Mtg 25/26: Tue-08-Apr-2025

Outline for Today

Lights

Administration

Today

For Next Meeting

Wiki

Link to the UR Courses wiki page for this meeting

Media

Transcript

Audio Transcript

  • Okay,
  • let's get started Here. I
  • so
  • from our quiz, what are two benefits of area lakes that I
  • wasn't really sure I put like
  • softer shadows and more wings,
  • deep lighting. Yes, that's what I was thinking of. Hello. I was
  • wondering if you're gonna join your backpack or not. I was
  • eventually,
  • the washer was closed, were closed. Had to go further. I
  • The squeaking on the whiteboard is new. I don't know if I like
  • it or not. Some some feedback that I'm doing stuff, I guess,
  • but
  • it's Not as bad as nails on
  • a chalkboard. Definitely. I
  • Is that supposed to be effects, more realistic lighting effects
  • looks
  • like an old from here
  • looks like old facts.
  • Thank you. You're welcome.
  • I'm not creating new coining new words in class here.
  • Okay? I just, I
  • so a comment he makes, or they make in chapter 12 here is that
  • the range of lights is limited by the Physically Based
  • Rendering paradigm they can't have lights that do non physical
  • things,
  • not physically accurate things. I
  • so that that could have, that has has been an effective mode
  • of operation in the past, but in the grand scheme of things, it's
  • better to use a physically based approach, because then Things
  • work together better and
  • so there are four different kinds of lights, four different
  • categories of lights,
  • point light sources,
  • directional light sources, area light sources and infinite
  • lights, so they don't have geometry associated with them,
  • But provide radiance to rays that escape the scene and
  • so it's a
  • when We have legs in The scene We Need I
  • so if we don't have lights, we're not going to be able to
  • see much of the scene.
  • If we have a case like this,
  • where there is an area late
  • so efficiencies can be gained. I
  • so if we focus samples, so
  • in that cone
  • that connects the point on the surface
  • to the sphere.
  • So those are the directions where
  • we'll find illumination.
  • Does that make sense? So if we just I,
  • yeah, if we just I to focus our samples and the directions from
  • which the light is coming, then we'll get a much quicker, more
  • Efficient rendering and less error In the image, less noise I
  • so here's a dragon with a point light source,
  • or lit by a point light source. What do you think about this one
  • hard shadowing?
  • Yeah, very hard shadows. The light
  • must be very close Because the shadows are very short. I
  • let's see what clues we have about where the light is. I
  • Okay, so this seems to be a shadow from here.
  • This is a
  • this is the shadow from here. And
  • I'd say if we were the dragon, it would be just over, like on
  • our left side, like right here. If we're the dragon,
  • let's go it's close.
  • If you're that close over here, I think this shadow would be
  • longer,
  • so maybe more to the
  • side, almost to the
  • side, yeah, maybe thinking like right here, because,
  • I guess, hard to say without duplicating it ourselves. Yeah,
  • I
  • so does it look impressive? This is a cool picture.
  • Depends on what you're trying to accomplish.
  • If I want to make a cool picture. Have I succeeded? I it.
  • Maybe this is a cool picture in the 90s. I
  • What do you think, cool or
  • not? Binous. Do you have a thought about this? No,
  • okay, I Okay. Astra full, what do you think about this image?
  • It's
  • different. I
  • so it's just
  • an expression of ambivalence. I
  • Okay, 12.4 I'm
  • so what do you notice about this? Do
  • so it seems like it's
  • it's not the same place as the first point light source is the
  • spotlight, definitely slightly different spot.
  • Yeah,
  • so it may or may not be the same position, But the difference is,
  • because It's a spotlight, so I
  • so we're specifying a cone.
  • So instead of having
  • the light emitted in all directions from that point,
  • we specify a cone, and that gives the
  • is area on the outside of the comb, like around the edges of
  • the comb are softer, where the shadow that's being reflected
  • from the dragon is still hard. Does
  • that make sense?
  • Yes, so there are two angles here.
  • So the total
  • the total width,
  • it's the whole angle. And then there's the theta i
  • Well, theta is in the i
  • Oh, so the two angles are fall off start and total width. So
  • anything within fall off, up to the angle of fall off start is
  • fully illuminated, and then
  • up to the total width, it drops off.
  • So we see the dragon in the in
  • the center here is fully illuminated. Then as the angle
  • drops, as the angle goes to the total width, it falls off to
  • zero.
  • Is this one a little more cool than the other? One
  • more interesting
  • to draw your attention, more stuff to analyze and look at on
  • it.
  • Okay, so here's steam runner with projection light using the
  • grid image.
  • So we have an image.
  • It's like a slide projector.
  • Anyone familiar with slide projectors?
  • I Okay, okay.
  • Using a projector image like that, you could put, like a
  • colored image in color
  • the light coming out.
  • Yes,
  • it's like if we had that exact same grid image. But the grid
  • image wasn't white background. It had like a orange. It would
  • project an orange hue over the Yeah,
  • so I, I wonder if there are, if
  • you just wanted one color, you could specify the spectra of the
  • light. But if you want to vary things, maybe that would be a
  • good way to do it,
  • to project the texture.
  • Go neophotometric diagram, legs and
  • do you remember
  • our Gonio photometer setup? We're measuring
  • the BRDF
  • for different materials by varying the light source and the
  • viewing position. So this is a similar idea. And
  • so the emitted intensity is defined in a fixed set of
  • directions on the unit sphere, and the intensity is found by
  • interpolating intensities of adjacent samples and
  • so the first one
  • mostly
  • eliminates in its up direction, only A small amount of
  • illumination in the down direction
  • so that's this is up and this is down.
  • Light B is a light that most illuminates in the down
  • direction. And
  • here's an example where they Have both up and down
  • contributions. And
  • I wonder if we can compare 12 point 12 with 12.3 we did You?
  • Did image diff. I
  • Let's try it out. I
  • Wow, that's a lot. I
  • No, it's not a banana.
  • So there's a lot of subtlety between those two Images. Ah,
  • more. Shadow Play around the outside.
  • Yeah, the heart shadow underneath looks like it's
  • staying in the same spot. I
  • off so
  • this shell, Shadow
  • of the dragon, it seems like light is from
  • behind the dragon, and
  • this part
  • is very subtle. It's hard to detect in the original image. I
  • I see we're sure.
  • Really look, you can see it's there, the 12, three.
  • It's hard to pick out.
  • It's there. It's actually, if you look, it's got like a little
  • bit of a red hue where that is actually
  • coming through. So it's like it's reflecting off the dragon
  • and going Down on the ground.
  • It's It's
  • a
  • bit there. I
  • So we're Interested in the emitted
  • power of distant light, and an approximation of that is to Find
  • the disk
  • that can be inscribed in the sphere that bounds the scene
  • elements and oriented towards the direction of the distant
  • light
  • that gives enough precision for what they need to do in you.
  • It's not a very it's not an accurate estimate of the emitted
  • power, but
  • I won't underestimate it.
  • So when we get
  • to area lengths, we have a geometry associated with the
  • emitter. And this is comparison of a small disk and a large
  • disk. Do
  • 12 point 15. Is this becoming more cool or not?
  • Yeah, I think that's I like Being
  • I like be overly, yeah, I
  • so in this case, we're Taking area lights and putting them at
  • infinity.
  • So environment lighting, so taking an image as
  • specifying the lighting for the scene in an
  • environment map.
  • So the first one is illuminated by a few area lights, and the
  • second one is illuminated with mid day sky
  • light from an environment map. And C is using a sunset
  • environment map. So this is area lakes. This is mid day.
  • This is sunset. And I
  • prefer infinite area lengths because I find it illuminates
  • the objects from all sides better and is more realistic.
  • So here's the mid day environment map.
  • Here's The Sunset And
  • so there's lots of
  • environment maps available On the web And
  • so this is environment
  • map illuminating the dragon Model,
  • morning skylight. Environment Map through two samples per
  • pixel. A is rendered with uniform sampling distribution. B
  • rendered with distributed samples distributed according to
  • the environment map image pixels and C, rendered using the
  • compensated distribution that skips sampling unimportant parts
  • of the image, all images took essentially the same amount of
  • time to render, though B has over 38,000 times lower MSE than
  • A and C further improves MSE by a factor of 1.52
  • what's MSE mean
  • squared error? Yeah,
  • that's quite an improvement. What
  • that's impressive.
  • Very Clean that last one I
  • so in this case, the portal is the window. So we're trying to
  • focus our samples. Our goal is to focus the samples out the
  • window and
  • see that looks a bit better.
  • Here's the idea Of the Four Pole and
  • any thoughts about that?
  • So I thought I would go through some of the questions that were
  • suggested for the final Going to
  • not show you. I
  • names of people i
  • What is the Nyquist rate? The
  • and what's the answer? I
  • two times, yes, remember it
  • was the two times something. I
  • so choices are ambidextrous, left handed, right handed. Nd,
  • indeed, there is no default coordinate system. I love
  • You. Kevin, yeah, I
  • What is that word of the minimal coordinate system is, is it
  • supposed to be used
  • in pbrt? Yes?
  • What?
  • One what needs to be accounted for to prevent textures from
  • being too blurry. You choices are camera integrators, angles,
  • bounding boxes or lights you
  • which one is not a member of the exposure triangle. Aperture ISO,
  • shutter speed
  • or angle and
  • what is the primary purpose of a bi directional scattering
  • distribution function, to define the geometry of a surface, for
  • array intersections, to model how light is transported through
  • participating media, to describe how light is reflected and
  • transmitted at a surface to simulate the spectral response
  • to the camera sensors, yeah,
  • describe how light is reflected and transmitted at A Surface i
  • What does the rendering equation primarily describe how light
  • travels through a vacuum, how surfaces reflect Light, how
  • textures are mapped, how pixels are colored.
  • So the suggested answer is B, how services reflect plane from
  • I'm wondering if that's
  • the Answer is more nuanced than that. I will
  • think about that a bit more.
  • Which of the following statements
  • about spectral rendering is true, but PBR TV four performs
  • rendering computation computations using only RGB
  • color values. Spectral rendering in PBR TV four involves
  • converting all spectral data to gray scale before processing PB.
  • PBR TV four performs rendering computation using Point Sample
  • spectra, limiting the use of RGB color to scene description
  • inputs and final image output. PBR TV four system does not
  • support spectral rendering. I
  • so use Point Sample spectra,
  • so we can specify things in RGB, but internally.
  • So that was the third answer. That was the third one, yeah. So
  • I was thinking it was, it
  • wasn't 100% Sure.
  • I What Is blends, Law I
  • as technology advances, rendering time remains constant
  • and
  • what are the three Main physical processes that infect The
  • distribution of radiance? What
  • I explain the importance of Monte Carlo integration In the
  • rendering algorithms described in the book.
  • Why is it commonly used in realistic image synthesis? I
  • A so Monte Carlo methods provide a way, a straightforward way, to
  • get a numerical solution to
  • integrals. That would be difficult otherwise I'm
  • you're still in class. We're still in class every
  • what is a texel And how is it different From a pixel? I
  • Yeah, I
  • so I will post
  • the questions after a bit of editing, I
  • any questions or concerns. So we'll do a review on Thursday,
  • and then the final will be
  • the following Thursday on my birthday, yes,
  • and which time I will be 60 years old. So
  • cupcake. I didn't think about that. You finish the
  • exam, you get a cupcake. We're supposed to bring them.
  • Actually, we're supposed to bring them. That's
  • right, yeah, cupcakes aren't we're thinking about doing
  • cupcakes for the 50th
  • celebration, but they were like, three or $4 a cupcake, so it was
  • more cost effective to get a slab cake. Anyway. Gathering
  • on the green in the summer.
  • Yes,
  • thank you for today.
  • See you on Thursday. Take care.

Responses

No Responses