I have to say it looks good to me! Has a nice clean almost air-brushed 
feel that distinguishes it from other images.

Though if you want the lighting to look more interesting, at the cost of 
looking more conventional, maybe stick a highlight light in there, i.e. 
one behind all the objects at a shallow angle so it adds some 
rim-lighting, particularly effective if you have some specular on the 
objects.

But I'd say just work on the details of the scene, I like the lighting.

Tek


Tom York wrote:
> This image was rendered with radiosity in two passes. There's an area light
> behind the camera, well off to the right and up (to cast some soft
> shadows). The light sources for radiosity are the sky, some yellow panels
> below the floor (they're responsible for the yellow shine on the underside
> of the exhibits) and a blue panel in the ceiling (you can see a bit of that
> panel).
> 
> Antialiasing used was method 1 with threshold 0.0 (anything more and those
> white lines on sharp features become more obvious and abundant).
> 
> What would be the best way to improve the lighting now? It's quite flat, but
> I want to keep a "daylight" look to it. Getting rid of those radiosity
> artefacts in the corners might be nice, but not if it involves
> substantially increasing render time (this already took about 15 hours, due
> mainly to radiosity). Radiosity settings used were:
> 
> First pass:
>    pretrace_start .08
>    pretrace_end .001
>    count 600
>    error_bound { .05 adaptive 1.5, 20 }
>    adaptive 2
>    recursion_limit 2
>    minimum_reuse 0.01
>    save_file "tvo.rad"
> 
> Second pass:
>    pretrace_start 1
>    pretrace_end 1
>    always_sample off
>    error_bound { .25 adaptive 1.5, 20 }
>    recursion_limit 2
>    minimum_reuse 0.01
>    brightness 0.75
>    load_file "tvo.rad"
> 
> I think these are based on some settings posted a while back in these
> groups. The geometry is mesh-based. There is a larger version of the image
> available here:
> 
> http://www.zubenelgenubi.34sp.com/temp/tvo_lrg.jpg
> 
> (1280x960, about 130kB)
> 
> 
> ------------------------------------------------------------------------
>

Post a reply to this message

Tom York nous apporta ses lumieres en ce 2005-07-19 14:35:
> This image was rendered with radiosity in two passes. There's an area light
> behind the camera, well off to the right and up (to cast some soft
> shadows). The light sources for radiosity are the sky, some yellow panels
> below the floor (they're responsible for the yellow shine on the underside
> of the exhibits) and a blue panel in the ceiling (you can see a bit of that
> panel).
> 
> Antialiasing used was method 1 with threshold 0.0 (anything more and those
> white lines on sharp features become more obvious and abundant).
Have you tyied method 2?
> 
> What would be the best way to improve the lighting now? It's quite flat, but
> I want to keep a "daylight" look to it. Getting rid of those radiosity
> artefacts in the corners might be nice, but not if it involves
> substantially increasing render time (this already took about 15 hours, due
> mainly to radiosity). Radiosity settings used were:
> 
To help reduce the artefacts you can try using 1 or 2 more pretrace steps, it can help
somewhat. Try 
pretrace_end 0.000625 or 0.0003125
Increasing count normaly help, but I'm not sure that it can realy help in this case.
Using a smaller low_error_factor than the default (0.5) may help in some case. It only
affect the 
last pretrace step.
The artefacts are mainly related to your yellow pannels, maybe some tweaking in there
finish, 
pigment or dimentions and placing could help. Some suggestions:
- make the pits deeper and lower the panels.
- make the pannels a bit smaller than the pits. or enlarge them to make them larger
than the pits.
- a little lower ambient with a higher saturation for the yellow pannels.
- make the yellow pannels transparent and put some weak yellow area_light spot_light
under them.

To Joanne Simpson, if the air is very clean and dry, there can be no notable fog at
all. In a 
futuristic space museum this would be the case: you cut on cleaning and get rid of
corrosion.

Alain

Post a reply to this message

New go, but to get it I altered assumed_gamma from 1.0 to 2.2 (my
display_gamma is correct for my monitor at 2.2). With assumed_gamma 1.0
(which I know is the "correct" setting) it seems quite difficult to avoid a
bright, flat lit image for some scenes (including this one). There is a
larger version at

http://www.zubenelgenubi.34sp.com/temp/tvo2_lrg.jpg

(1280x960, 150k)

"Ross" <rli### [at] everestkcnet> wrote:

> I don't know the best way, but I think a very strong bright light comming
> through the top on one side, like a sun at 2 o'clock, could help give some
> starker contrast. nice scene, you really achieved a sense of scale.

I've added another area light "high". Thanks for this suggestion, I think it
improved the shadows a lot!

Post a reply to this message

Attachments:
Download 'tvo2.jpg' (72 KB)

Preview of image 'tvo2.jpg'

"s.day" <s.d### [at] uelacuk> wrote:
> Nice image, I think you are right that it is too light intense which makes
> the image look a bit flat, especially the two white panels in front of the
> ships, they almost look 2D have you used a high ambient setting for those
> or rgb values > 1? I assume you are using ambient value of zero and light
> fading I find this helps to add realism to the lighting. Still it is a very
> good image I like the way you have shown the scale with the people (are
> they blobs?)

Yes, for those panels I used a large ambient value to make them cast light
in the radiosity step - completely pointless as I realise now, because
there's very little geometry facing them. I should have dialed it down in
the second pass but didn't. I've now fixed that. Not too sure about light
fading - I think the "unreal" flavour in the first image is mainly
contributed by the extreme flatness of the radiosity, and I don't want to
weaken the (area_light) sun too much.

The people are meshes of about 700 hundred polygons each. I've tried to keep
them back from the camera, but I wanted something to add scale and break
the symmetry a bit.

Post a reply to this message

"Joanne Simpson" <cor### [at] onewhiteravencom> wrote:

> ditto - maybe use a lower rgb, e.g. 0.6 and a rough normal to reduce the
> flatness. Also a chamber that size would have a visible atmosphere. I would
> use a little ambient fog to soften things further away and give a sense of
> distance.

Yes, I've reduced the ambient, but those panels still need work. I've added
a little fog to see how it runs in the newer image, but I think I
underestimated the amount needed to get good depth cueing. I'll turn it up
in the next attempt.

Post a reply to this message

Tek <tek### [at] evilsuperbraincom> wrote:
> I have to say it looks good to me! Has a nice clean almost air-brushed
> feel that distinguishes it from other images.
>
> Though if you want the lighting to look more interesting, at the cost of
> looking more conventional, maybe stick a highlight light in there, i.e.
> one behind all the objects at a shallow angle so it adds some
> rim-lighting, particularly effective if you have some specular on the
> objects.
>
> But I'd say just work on the details of the scene, I like the lighting.

Thanks! I haven't given up on the airbrushed look, and I might well do two
versions. I really imagined this scene more as a kind of architect's line
drawing or sketch, but it seems to have taken on a life of its own.

Post a reply to this message

Alain <ele### [at] netscapenet> wrote:

> Have you tyied method 2?

I hadn't, but following your suggestion I tried it on the new image. It
seems rather better at dealing with bright speckle artefacts (not visible
in either image, I hope) with lower threshold settings. I only used
threshold 0.3 for the new image, which is why some of the posts surrounding
the exhibit pits have disappeared.

> To help reduce the artefacts you can try using 1 or 2 more pretrace steps, it can
help somewhat. Try
> pretrace_end 0.000625 or 0.0003125
> Increasing count normaly help, but I'm not sure that it can realy help in this case.
> Using a smaller low_error_factor than the default (0.5) may help in some case. It
only affect the
> last pretrace step.
> The artefacts are mainly related to your yellow pannels, maybe some tweaking in
there finish,
> pigment or dimentions and placing could help. Some suggestions:
> - make the pits deeper and lower the panels.
> - make the pannels a bit smaller than the pits. or enlarge them to make them larger
than the pits.
> - a little lower ambient with a higher saturation for the yellow pannels.
> - make the yellow pannels transparent and put some weak yellow area_light spot_light
under them.

Thanks for this advice! I will try out your suggestions.

Post a reply to this message

I think that looks a lot better. It makes the ship's textures pop a bit
more.

Post a reply to this message

"Ross" <rli### [at] everestkcnet> wrote:
> I think that looks a lot better. It makes the ship's textures pop a bit
> more.

Yes, agreed. In other (less uniformly lit) scenes using those ships, the
material seems to go from plastic to metal just with the increase in gamma.
I wonder if it's better matched to the textures, or something. Anyway:

To get rid of some of the artefacts, I followed part of Alain's excellent
advice. The yellow light panel buried in the floor pits has been lowered
further, and turned down a bit. I also decreased the pretrace_end value.
This did have the side effect of punishing my computer with swapping for
about a day, but the artefacting is less noticeable and I think the
radiosity solution is of higher quality. The fog has been slightly
increased compared to the last attempt. These images use antialiasing
method 2 with threshold 0.1. Attached is a smaller version of the image
with assumed_gamma 2.2. A full-size PNG image has been placed at

http://www.zubenelgenubi.34sp.com/temp/tvo_hg.png

(1280x720, 430K)

To appease Tek, here is the new radiosity solution with assumed_gamma 1.0,
and without the fog or additional sunlight. The main difference is the
colours are a little fainter, but they both look similarly "clean" or
airbrushed to me.

http://www.zubenelgenubi.34sp.com/temp/tvo_lg.png

(1280x720, 420K)

Thanks for all the advice!

Post a reply to this message

Attachments:
Download 'tvo_sml_hg.jpg' (70 KB)

Preview of image 'tvo_sml_hg.jpg'

It is absolutely perfect for a "neutral", "boring",  or "daylight" mood.  I 
mean perfect in that it looks like a photograph.  If you want it to be 
exciting, to have the lighting to show a mood, you might have some work to 
do, but it would be hard not to make it "annoyingly obvious raytracing".

Post a reply to this message