Planetside Software Forums

General => Image Sharing => Topic started by: lonewolf on January 20, 2008, 01:20:21 AM

Title: point light
Post by: lonewolf on January 20, 2008, 01:20:21 AM
Here's a quick test render of a WIP lighthouse. I would like a point light at the top, but all TG has is a light source (sphere). Anyone know how I can change this?

Iain
Title: Re: point light
Post by: Oshyan on January 20, 2008, 01:44:32 AM
TG's light source *is* a "point light". Just leave the Source Radius at 0 and increase strength to the desired level. However what you may want is a so-called "spot light". TG2 doesn't include such light source types, but you can model something with a spherical opening in it and put a Light Source inside it and that will basically do it.

- Oshyan
Title: Re: point light
Post by: lonewolf on January 20, 2008, 04:52:03 AM
Thanks. Will give it a go.

Iain
Title: Re: point light
Post by: dhavalmistry on January 22, 2008, 12:07:05 AM
please DO share the final.....
Title: Re: point light
Post by: lonewolf on January 22, 2008, 07:19:51 AM
Haven't had much time to remodel the lighthouse, but here's a quick lightbox test I made up. The pictures are with light strength at 100 and 500. After that the box can't hold it in.

Iain[attachimg=#]
Title: Re: point light
Post by: old_blaggard on January 22, 2008, 12:52:22 PM
Your box might be able to hold in the light better if you tried using raytracing in the atmosphere.  I'm not sure, though, and it will jack up render times.
Title: Re: point light
Post by: Mr_Lamppost on January 22, 2008, 04:44:33 PM
I had a similar problem with light leaking through the top of the lantern on my lamppost  :D

http://forums.planetside.co.uk/index.php?topic=2259.0

I know I was using an extended light source so that I could cast soft shadows but don't remember any or the other settings.  I have another render on the go just now but will have a look when that has finished.
Title: Re: point light
Post by: j meyer on January 22, 2008, 05:30:56 PM
I did several tests recently and the results suggest that there
is a light through issue similar to the sun through terrain bug,
which is well known.The light source shines easily through
double hull walls or whole models aswell.
I have to admit though that i haven't tested o-b's suggestion,
so maybe there's still a chance.Otherwise we will have to wait
until planetside can fix it.
Title: Re: point light
Post by: lonewolf on January 23, 2008, 03:47:03 AM
Here is a modified lightbox with 3 wall layers and a protruding cone. light strength 500 and raytraced shadows turned on. took 5 times as long to render a quick test. so final render should take some time. fixes the problem though.
[attachimg=#]
Title: Re: point light
Post by: sonshine777 on January 23, 2008, 08:53:15 AM
It will be interesting to see if you get a volumetric beam when you put haze or low clouds in the scene. Looks good as far as the testing goes though. :)
Title: Re: point light
Post by: j meyer on January 23, 2008, 11:26:10 AM
Did a few more tests yesterday.I used 2 spheres for this
and the atmo samples were set to 32,light strength was
2500 iirc.Have a look.
Hope you don't mind me posting it here.
Title: Re: point light
Post by: rcallicotte on January 23, 2008, 12:58:48 PM
You mean enabled ray traced shadows?
Title: Re: point light
Post by: j meyer on January 23, 2008, 04:53:25 PM
Aye.
Title: Re: point light
Post by: rcallicotte on January 23, 2008, 07:56:03 PM
I'm going to want to try that for darker scenes to see what it does.  Thanks.  This is useful to understand.
Title: Re: point light
Post by: lonewolf on January 31, 2008, 02:05:52 AM
Been working on my new lighthouse. This one has a hollow lightbox in the top for my pointlight to go in.
Initial render of the object tells me I have to work on the shaders a little, but most of the detail I wanted is there.
When I get it right, I will be making a day image and a night image.

Iain
Title: Re: point light
Post by: Oshyan on January 31, 2008, 02:38:17 AM
Wow, excellent detail! Can't wait to see a finished render. :)

- Oshyan
Title: Re: point light
Post by: old_blaggard on January 31, 2008, 03:01:26 AM
Looks great!  If you're so inclined, we wouldn't mind having that file in our repository over at www.ashundar.com.
Title: Re: point light
Post by: lonewolf on January 31, 2008, 05:07:33 AM
Thanks guys. It's my first go at a proper 3d model. I made it in Google Sketchup.
Here is the picture I used as a reference.

Iain
Title: Re: point light
Post by: rcallicotte on January 31, 2008, 09:40:23 AM
Wow, great model and great potential.
Title: Re: point light
Post by: JimB on January 31, 2008, 03:13:12 PM
Quote from: lonewolf on January 31, 2008, 02:05:52 AM
Been working on my new lighthouse. This one has a hollow lightbox in the top for my pointlight to go in.
Initial render of the object tells me I have to work on the shaders a little, but most of the detail I wanted is there.
When I get it right, I will be making a day image and a night image.

Iain, that's a really nice model. Good job!

On another note, I think there's a classic CGI lighting problem here, and a soft shadow sun should possibly be used. It's when a curved surface with a specular material is strongly lit from one side, and what you see is that incredibly strong contrast between light and dark running the height of the lighthouse, instead of the division being slightly blended as the light curves around (the Sun is a big sphere of light that wraps light around objects, not a singularity). It's something Planetside should be aware of. I always use an area light as a sun when creating 3D environments for jobs. That way, the light wraps around slightly to get around that hard cut-off contrast issue. It's usually not a model resolution issue either. You can see what I'm talking about if you compare your render to the photo of the real lighthouse.

If it's any consolation, Renderman used to do it which I found out on a job involving a white space rocket a number of years ago, and I ended up coming up with the above solution in Mental Ray.  ;)

I'm also not taking anything away from your render, please understand. It's just a CGI thing. Hope I'm not hijacking the thread.  :-\
Title: Re: point light
Post by: rcallicotte on January 31, 2008, 05:44:54 PM
Jim, does this mean using something beyond the soft lighting provided in TG2?
Title: Re: point light
Post by: JimB on January 31, 2008, 05:52:05 PM
Quote from: calico on January 31, 2008, 05:44:54 PM
Jim, does this mean using something beyond the soft lighting provided in TG2?

I've no idea, but I'd recommend playing with specularity to try and keep the spread of the highlight down.

Just to add, I think the best way to illustrate what I'm talking about is by looking at photos of Earth from space, noting how the division between day and night is soft, not very hard:
http://www.harmsy.freeuk.com/images/earthmoon.jpeg
http://hyperphysics.phy-astr.gsu.edu/hbase/solar/picsol/earth5.jpg
http://chamorrobible.org/images/photos/gpw-200702-75-NASA-ISS002-E-7377-space-ocean-terminator-night-and-day-20010617-Earth-medium.jpg (http://chamorrobible.org/images/photos/gpw-200702-75-NASA-ISS002-E-7377-space-ocean-terminator-night-and-day-20010617-Earth-medium.jpg)
http://www-misr.jpl.nasa.gov/education/images/Apollo8_Earth2.jpg
http://www.dkimages.com/discover/previews/794/976497.JPG
Title: Re: point light
Post by: rcallicotte on January 31, 2008, 07:18:29 PM
Thanks, Jim.  I'm learning lighting and this helps.
Title: Re: point light
Post by: Will on January 31, 2008, 07:23:48 PM
Nice, I can't wait to see how this comes out, a lot of interesting lighting going on here.
Title: Re: point light
Post by: Matt on February 01, 2008, 12:53:29 PM
Hi Jim,

I agree with your observations about the hard edge on the terminator of hard surfaces and how they can be improved by using an area light, but I think the Earth images show something different. In those Earth terminator images most of that softness comes from the scattering effects of the atmosphere. The angular diameter of the sun as viewed from the Earth is only about 0.5 degrees, so on a hard surface there is only about 0.5 degrees of additional softness that wouldn't be there with a point light source.

Enabling soft shadows in TG2 may help to slightly soften the terminator on hard surfaces, but it doesn't simulate all the softening effects you'd get from a true area light source. Not yet anyway.

Matt
Title: Re: point light
Post by: rcallicotte on February 01, 2008, 05:58:09 PM
Hmmm.  Mmmm, hmmm.  Interesting. 


Quote from: Matt on February 01, 2008, 12:53:29 PM
Not yet anyway.

Matt

Title: Re: point light
Post by: Mr_Lamppost on February 01, 2008, 06:39:28 PM
Nice model  :)

I know you are duplicating it from the photo but what is all that shoreing up around the base?

Quote from: old_blaggard on January 31, 2008, 03:01:26 AM
Looks great!  If you're so inclined, we wouldn't mind having that file in our repository over at www.ashundar.com.

Didn't expect that  ;D then "Nobody expects the Spanish Inquisition"  ;)

Quote from: Matt on February 01, 2008, 12:53:29 PM
I agree with your observations about the hard edge on the terminator of hard surfaces and how they can be improved by using an area light, but I think the Earth images show something different. In those Earth terminator images most of that softness comes from the scattering effects of the atmosphere. The angular diameter of the sun as viewed from the Earth is only about 0.5 degrees, so on a hard surface there is only about 0.5 degrees of additional softness that wouldn't be there with a point light source.

Enabling soft shadows in TG2 may help to slightly soften the terminator on hard surfaces, but it doesn't simulate all the softening effects you'd get from a true area light source. Not yet anyway.

Matt

I am sure you mentioned elsewhere that the GI does not take account of secondary forward scattering of light in the atmosphere, which I guess is at the root of this problem, not confined to TG. 

Title: Re: point light
Post by: Will on February 01, 2008, 06:45:31 PM
heh I just realized your modeling a model :)
Title: Re: point light
Post by: JimB on February 02, 2008, 07:45:03 AM
Quote from: Matt on February 01, 2008, 12:53:29 PM
Hi Jim,

I agree with your observations about the hard edge on the terminator of hard surfaces and how they can be improved by using an area light, but I think the Earth images show something different. In those Earth terminator images most of that softness comes from the scattering effects of the atmosphere. The angular diameter of the sun as viewed from the Earth is only about 0.5 degrees, so on a hard surface there is only about 0.5 degrees of additional softness that wouldn't be there with a point light source.

Enabling soft shadows in TG2 may help to slightly soften the terminator on hard surfaces, but it doesn't simulate all the softening effects you'd get from a true area light source. Not yet anyway.

Matt


Good point Matt. However, we don't see a harsh terminator on planets without atmosphere. I personally put it down to the Sun being much larger than any of the planets, which means light is hitting the 'spheres' from a wider source and wrapping around slightly. I've used the technique before and you get two desirable things; a soft division between the day and night side, and a more even spread of light on the day side which smooths out the distribution of light. The latter means you don't get a very harsh polarisation of light where the planet's surface is pointing more directly towards the Sun, which we see in actual planetary photos (it was one of those things that always bugged me about singularity light sources).

That 0.5 degrees makes a big difference because the surface area being illuminated is actually quite large.

I did a very quick test in TG2 last night and you are definitely correct that the Atmosphere is diffusing the light and softening the terminator. However, if you have clouds that are very (and unrealistically) high they show a harsh terminator, possibly because of the scattering which might be similar to specular shading?

I think my point is, more importantly, that when applying a specular surface, be wary of making the specularity too rough and tweak it. Far better to use proper soft reflections with just a little bit of specularity thrown in. There are CGI lighting techniques that can throw out ambient occlusion and final gathering completely by using soft reflections only.
Title: Re: point light
Post by: nikita on February 03, 2008, 08:10:43 PM
Quote from: JimB on February 02, 2008, 07:45:03 AMGood point Matt. However, we don't see a harsh terminator on planets without atmosphere.
But that's not due to the sun being an area light. It's more or less the small structures that throw longer shadows when they are nearer to the dark side. So, shadows get more frequent the closer you get to the day/night-border. When watched from the distance, you don't see the single shadows, but an overall darkening effect that results in a seemingly smooth transition from night to day side.
The effect can be seen on this photo: http://www.lighthouse.net.au/Lights/TAS/Cape%20Wickham/Cape%20Wickham%20Tower%20rc%203.jpg

That being said - if the lighthouse had a rough but subtle surface, that might do the trick.
(IF its walls are made of stone)
Title: Re: point light
Post by: lonewolf on February 04, 2008, 01:24:40 AM
Quote from: nikita on February 03, 2008, 08:10:43 PM
Quote from: JimB on February 02, 2008, 07:45:03 AM
That being said - if the lighthouse had a rough but subtle surface, that might do the trick.
(IF its walls are made of stone)

That might be one of the steps I will take on my daytime picture. By using a small amount of displacement on the main surface and perhaps a soft fill light, it might look real enough.

Iain
Title: Re: point light
Post by: JimB on February 04, 2008, 09:03:11 AM
Quote from: nikita on February 03, 2008, 08:10:43 PM
Quote from: JimB on February 02, 2008, 07:45:03 AMGood point Matt. However, we don't see a harsh terminator on planets without atmosphere.
But that's not due to the sun being an area light. It's more or less the small structures that throw longer shadows....

Sorry, but I have to thoroughly disagree with that. If it were the case then white reflective spacecraft would show a completely sharp contrast between their light and dark sides, which they don't. There's just no getting around the Sun not being a singularity point light, and that it has a surface that emanates light from its surface (edge to edge, so to speak), in all directions. Even seemingly hard shadows from an object onto another object scatter at the edges the further they get from the source shadow casting object. This happens in space as well. An eclipse seen from space illustrates this nicely, where there is no atmosphere on the Moon to diffract the rays:

(http://www.mreclipse.com/SEphoto/TSE1991/image/TSE91-GOESw.JPG)
http://www.mreclipse.com/SEphoto/TSE1991/TSE1991galleryB.html (http://www.mreclipse.com/SEphoto/TSE1991/TSE1991galleryB.html)

(http://ali.apple.com/space/space_images/EuropikeEklipsis.jpg)
http://ali.apple.com/space/images_earth.shtml (http://ali.apple.com/space/images_earth.shtml)
Title: Re: point light
Post by: nikita on February 04, 2008, 09:28:20 AM
I don't disagree with that, I'm just saying that there are several effects, that cause soft shadows. The area light effect is just comparatively small.
(I'll do some calculations on that. If I'm wrong, I'll find out. :) )
Title: Re: point light
Post by: rcallicotte on February 04, 2008, 09:58:21 AM
This is interesting, Jim.  Thanks for the explanation and I especially appreciated the examples.  I hadn't seen these.
Title: Re: point light
Post by: nikita on February 04, 2008, 10:14:31 AM
Soooo...
The transition between light and dark side is about 0.465% of the diameter of the object in question. Here are some examples:
Moon: 16km (10miles). So, if you take this image http://upload.wikimedia.org/wikipedia/commons/1/18/HalfMoon.jpg that zone is about 3.5px wide.
Earth: 60km (38miles)
Space Shuttle: 8.1cm (3.2inch)

Now it's up to you to the artist to decide if that's harsh or not.

I admit, that effect is stronger than I thought.  :)
Title: Re: point light
Post by: Matt on February 04, 2008, 12:17:13 PM
Quote from: JimB on February 02, 2008, 07:45:03 AM
Good point Matt. However, we don't see a harsh terminator on planets without atmosphere. I personally put it down to the Sun being much larger than any of the planets, which means light is hitting the 'spheres' from a wider source and wrapping around slightly.... That 0.5 degrees makes a big difference because the surface area being illuminated is actually quite large.

The absolute size of the sun isn't important, only the angle subtended by its visible disc at the distance from the sun to the planet. As Nikita says that wraps around about 0.4% to 0.5% of the diameter of the planet (or cylinder, whatever). In some scenes you might notice the difference. On the moon, however, near the terminator the roughness of the terrain casts shadows over much larger distances than that, so you would mostly notice the difference on smooth plains or if you look at the edges of individual shadows.

QuoteI've used the technique before and you get two desirable things; a soft division between the day and night side, and a more even spread of light on the day side which smooths out the distribution of light. The latter means you don't get a very harsh polarisation of light where the planet's surface is pointing more directly towards the Sun, which we see in actual planetary photos (it was one of those things that always bugged me about singularity light sources).

Yes, that wrap around is an important effect that should be simulated. TG2 doesn't do this yet. You will get better results with a proper area light.

The second benefit you mention above about reducing the extreme variation in brightness is something you should only see if you use a much larger angle. The real reason the moon, for example, doesn't behave like a CG sphere (has a much flatter appearance at full moon etc.), is mostly down to its reflective properties, and 0.5 degrees of area light can't correct for that (or if it does then there is a bug in the area light model). If a large area light solves that problem then I wouldn't argue that you shouldn't use it :) but I'd worry about it then creating too much softness elsewhere.

Quote
I did a very quick test in TG2 last night and you are definitely correct that the Atmosphere is diffusing the light and softening the terminator. However, if you have clouds that are very (and unrealistically) high they show a harsh terminator, possibly because of the scattering which might be similar to specular shading?

I wouldn't rely on TG2 to test these theories ;) You are right that high clouds in TG2 show a hard terminator, and that's a combination of various limitations in TG2's models, including the lack of decent area light support that you're looking for.

Quote
I think my point is, more importantly, that when applying a specular surface, be wary of making the specularity too rough and tweak it. Far better to use proper soft reflections with just a little bit of specularity thrown in. There are CGI lighting techniques that can throw out ambient occlusion and final gathering completely by using soft reflections only.

Definitely.

Matt
Title: Re: point light
Post by: Matt on February 04, 2008, 12:32:37 PM
Another thing to think about... even if the sun were a singularity and it were illuminating a rough surface there would be a wider transition zone (just as the craters on the moon can sometimes catch light beyond the theoretical terminator line because of chances of line-of-sight). With displacement, TG2 can simulate that to some extent, but its accuracy depends on the detail levels. So if the surface shaders are not designed to simulate these statistical variations (and in TG they're not), then an area light source with a larger diameter could help with that. As long as you don't mind it affecting the softness of shadows elsewhere.

Matt
Title: Re: point light
Post by: JimB on February 04, 2008, 01:21:14 PM
Righto  ;) I'm gonna check all this out once I've finished my current job, and set up a mini Solar System. But one thing I haven't figured out is if the Sun's corona adds illumination, extending the illumination further than the surface  ::) ???
Title: Re: point light
Post by: Matt on February 04, 2008, 01:56:45 PM
In the real world? Theoretically yes, but it's got to be some rediculously miniscule percentage of the main illumination from the photosphere... You'd have to have the camera's exposure cranked up so high that any illumination from the photosphere would blow everything out. But I think you know that already :)
Title: Re: point light
Post by: Cyber-Angel on February 04, 2008, 07:06:41 PM
Quote from: JimB on February 04, 2008, 01:21:14 PM
Righto  ;) I'm gonna check all this out once I've finished my current job, and set up a mini Solar System. But one thing I haven't figured out is if the Sun's corona adds illumination, extending the illumination further than the surface  ::) ???

Jim,

What kind of information would you need I might be able to do some research and find some for you, if I knew the kind of data or information that would be the most helpful in this matter?

Regards to you.

Cyber-Angel   
Title: Re: point light
Post by: JimB on February 05, 2008, 02:14:23 AM
Quote from: Cyber-Angel on February 04, 2008, 07:06:41 PMJim,

What kind of information would you need I might be able to do some research and find some for you, if I knew the kind of data or information that would be the most helpful in this matter?

Thanks CA. All I really need to know is if there is any perceivable added illumination from the corona, and what that might be compared to Sun's photosphere? As Matt says, it's likely negligible, but it would be good to know for sure. Would the corona be bright enough to increase the softening of the Moon's shadow on the Earth?

I did a quick setup last night, and I was more than surprised that the shadow cast from the Moon onto the Earth is far harder (although still soft edged) than the photos above. However, in a turn of the expected tables, it was similar to the time lapse frames of a lunar eclipse in this link (green text): http://www.geocities.com/the_150mm_reflector/moon.htm and like this photo:

(http://msnbcmedia4.msn.com/j/msnbc/Components/Photos/041027/041027_space_eclipse_hlg10a.hlarge.jpg)

Does anyone know if the Moon interferes with the light at its edges during a solar eclipse?
Title: Re: point light
Post by: Cyber-Angel on February 05, 2008, 06:27:58 AM
It is highly likely that any perceptible additional illumination from the corona is more then likely to be swamped by the intense glare emitted by the photosphere, and in all likely hood would probably contribute (And this is speculation at this point) about 0.3-0.5 of a percentile of relative over all Lux value to the apparent illumination of the sun at lest as far as the visible spectrum is concerned: you could look at the sun with other spectra such as Gamma and X-rays but they don't apply in terms of rendering applications.

I have done some preliminary searching but cannot find any hard data at this time: this is due to me probably using incorrect search criteria; plus me been out of my field (Deep Space Exploration) with this.

I am wondering if the lunar regolith is responsible for defuse scattering of the light across the surface relative to the angle of incidence with some kind of subsurface scattering taking place on the lunar surface (Speculation on my part) and that this is causing the softness your seeing in the photographs, since TG2 dose not account for the aforementioned then it produces a hard shadow edge instead of a defuse softened one.

Regards to you.

Cyber-Angel                 
Title: Re: point light
Post by: nikita on February 05, 2008, 07:52:47 AM
I think, the soft shadow of the moon on the earth results from clouds scattering light into the shadow. You can see that the shadow on the photo you posted (http://ali.apple.com/space/space_images/EuropikeEklipsis.jpg) is yellowish or reddish because red light is more likely to be scattered than blue light.
I guess TG2 just doesn't respect that effect to the extent it would be necessary for such special applications?

Maybe there's also a small amount of light diffracted by the moon but that's probably not that much?

As for the coronas brightness. The german wikipedia says it's about as bright as a full moon.
Title: Re: point light
Post by: JimB on February 08, 2008, 03:50:11 AM
Thanks guys. Will play with all that soon.
Title: Re: point light
Post by: lonewolf on February 18, 2008, 04:41:51 AM
Ok guys, here is my daytime lighthouse image. Not sure what happened to the water on the right hand side, and the piece of land directly behind to the left of the lighthouse looks a little unnatural.
Overall though, I'm quite happy. Comments, critique and suggestions welcome.

Iain
[attach=#]
Title: Re: point light
Post by: rcallicotte on February 18, 2008, 09:37:50 AM
Love the foreground rocks.  Love the modeling and the lighting.  Great clouds, too.
Title: Re: point light
Post by: nikita on February 18, 2008, 09:56:09 AM
Those metallic surfaces look a bit like plastic. But that might also be a matter of taste.
The rock the lighthouse stands on looks really cool.  :)
Title: Re: point light
Post by: Mr_Lamppost on February 18, 2008, 05:59:22 PM
While all the elements look fine, nice texturing on the rocks by the way; the scale looks a little off.  My guess is that the lighthouse model is very big.
Title: Re: point light
Post by: nikita on February 18, 2008, 06:19:20 PM
btw Are you going to share your lighthouse with us?
I got a nice idea for an image with that object.
Title: Re: point light
Post by: lonewolf on February 19, 2008, 01:13:01 AM
Thanks all for commenting.
Mr_L: The scale does seem a little off. I had to scale up the lighthouse quite a bit to fit the landscape, but I may have overdone it a little.

Nikita: I am going to share the model, I was a little preoccupied making this picture to zip it and ftp it.
Here is the link:
Lighthouse.ziphttp://home.exetel.com.au/iaintarrant/visual_experiences/index.html (http://home.exetel.com.au/iaintarrant/visual_experiences/index.html)

Iain
Title: Re: point light
Post by: lonewolf on February 22, 2008, 07:29:45 AM
Ok, so here is the dark version of my image. It was going to be a night time image with a beam of light coming from the top of the lighthouse, but I am having difficulty with the beam of light and light coming out from the light box (except when I use ray traced shadows, which multiplies the render time by 5, so my 58 hrs would have been nearly 300 hrs).
So instead I decided to make it a daybreak image, with just enough light in the atmosphere to see the lighthouse and it's surroundings, but also enough to make sure the light could not be seen. I also placed a small light in the lower story (those of you who downloaded the model may notice the lack of window panes) so you can imagine the lighthouse keeper having a morning cuppa before going up to switch of the light.
I also tried saving the image in the openEXR format this time as well as TIFF. With a little manipulation in FDRTools and then a little more in ACDSee (still learning the OpenEXR stuff), I came up with this. I hope you like it.

Iain
[attach=#1]
Original TG2TP output
[attach=#2]
Modified Image
Title: Re: point light
Post by: sonshine777 on February 22, 2008, 08:35:58 AM
Looks good Iain, I posted over at the file sharing thread, and found that the only way to get beams was with ray-traced shadows enabled also. And yes it does increase render time (lets hope the new render engine speeds thing up) :).
Title: Re: point light
Post by: rcallicotte on February 22, 2008, 09:47:03 AM
Another good way to get beams is to create three thick cloud layers and then disable the Enable Primary on all of them.  Sink the sun, crank it's strength and find the beams.