Outline for Today
Midterm review
Administration
Today
- Going over the midterm exam
For Next Meeting
Wiki
Link to the UR Courses wiki page for this meeting
Media
Transcript
Audio Transcript
-
So Did everyone see The
grade scope? Email? I
-
this is what it looks like. So
it'll have a link to view your
-
graded work, and it'll Be from
no reply@gradescope.com
-
so what Is sRGB?
-
Stand it up in I
-
Five in the 1990s I
-
so it was defined in the 90s
-
so that
-
we'd Have a standard
representation of colors on the
-
web and
-
this is a color space.
-
So what about a gannet? I
-
What is? What do we mean? We're
talking about a color gamut for
-
so gamut is just means, like a
range. So when we see out of
-
gamut, it just means something's
outside the range. So in our
-
case, the color space, so if
it's not represented in the
-
color space, it would be out of
gamut. Yeah, I
-
so how do we define the gamut
for an RGB color space
-
would that just be considered
the zero to one or the zero to
-
255, depending on like, which
version you're doing or and then
-
anything above those numbers
would be considered, I Guess,
-
out of gamut, or they get
truncated normally.
-
So the zero to 255
-
talk about
-
zero to full
-
color? Yeah, I
-
so typically, we're defining a
gamut,
-
so we have the tri stimulus
theory of colors, so we can
-
represent colors based on red,
green and Blue, and the gamut is
-
defined by the particular
chromaticity of those three
-
colors.
-
So we can make
-
all the point. All the colors
inside the triangle can be
-
represented by the sRGB color
space.
-
So how does the gamut for sRGB
compare to other gamuts we
-
looked at and that are available
now,
-
bigger or smaller?
-
Bigger?
-
No, I remember the diagram. I
just don't remember which was
-
labeled what? There's just one
that was bigger than all the
-
rest, though. Yeah.
-
SR should be as the smallest
gamut in That picture you
-
Does That makes sense. You.
-
No. Sorry about That. I
-
so here's the section of the
book
-
from which this question came.
-
So in pbrt, we assume that
-
the geometric, geometric optics
is adequate model for late
-
scattering. So we have linear,
linearity, okay. Linear
-
linearity, combined effect of
two inputs an optical system is
-
always equal to the sum of the
effects of each input
-
individually. Energy
Conservation, when light
-
scatters from a surface or
participating media, the
-
scattering event can never
produce more energy than it
-
started With no polarization,
-
no fluorescence or
phosphorescence and steady
-
state. So the environment is
assumed to have reached
-
equilibrium, so the distribution
of radiance is not changing over
-
time. So realistic situations,
as happens very, very quickly.
-
So we lose a geometric up with a
geometric optics model,
-
diffraction and interference
effects and
-
does that make sense?
-
And just to confirm the thing,
steady state, that's like, if
-
you were taking, let's say, a
picture of the stars at night,
-
and then all of a sudden, a
street line turned on beside
-
you. It's assuming nothing like
that street mount will appear.
-
You're capturing the light as if
it was all steady, so no sudden
-
light turns on.
-
No cars are driving by,
therefore causing new lights to
-
appear for just moments and
things like that. Consistent
-
luminance, right? I
-
I would say even when you have
new lights appearing,
-
their contribution is
-
reaches a steady state almost
instantaneously.
-
So it's not
-
that you're avoiding people.
You're disallowing people from
-
turning on lights or street
lights or
-
driving by a scene,
-
but the way we model it is that
-
we use exposure settings right
where there's like a long
-
shutter time or whatever. So in
that time you're capturing light
-
as though it was steady or
consistent. So if you've had
-
suddenly something turn on
midway through, would that would
-
interfere with an actual
photography so I assume it has a
-
thing like we're assuming that
can happen. That's one of our
-
assumptions with the steady
state.
-
I think the steady state, it
means that we're not, we're not
-
dealing with a period of time
when the light comes On,
-
and seeing the effect of the
light coming on
-
over the period of our exposure,
-
that there's A difference in
-
the inner energy from the light.
Rather,
-
if the light appears,
-
then we're just we just accept
that it's
-
In the steady state, there's a
new steady state, And
-
that might be Interesting to To
Explore I
-
so what is literate programming,
and Do you have examples aside
-
from pbrt?
-
How about Writing Code. I'm
-
I think that's a Good
explanation of literary
-
programming. I
-
anyone. What do you mean by the
narrative?
-
So describing how code is
designed or written? Do
-
you want to create an exposition
of the design of the code,
-
and so You're writing it a way
that supports the narrative So
-
does anyone have examples of
literary programming systems
-
I've done pbrt Do thanks.
-
Yeah, It's pronounced tech. I
-
What do you think about Jupiter
Notebooks? Does anyone use those
-
before?
-
So just to clarify, literate
programming is not like C Plus
-
Plus, where you have to declare
the function at the very top,
-
and then later you can do the
function, but you have to do
-
this here and stuff like that,
like a very ordered and
-
structured thing. What exactly
is literate programming, as far
-
as Like how does it differ?
-
So the documentation I
-
and Then there are Two pieces
and
-
so in Python with These Jupiter
Notebooks, you can have
-
documentation, and then you have
code that runs. So it's not
-
necessarily the same, but it's
an
-
so first of all, how many people
are familiar with tech and late
-
tech.
-
Cool, so,
-
yes, with a computer notebook
example, it's more like, it's a
-
coding environment that
-
also kind of, like allows you to
work with the building blocks.
-
It allows you to build up
various blocks until you reach
-
the very end. Because, like,
that is, I mean, that's kind of
-
nice. Mean,
-
I think with these notebooks,
you can intermingle descriptions
-
and and code, and so they're a
little bit separate. So you
-
you're including the code in the
document, but the document and
-
the code are different, so it's
about having a single source
-
that can be parsed into these
two different streams. I
-
so it's an interesting
discussion. Okay, who's who
-
here. Is familiar with Emacs?
Does anyone know what emacs is,
-
what kind of application it is?
-
So editor, I
-
so we can have a continued
discussion about this. I'll put
-
some resources online, but it
won't be on the final exam. The
-
other things will be on the
Final Exam. I
-
it rasterization versus ray
tracing. So we talked a little
-
bit about, very little bit about
Z buffer and being a
-
rasterization algorithm. We're
keeping track of the closer
-
pixel that's closer. We're
testing against a Z buffer to
-
see whether the new picture
pixel will be computed is closer
-
than the existing one. If it is,
we update the pixel. If it's not
-
We discard it. You
-
writing a little bit too small
there, I apologize. So
-
rasterization says for each
object in the scene, then for
-
each pixel i
-
that is possibly associated with
the object as it gets scanned
-
converted. Is it closer than
what's stored in the set buffer?
-
Okay, so what does ray tracing
do? I
-
to do the same loop, To organize
the same way I
-
so who says it's the same loop
as rasterization for each object
-
done for each pixel? Anyone?
-
My brain's not working today,
but that's how I interpret it.
-
Okay?
-
So actually, we're starting from
the pixel,
-
yeah, it comes, it comes towards
that Ray, right?
-
The ray comes out of the pixel?
Yeah, so the eye, which is the
-
camera, we trace a ray through
the Pixel and into the scene.
-
So for each pixel, then
-
so what's as we Go through the
scene, what's the closest and
-
then how is the value of the
pixel determined? So we're
-
tracing rates. If it's a
reflective surface that we
-
bounce the reflection rate,
retrace rate in light source,
-
see if it's in shadow and so
forth. We're
-
Does That make sense? I
-
so the depth of field is what,
what happens
-
in the space? Sorry,
-
I'm really bad headed right now
my brain can the best I can give
-
you is This pretty much a second
of
-
that. So what's in focus
compared to what's out of focus?
-
I
-
so when it's not in focus, the
lines don't converge to a point
-
on The film. They convert it
comes towards a desk. I
-
boundary of the disc.
-
So as to focus, the larger this
disc is the film, the more out
-
of focus the image will be
without the image will be or
-
that part that's out of focus,
-
so what are the different court
assistance we talked About in
-
pbrt. So
-
the world, the object, the
camera, then I also put texture
-
coordinates which I didn't know
if it was a Click of
-
one, yeah. World, object,
Camera,
-
camera, world, i
-
i also seen a normalized device
coordinates discussed in Chapter
-
Five about the projections.
-
So if we need to move between
them, we use matrices. So why,
-
when we're doing projections, do
we have four by four matrices
-
discusses
-
the classes? Because when we're
doing translation, stuff like
-
that, I so we
-
Got homogeneous coordinates and
-
so for registers allow rotation,
scaling and translation to be
-
expressed in the same matrix.
-
That's why we
-
a four by four matrix for three
dimensional objects is a useful
-
thing. Take A
-
Picture. I
-
it. Somebody asked, where this
was mentioned? Does it be the
-
four that we talked
-
about? So we have a square,
-
and then we have
-
an arc, one quarter, one
quadrant in a circle. So just
-
imagine that's that's part of a
circle. So when we're doing
-
Monte Carlo and
-
and number
-
of the it's basically trying to
find The ratio of points that
-
are inside the circle compared
to inside the Square.
-
I'll put to the link to that
administration and anything
-
clarificational to that as well.
Does that make sense? I
-
spherical geometry and
-
why is it important for
rendering? I
-
So why can we are we supposed to
specifically worried about, or
-
concerned about spherical
coordinates vertical Geometry
-
and ray tracing.
-
The so we're dealing with all in
coming and outgoing directions.
-
Look at that in terms of these
directions are expressed in
-
spheres or hemispheres,
depending on the case.
-
We got to translate all of that
to be a two dimensional screen
-
at the end, because it's got to
be viewable.
-
Yes,
-
you're taking in like every
single direction you could
-
possibly go up your day to get
all that information.
-
But also, if you did every
single direction, you would be
-
there for more than a day. I
think,
-
yeah, we do it every day.
-
It just works in nature.
-
It's a simulation that's the
hard part, and we're getting
-
more fidelity with our
simulation
-
so
-
and if we have
-
a bright light source
-
that makes it easier for having
dimly lit space, then it's hard
-
to or to Find the light in the
scene.
-
But anyway, so I
-
so what are the three
parameterizations for spherical
-
coordinates,
-
octahedral encoding, equal area,
mapping spherical coordinates, I
-
so this spherical coordinates
gives us an easy representation,
-
easy To convert between this and
Cartesian coordinates.
-
Octahedral encoding Is space
saving. I
-
so I wrote down E coli mapping
some reasonable equilibrium
-
mapping is important, because if
you're sampling
-
with spherical coordinates, you
might get, we're not going to
-
get a uniform distribution, or
we get extra stuff in the poles
-
and some distortion there as
well.
-
Okay, Does that make sense? I'm
-
so how do we use bone and
glasses to accelerate re
-
intersection tests? You
-
can use it to truncate rays that
aren't hitting your object. And
-
rather than checking like the
very fine shape of the object,
-
you just search a general area.
If you don't hit that general
-
area, well, you're not going to
hit the shape. And also, when
-
the ray leaves that general
area, if it doesn't hit the
-
shape, well, it's not it's never
going to hit
-
the shape unless it bounces. I
analysis.
-
So imagine
-
these are two axis aligned
bounding boxes.
-
What can I what other
improvement can I do in CS
-
having a collection of bounding
boxes at the same level, so very
-
shallow tree can
-
have multiple different like
levels of it. So you have one
-
boundary box that maybe contains
both of those ones. If you don't
-
hit that one, you're not gonna
hit any of the ones inside of
-
it, either, right? And you can
just keep going outward.
-
So hierarchical, I
-
realize we're in a time here. So
Ken,
-
the chapter seven, kind of was
about that when you told us
-
to read, yeah, that's making the
the data structures for
-
for the hierarchy hierarchical
bounding structures. So there's
-
a list of things that are common
to Ray tracing systems,
-
cameras and lights and
visibility and light scattering
-
and so forth.
-
I'll post A link to that. I
-
so question 11 about the
importance of choosing an
-
integrator in the sampler
-
that's from the user guide.
-
That's a guide for the answer
anyway, But I
-
tried to be open. I
-
so bi directional distribution
functions and
-
the bi directional reflectance
function, bi directional
-
transmit distribution function,
bi directional surface
-
scattering, reflectance,
distribution function. These are
-
the three we've talked about. So
why are they bi directional? You
-
equation, incoming direction and
not going direction. So we have
-
those pair.
-
Okay, so it seems like we're
just at a time since I'm
-
learned about my next meeting.
-
So that makes sense. Okay, so I
was generous with marking. So
-
this is an opportunity, so 15
marks for the midterm won't sink
-
you,
-
but if so, be clear about this
stuff for the final.
-
Okay. Thank you for today. Now
the winters back, stay warm and
-
take care, and we'll see you on
Thursday. Oh, have a look at
-
chapter 8.1, section one of
chapter eight. Okay, thanks
-
again.
-
again.
Responses
No Responses