Why doesn't this work? This attachment should be named [whateveryouwant].pz3 This is not hacked into a character or anything. It's just a saved poser file with three spotlights parented to a little cube with a normal size cube underneath them. Images coming...

Here's a picture from the main camera undendered so you can see the ground shadow.

This is a rendered version of the above scene

This is the view from above

and finally the view form the top light's shadow cam, which should be proof positive that there should be a shadow cast on the ground by the large cube...

Is it documented anywher ethat as you increase the Angle End of a spotLight, the shadow cast by it decreases in sharpness? It appears that somewhere between 140 and 150, the shadows cast and the outer perimetre of the light's area of effect begin to bleed into one another to the point where you can no longer see the cast shadow, and the overall effect of the light is merely dimmed. Crap crap crap. That means it's going to take more than 4 lights to make an omnidirectional. I was hoping an equalateral pyuramid would do the trick. But there needs to be about ten degrees of overlap, and while a d4 shape will provide 120-degree ranges, for that overlap there needs to be a 140-degree angle, which means that at that point the shadows fuzz out.

Very interesting. Thanks for the info.

I'm developing my own ERC controlled OmniLight, un-rendered image above.

And I seem to be having some success with shadows (see rendered image above). I'm not sure why it is working yet, I was having the same trouble as you for a while, but note that the Angle End is set to 160.00, and the Dist End to 2.2000. Still very much in the experimental stage at the moment. I will post more info when and if I have it.

Okay, maybe I'm not the greatest in the realm of physics, but -- if I'm not mistaken, light is additive. That is, when you take two lights and shine them in the same place, it's twice as bright. When you have two lights effects crossing one another, there's a brighter part where they cross. I took and made a cube of lights and thought that I could set the outer falloff angle to 110 to get just the faintest bit of crossing, and got nothing but the main light. SO, to make a short story long, I went and set the start and finish falloff to the same value to get a crisp edge to see where the light was falling, and only got what appeared to be a single light cone. So I tilted one of the side lights up ten degrees to try and force the issue. This is the unrendered image which looks about right...

This is the rendered image. Can everyone tell what's wrong with this picture?

:) Have got tthe shadows working, will e-mail you the cr2.

This is an omni light set with 8 lights. Note the bright spots where the light cones overlap. Angle Start=100.00, Angle End=160.000, Dist Start=5.000, Dist End=10.000. Problem is we are trying to fit circles together. I think we need square lights. Perhaps masking is the answer?

This is the light set I used, it's actually two of your light sets combined.

More weirdness! this was my light set up, 18 white lights...

This is the rendered image, looks smother than the 8 light set, but still has hot spots, and i hate using that many lights Now for the weirdness...

The only change I made from the previous image was to add a cone on the floor directly below the light set. So what's that new shadow doing there??? The lights are at an altitude of y=0.550, well clear of the cone, and directly above it. I just don't get this!

I did the 8-light one right after, too. It works best if the whole thing is y-rotated 30 degrees (and, of course, x-rotated 180). Like the 8 corners of a cube turned at a 45 degre angle. Technically flipped both ways, though it only takes a 1/12 turn to get the same shape as a 1/2 turn since it's in thirds. Yours appears to be only flipped vertically, and not rotated, but that may be just how it looks in that preview. It looks like the top and bottom angles mirror one another. A render of my 8-lite is attached. Angle start 30, angle end 160. Still some hot spots but smoother. I'm thinking of ways to get rid of the hot-spots. What if -- this is just an idea -- what if the intensity were turned up to 800% so that the lights were just really, really bright, but the distance attenuation was dropped to a really small number and the start attentuation distance was dropped to almost zero, so the lights' overlap was effectively no different from the lights' non-overlap, thus making the distance attentuation the only effect... ? I'll try it.

I did rotate the second set, and pushed the Angle End to 100, that last was not the most flattering, but it demonstrated the problem better. I still don't think that 8 is enough, but I hope I'm proved wrong. 18 is a lot so perhaps there is a happy median. Any way for the moment I plan to push on with 18 light set and try to make the ERC a bit more user friendly before I put it up in the freebies.

From what I can see, the angle of a spotlight seems to be 60 degrees no matter what. I'm not entirely sure what the attenuation parametres are based on. Based on this I've built a 14-pointed arrangement of them where they are placed at the vertices of a dodecahedron (20-sided pythagorean solid). This completely minimises the overlap which, despite the numbers given for the angles, don't seem to have anything to do with the actual angle of the cast light. Why? Because at 160 degrees (the default angle limit), almost half a sphere would be lit -- but isn't. Further, it would make sense that if the angle really described the angle of the light, then upping the limit to 360 and setting it there would then create almost a real omni. But that hasn't worked at all.

Here's the 14-point effect, which seems to work...

Here's the pz3 file (just rename it to [whatever].pz3)

It's looking good dodger, thanks for sharing. You worked out the angles for positioning on a dodecahedron, wow, I cant even work out how to stand up straight (perhaps its the Vodka). Unfortunately Im at college for the next 7 hours, so I wont be able to play with it for a while, the light set that is ;).

_dodger, I don't know if you ever saw this, but I conducted a similar experiment in omnidirectional lighting earlier this year.

http://www.renderosity.com/messages.ez?Form.ShowMessage=680431#8

I needed omnis, or at least something close to omnidirectional, in order to simulate naked lightbulbs. I went a bit overboard with the spotlights (especially since I only had 128MB of memory), but basically the technique was the same.

I've since changed hosts, so here's a new link to the video:

Video clip (MPEG format, 1.37MB)

I hope that DAZ Studio and future versions of Poser will feature omnis, because interior lighting is a pain-in-the-tail without them.

Here's a scene rendered with the omni parented to a prop that's parented to the flame. I'm still getting weird results trying to parent a light or another figure by a gravity-affected part. Thus the prop in-between.

'You must specify a body for your message' -- even when attaching a file. Yup, it's in no way surprising that Renderosity and Curious Labs are in bed together. I use them both all the time, but they both piss me off equally with their bad programming and stupid bugs.

Poser spots with angles greater than 90 has strange behaviour, you can have ufo lights (light pass through a wall that cast shadows),etc, etc, etc

Stupidity also evolves!

Attached Link: OmniLight

I've posted an ERC controlled OmniLight in the free stuff. It's far from perfect, but at least it's a start, and perhaps others will get inspiration to develop the ideas further.