity. The scene would contrast more
> >>>
> >>>
> >>> Some of the colors are too saturated. (srgb? assumed_gamma?)
> >>>
> >>> the reflective elements, show too strong a total reflective amounts. You can use
> >>> conserve_energy and albedo keywords to solve that.
> >>>
> >>
> >> Thanks for the comments.
> >>
> >> I have reduced the gamma, gray_threshold and brightness.
> >>
> >> The steel-work is now a darker pigment and less reflective.
> >>
> >> Getting there slowly, this radiosity thing is complicated :-)
> >>
> >> m@
> >
> > great improvements! Now the small dining table appears as a nice place to rest
> > the wandering eye... Cool details. Now the image is really passed the finished
> > state, but of course, one could always want to iterate more... so :
> > -I suspect the light outside is not parallel  sun like type and in that case too
> > close to the building because it makes beams directions diverge instead of
> > falling in the same direction if they were generated by sunlight.
> > -the red of the machine still slightly over saturated
> > -The mapping of what I assume to be metallic grids on the floor seems stretched
> > or unreadable.
> > -the floor seems too bright, clean and flat.
> > -The walls miss some history too. maybe just a little dirt in the corners,
> > textures allowing to identify what they are made of ?
> >
>
> The sun is 100 million units away and the beams are still not appearing
> to be parallel, any thoughts?
>
> light_source{<-40, 55, 100>*1000000 color rgb<1,1,1>*10 parallel
> point_at <0,0,0>}
>
> camera{ angle 50 location <-24,-15,-50> direction 1.5*z right
> x*image_width/image_height look_at <5, -10, 15>}
>
> I have de-saturated the red and green a bit and increased the normals.
> Added some bozo to the walls.
> Made the floor grids a bigger scale to stop the interference pattern.
> Increased the fade distance on the lamps to give some shadows on the wall.

Wow! The magic is comming in... And now they are indeed parallel enough to my so
unscientifically nitpicking taste :-D

Thanks Alain and Subclick for the great tips, I suggested a location tip
addition to parallel light wiki doc section.

I keep adding critics for constructive feedback believing in a great potential,
but be aware that you are already way above what I would be able to do :-))
Things that would I believe payback the most in that order :

Cracks / planks /tiles asperities in the floor, any kind of normal map texture
of your choice used to give it small shadowed and light reliefs that will
fragment the colored projections currently too neats but at a lower frequency
than pure noise.

I feel the wall could benefit slightly smaller bump scale (its depth is perfect
as appears in the animation. An aoi pattern or any other way to darkening its
corners might pay.

There seems to be some low frequency noise in your radiosity / photons solving,
if that is the case, could you try dividing the photon spacings for wall and
floor objects and increase their importance values ?

The animation is too continuous, and would benefit from a few sunden frame step
stops and jumps (as in an edited cut) when trajectory feels like waiting.

Now the reflectivities are perfect, the relief on the green painted metal looks
awesome... and mostly, the media is so promissing ! thanks for your inspiring
work.

Post a reply to this message