Lighting Advice on a background

Matt Frith · Wed 03/02/2010 23:30:30

Hi there, I have been working on this background for a few days now and have come to the stage where I need to get the lighting correct.

I think the building shadows are okay but I just cant seem to get the oil lamp to work. Does anyone have any advice they wish to offer. In exchange, I can offer cookies.

Here is my attempt at the streetlight

Oh and if anyone has any general critique, it is more than welcome. I am aware that some details are not done yet though.

Thanks in advance.

auriond · Thu 04/02/2010 00:15:33

May I compliment you on that lovely sky

One thing though: the patch of sky to the right seems like it should be much darker. Otherwise it looks like there's a second light source in the sky behind the buildings.

As for the lamp, I would make the glow more rounded, rather than beamed down as if from a modern electric lamp. There would be a pool of light just below the lamp as well. Here are some examples of gas lamps from a quick google:

Ryan Timothy B · Thu 04/02/2010 00:36:00

I played around with your image and tweaked a few things.

(I may have made the moon a little too dark)

I didn't like how the main building was straight and parallel to the side walk, so I brought the front out, and the rear in. Darkened the rear half as well.

I felt it was a little too bright on the right side of the image, so I darkened that a little too (forgot to adjust the fence though).

I wasn't really liking that sky. Felt it was too distracting, so I tossed in a gradient to quickly show what it should possibly look like. Also added a few gradients here and there as well (I know it's a pixel art type background, but it's only an edit

).

Added a shadow to that sign, and corrected the shadow on the sidewalk.

Anyway, gave a quick stab at the light post too. Not sure if it's as fitting though.

Anyway, I like your style, it's interesting.

LRH · Thu 04/02/2010 01:03:50

Just toying with it. I don't know about this, but feel free to take a look.

If you need to know how it was done I can tell you all about it.

Jim Reed · Thu 04/02/2010 01:16:38

Matt, shame on you, weren't you lisening the other day?

The mast shadow is perspectively incorrect. The upper grill on the window also.

As for the lamp...well, Ryan drew a nice halo, and it should (the lamp) make an oval lighted area just beneath with correction to the curb. Also don't forget to light the building a bit, 'cause the lamp is there.

Shame on you, really =P

Ryan Timothy B · Thu 04/02/2010 01:20:52

QuoteThe mast shadow is perspectively incorrect. The upper grill on the window also.

Ahh yes, forgot to mention about the shadow of the mast. I had moved the moon over in my edit to balance it out a bit.

Jim Reed · Thu 04/02/2010 01:26:34

On second thought:

The light from the lamp should actually look something like this.

Also, there was an perspective issue (in green) from the building shadow falling on the curb.

Now gimme my cookie =)

Andail · Thu 04/02/2010 08:12:48

Funny, I don't perceive the cast shadow of the mast to be incorrect. I also fail to see how your red line is actually helping.
Either way, the moon interferes with the mast a bit awkwardly - the mast should either cover it partly, or there should be some space in between. However, I don't think Ryan's darker moon serves any purpose. A full moon can be very bright, and it just adds a nice dramatic effect.

When drawing the lamp light, don't overdo the light cone, or the spot where it hits the ground; street lamps are not spotlights, they're not used to illuminate one perfect circular area of the street. Give the ground some diffuse highlights, and skip the beam itself.

Ryan Timothy B · Thu 04/02/2010 12:03:46

QuoteI don't think Ryan's darker moon serves any purpose.

Ya, I agree that it's a wee dark (mentioned it in my post too

). The only reason I darkened it like that, is for the heavily cloudy sky he has. Since his moon is drawn over top the clouds, to me, it looks like it's pasted on there.

Meh.

Jim Reed · Thu 04/02/2010 12:18:34

It's clearly visible from the shadow that the moon isn't in the right place to cast it. And, yes, the lamp should eminate a soft yelowish glow on the pavement, certainly not a spotlight effect there.

Jim Reed · Thu 04/02/2010 12:34:45

On second thought:

A is not correct
B is how should the shadows fall
C is how long the shadows are supposed to be

I was wrong.

Ryan Timothy B · Thu 04/02/2010 12:41:17

I was actually just about to post to correct ya. But you've done a better job explaining it.

If you were to use this method with other light sources (closer ones, not the sun or moon), you wouldn't use the horizon as the guide, you'd use the point of the ground the object is over. So if the light source was a lamp, you'd use the base of the lamp as the shadow guide (instead of the horizon).

Edit: Or rather I should say that you use the vertical intersection point of the light source to the ground or ceiling, if it's a close light source. Shadows aren't always on the ground.

Babar · Thu 04/02/2010 12:45:12

For my own knowledge, what about the shadows cast by objects on the horizon (or above it?)

GarageGothic · Thu 04/02/2010 12:57:04

Aren't shadows cast by objects illuminated by sun/moonlight supposed to be parallel (since light rays from heavenly bodies are essentially parallel by the time they reach earth)? If the object in Jim Reed's example C moved around, the shadow would change direction. I used a similar technique for calculating shadow length/angle for my dynamic shadow module, but as far as I could tell it only works that way with point lights.

Edit: Actually Babar's question illustrates quite well why this technique doesn't work for the sun. If an object was at the horizon, the shadow would be cast horizontally (along the horizon), thus doing it this way mimics a point light source located directly above the horizon rather than one far away in space.

Jim Reed · Thu 04/02/2010 12:59:06

I think it's something like this, with this here UFO, but I'm not absolutely certain.

Calin Leafshade · Thu 04/02/2010 13:00:42

I agree with GG, Jims model assumes the moon IS at the horizon which of course it isnt.

the light source should be a huge plane perpendicular to a line between the moon and earth... the moon is not a point light at those distances.

EDIT: Or rather you COULD assume the moon is at the horizon but it would have to be ACTUAL size.

Ryan Timothy B · Thu 04/02/2010 13:06:04

QuoteFor my own knowledge, what about the shadows cast by objects on the horizon (or above it?)

*cough* you fake it.

lol That's the drawback for using the horizon as the intersection point for the sun, since the sun is indeed further than that (the REAL method, I do not know, but in most cases the horizon works for the sun/moon). It looks worse here on Jim's demonstration because he uses a black line for the horizon making it look like it's really close.

The shadow is also supposed to widen the further from the light source. You're supposed to draw two lines from the sides of the light source, and cross them when connecting it to the obstruction to determine the widths. So left side of the sun, to the right side of the obstruction, and right side of the sun to the left side of the obstruction. (which Jim didn't do on his UFO, or pole shadow)

Jim Reed · Thu 04/02/2010 13:15:16

Excuse my crude attempt =) I stand corrected.

It's essentially all about raytracing. I do my shadows by heart, but I see I can improve on that particular area quite a bit now.

Thank you for the critique...errr, wasn't this Tiers thread...? =)

Matt Frith · Thu 04/02/2010 13:36:13

Hello again, and thank you for all your advice! I think I have taken onboard the advice and have come up with this new edit. Hopefully the light cast by the street lamp makes more sense.

As you can see, I still havent resolved the mast shadow, but it seems to have made for an interesting debate so we will see where that goes.

Edit: I realised I hadnt added any light on the cobble below the street light.

Snarky · Thu 04/02/2010 14:17:30

For purposes of drawing, it's perfectly fine to assume that the sun and the moon are point lightsources at the horizon.

Quote from: GarageGothic on Thu 04/02/2010 12:57:04
Aren't shadows cast by objects illuminated by sun/moonlight supposed to be parallel (since light rays from heavenly bodies are essentially parallel by the time they reach earth)? If the object in Jim Reed's example C moved around, the shadow would change direction. I used a similar technique for calculating shadow length/angle for my dynamic shadow module, but as far as I could tell it only works that way with point lights.

Edit: Actually Babar's question illustrates quite well why this technique doesn't work for the sun. If an object was at the horizon, the shadow would be cast horizontally (along the horizon), thus doing it this way mimics a point light source located directly above the horizon rather than one far away in space.

Yes, the shadows are parallel, but because of perspective effects they all seem to point towards a point on the horizon directly under the sun. In effect, shadows from celestial objects are always drawn in one-point perspective. Think about how you would draw the shadow of the building on the left edge of the screen (if you extended the canvas to the left), and you'll see that it works out.

As for objects on the horizon, there aren't any objects on the horizon. The horizon is infinitely distant (your perspective horizon, that is, not the geographical horizon), and objects at the horizon are scaled to 0. Only the sun and moon are so far away that you can consider them to be on the horizon for the purposes of perspective. For objects that are so far away that they appear close to the horizon, just don't have them cast any shadow on the ground, since you wouldn't really see it from your POV.