Marshall: Close your eyes, we're about to
begin. I can sing while we're waiting for Stan. I can sing while we're waiting for Stan,
cause' we're waiting for Stan and it's good time to sing. I can sing while we're waiting
for, waiting for Stan. We're just gonna do, it even if it's garbage. Stan: Even if it's, oh. Marshall: We're just gonna do it and we'll
make it work. We are going to be as relaxed as humans can be. Stan: Well, things might leak out if you relax
too much. Marshall: I'm not going to sing and you won't
be a jerk. Stan: [Chuckle] What the hell man! And you
will. Marshall: As a counterpoint. Stan: Welcome to the Draftsmen Podcast - Marshall: Welcome indeed. Stan: - ladies and gentlemen. I am Stan Prokopenko,
creator of Proko. Marshall: I am Marshall Vandruff, art instructor. [Intro] Stan: Here we go. Marshall: What's the subject today? Stan: The subject today is How to Study Perspective. Marshall: Oh, okay. That's something that
I'm interested in. Stan: It's uh - Marshall: You too, huh? Stan: - Convenient that we have a professional
perspective teacher here. Marshall: Uh... Stan: You. Marshall: Yeah, you're a professional - you're
gonna be a professional perspective teacher soon. Stan: Uuh, kind of. I'm gonna have a few perspective
lessons in my basics course. Marshall: Okay. Stan: Bare minimum perspective. Marshall: Okay. Stan: We did an episode on how to study anatomy
a month or so ago. Marshall: Yeah and that was - Stan: More than that, way more than several
months ago. Marshall: Walked through the books. Stan: We walked through the books, we walked
through all the ways people should - how they should approach learning anatomy. We didn't
teach them anatomy, we taught them how to learn anatomy. This is gonna be the exact
same thing except perspective and - Oh Jesus Christ. Marshall: Whoa! And we might walk them through
fewer books. I think will be less time on books. Stan: And at the end - Marshall: Yeah. Stan: The - what do you call it? The um - Marshall: Drumroll? Stan: No, the - Marshall: The debut. Stan: The debut. Marshall: The introducing, the premiere. Stan: The premiere, of the drawing - the box
checking AI App that I've been working on. Marshall: This is the Proko box checking AI
App. Stan: Checking AI App yeah. Marshall, the
perspective expert is going to be tested with my App. He's gonna draw a bunch of boxes and
then we're gonna run it through the algorithms. Marshall: This is the test of whether - Stan: Robots. Marshall: - the guy who teaches perspective
can draw a box, I'm nervous. Stan: Oh boy Marshall. Marshall: Whoa! Stan: Yeah and people are listening, we're
gonna have a link in the description of all the results. Marshall: Yeah. Stan: So you just click on it, take a look.
We're gonna have all the boxes numbered. Marshall: You're gonna have cheers and jeers? Stan: What? Marshall: I mean, well you know if I do one
that actually passes the test of AI and say, "yeah, you got it". Stan: Yeah, we'll see, I don't know. Marshall: What do you - do you get percentages?
Like if you're 90% - Stan: No, it doesn't grade you, it literally
just overlays the correct version of the box that you drew. Marshall: Oh, okay. Yeah. Stan: That's all it does. Marshall: Okay. Stan: Yeah, we're starting with one specific
task, we're gonna expand on it later. Marshall: 'Cause this is about as simple as
you can get. Stan: Yeah, check a box. Marshall: For perspective is to - Stan: It doesn't have to be a cube, just a
box. Marshall: Just a box, right, right. Stan: Yeah. Anyway, I'll explain more of how
it works a little later. Marshall: Okay. Stan: As we get to the end of the episode.
For now, let's teach people how they can study or learn Anatomy. Marshall: Okay. Stan: No, perspective. Marshall: Perspective. Stan: Yes. Marshall: Okay. Stan: Marshall you start. Marshall: I'll start by telling everybody
something that I said many times before which is that I didn't learn perspective until I
was 28 years old and people said, "it may be that you can't learn it after 28". Stan: What? Marshall: "If you didn't learn it when younger,
it's like trying to learn a new language" and I thought "no, I'm gonna learn this".
I knew that I needed it, I've told that story before. Stan: Wow! You didn't learn perspective till
you were 28? Marshall: No and here's why, they had a perspective
class at the junior college that I did not take and then when I wanted to take it, the
teacher who taught it didn't want to teach it anymore because most teachers don't want
to teach perspective the way most professional writers don't want to teach grammar. Stan: Ooh, that makes sense. Marshall: There were all sorts of other teachers
who couldn't teach it because it's - it was sort of a lost art in the 70s. If you were
old enough to have studied earlier, I tried to study it from books and I started with
a book that my rendering teacher suggested, it was by Joseph D'Amelio and I could not
understand that book. It's a good book, but I couldn't understand it because I was one
of those students who in junior high couldn't get through the algebra course and like another
- a number of other people in our class, the boys - girls got through the algebra course
and boys couldn't for some reason but we had - in order to alleviate it - Marshall: Because you had to take algebra.
They had a pre-algebra course and so, they funneled those of us boys who couldn't get
through algebra into the pre-algebra course and I learned nothing about pre-algebra because
our teacher was a newly married woman who told us more about her and her husband's sex
life than she did teach us pre-algebra because I don't even think she knew algebra, and we
all managed to pass that course and I was so relieved I'll never have to study math
again. My brain just didn't work that way and so trying to understand it was really
difficult for me and the way I learned it, is that - there was another guy who taught
perspective and I was teaching basic drawing and I thought "if I teach it I'll learn it".
And I came to him and said, "your teaching perspective, would you be willing to swap
your -? Yes I would." Because he did not want to teach perspective anymore. So that was
how I did it, I taught five semesters. Stan: And nobody has come up to you and offered
the same thing, you're still teaching perspective? Is that the story? Marshall: Uh, actually yeah. When I stopped
teaching other people, there were some - some of my students, in fact, some of my students
had become perspective teachers. Stan: Nice. Marshall: Yeah, one of them is doing a perspective
boot camp. Stan: 'Cause you're a good teacher. Marshall: Uh, also I really got into perspective
and started to find out that it was - it was fun. Stan: Yeah. Marshall: But it's tough and I think most
students neglect it because it's tough - Stan: Because math - Marshall: Because yeah, they fear "I'm not
good at math" and perspective is quantifiable but it is also like puzzles. Stan: Yeah, but a lot of perspective I mean,
yeah, a lot of perspective is math but it gets fun when you start trying to make it
intuitive, then it becomes much more just visual and less calculation and more visually
correct, kind of like how shapes and values and all that stuff can be. Marshall: The book that changed it around
for me, this was in the 80s, Joseph D'Amelio's book I still do recommend but you - here's
what happens, you may be just - at the very beginning, it may be too tough to grasp. Ernest
Norling has two perspective books, one of them is Perspective Made Easy. That was a
good one, Joseph D'Amelio's book is good but the one that changed it around for me was
the one that I read the summer before I was going to teach the fall semester of it and
it was Ernest Watson's book, that's Watson of Watson Guptill and it was retitled "How
to Use Creative Perspective" Dover now publishes it. If you want these books, go to my website
at martialart.com under reviews and go to the section called "Draftsmanship" and I'll
give you the list of books that I recommend. Of course, Scott Robertson's is high on the
list now but Scott Robertson is a tough book, it's more advanced. Stan: Yeah. Marshall: That creative perspective by Ernest
Watson, if you are willing to give it two or three months of attention, is a great book. Stan: For beginners? Marshall: Uh, for either really smart beginners
or people who have gone through something a little more - people who've gone through
your perspective basics or you're normal - Stan: I haven't done that yet. Marshall: But when you do it. Stan: Okay. Marshall: If you do that first, I would guess
the next thing would be to go to Ernest Watson. But let me tell you what Ernest Watson got
at, the - the theory of how perspective works should be about one portion for something
like 50 portions of practical exercises and his right from the beginning explains that
masters of perspective don't think about accuracy, they get it into their subconscious where
that when they draw a line, they aim it away or know that it's coming toward it and they
feel it because - they feel it out so that it becomes intuitive which is what you're
mentioning. Stan: Yeah. Marshall: And I don't know what it was about
Ernest Watson's writing but when he explained this, I thought that makes sense, you learn
perspective to forget it and he called it "Creative Perspective" because by the second
half of the book, he's showing you, you don't follow the rules at all. You use what you
know about how to bend these rules to make effects. Stan: Yeah, Kim Jung Gi? Marshall: Yes. Stan: He doesn't plot anything out, it's completely
intuitive perspective because he knows it so well, he can visualize it. Marshall: Now, I know that since my time,
there are all sorts of good boo - that - that - books on perspective. That one on perspective,
a comic book artist, I thought was really good. I think Marcos book on - on framed perspective,
I haven't read those books from cover to cover but they seem like they're really good to
me. Stan: Yeah, I've flipped through them and
I've looked at a lot of the visuals and they all look really good. Marshall: Yeah. So there's all sorts of resources
in books. However, books are a hard way to learn it. So, video lessons can be - be better
but I'm wandering off what was it going back Stan: Well, video lessons. Marshall: I knew that I needed to learn it. Stan: You're working on a prospective course. Marshall: I am. Stan: But you already have one. Marshall: Yeah. Stan: In the meantime - Marshall: Let's - let's - should we start
with the one that I'm working on first? Stan: Uh, maybe not because you probably will
never finish it. Marshall: No, I will finish it. Stan: Are you sure? Marshall: God willing. Stan: That's what you said to me five years
ago. Marshall: I know. Okay then let me give my
excuse. Here is - is - it's six years ago. Stan: Six - wow! Marshall: We started on this six years ago.
Stan and I were talking about doing a course together and we finally decided on doing a
perspective course because I'd wanted to redo the prospective course that I'd already done
on video. Stan: Yeah. Marshall: And so, I got to work on it and
I got to work on it with no results and there were a couple times where you were getting
irritated with, "come on, is this ever gonna happen?" But I want to explain why it took
two years to produce nothing, is that I figured this is the last time I'm going to teach perspective,
it's going to be a Proko course, I want to make sure our mission statement that we wrote
together, is that the 11 yr old will take this course and understand
everything that you would learn in the highest level perspective courses and see if we can
get that for an 11 year old. And the most important thing was
the macro structure of the course; where to start, way to make sure we don't introduce
this but people aren't going to understand it and so, I worked on the outline, the macro
outline for a long time and I eventually was happy with it and then we had the big pieces
in - in - in place and I'm doing this essentially on the side, I'm not making my living with
it yet. So that's why it's taking so long. But the big pieces are in place, we're getting
little portions of it put together and - Stan: It's moving along. Marshall: I am committed that if I die right
after it comes out, that there'll be a million 11 year olds who
say "I want to master perspective" and they will know what they need to know by the time
they're 12 or 13 because I'll walk you through the process
with as little tripping as possible. Should I explain what the structure -? Stan: Yes, I mean I - I don't want this episode
to just be about like "hey go watch my perspective lecture and wait for my perspective course".
I think we should - Marshall: Yeah. Stan: We should summarize and tell them what
they need to learn. Marshall: Yes. Stan: Yeah, do it. Marshall: There are all sorts of ways to learn
perspective, all sorts of ways to go at it. The most common way is you start with - with
depth measuring tricks and the depth measuring tricks if that's all - if you had a crash
course in perspective, I've got one or two weeks to learn perspective, go for the depth
measuring tricks. There are five of them, that's one of the - the most common one is
you say "I've got picket fences that are going back in space, how do I make it accurate?",
and you establish the relationship of the first two and then you run diagonals through
a center of that back picket, not through the top, through the center of it and it'll
find the one behind it and you continue to do that and do that and do that. Every perspective
course teaches this and they usually teach it early on, and they teach you not only with
a perspective picket fence but it works with a ceiling and it works with floor tiles and
it works on any axis that you're aiming back. And to do a bunch of that sensitizes you to,
how things get smaller as they go away, which is a first principle of perspective, diminution,
but it depends on how close you are to things. So, it gets more complicated than that trick
but at least that trick will get you thinking flat surface, measure back, have control of
it. Stan: So, people shouldn't start by learning
about one-point perspective, two point and then three point and just understanding that
kind of stuff? Marshall: That is the - that is usually the
very first thing introduced. Stan: Okay. Marshall: And that's valuable but one of the
problems I've had with perspective teachers that teach one point, two point and three
point perspective is that they don't explain why. One point is specifically for when you're
looking right at the face of something and it's going back to only one point and as soon
as anything swivels, my head might be in one point perspective to this camera but as soon
as it swivels one way or another the block that it represents means things that are going
- that were just going left to right, are now going left to right and away. And so,
when a block swivels, when you step off the railroad track so you're looking at an angle,
you introduce a second point that is necessary. Stan: Yeah. Marshall: And as soon as you look down extremely
on something or up extremely at something, as long as we're still using the block as
the fundamental, you introduce a third point and third, and three-point perspective did
not get popular, I'm told, until skyscrapers came in and helicopters came in, because now
we've got more extreme views and so, they just pretty much left it out of most art history.
It was still there but the biggest thing about three-point perspective is that we see almost
everything in three-point perspective in life and I know that sounds advanced but what Kim
Jung Gi said in that AI Inland Empire lecture that he did in 2014 is that he didn't really
learn one, two and three point perspective, he just learned the cube and how to spin it
around. And I think that is the most practical way is you figure you've got three axes - we
have a 3-dimensional world, you have three dimensions. They are the things that go this
way, this way and this way. You can boil them down to a block, a block is useful to keep
track of those three dimensions and everything in the world can be found as a compromise
between those three dimensions. What this means is that the foundation of understanding
perspective is to understand the block and specifically the cube because it says - it's
such a simple one by one by one aspect ratio. So, once you've got that, then you've got
these things that will be useful for the next thing which is curves, which are complicated
and that cylinders and these cylinders each will be aiming at an axis that determines
the position, openness and, as they're called "eccentricity", an angle of those ellipses
and you cannot master, no way to master those ellipses until you've got some idea of them
on a block to keep track of. That's how your 3d program does it and it's how artists have
been doing it for hundreds of years. Stan: Yeah. Learn cylinders by learning boxes. Marshall: Yeah. So I start with right angles,
learn about what a right angle is and then a block has three sets of right angles and
then the round things that we put on that block open up a whole new world of complexity,
all of these curves that are manageable curves now because we place them on a block and then
all of the compromises in between mean that we've got what a - what a 3D program can do.
It can figure out where any plane is and that is purely line and it is with that knowledge
of line, sections, cross contours, whatever you call them, that we will then have the
ability to predict how light will fall on it by figuring if the light source was from
this angle, now I know which plane is facing away from the light. The fact is, on a flat
surface, no plane is facing away from the light. So, it is an illusion to say inside
that world of these wireframes that I've built, I can pretend like there's a light in there
and it will make the shade happen over here. Robert Beverly Hills termed it, it's illusionary
drawing. Stan: Yeah. Marshall: You're creating the illusion of
3D on a 2D surface, that's what it's all about, and so, how do we do that with the science
of perspective? Which is predictable, it's manageable, it can be learned, it can be learned
in six months to three years depending on how much energy and how much brightness you've
got in understanding this kind of stuff and - Anyway, we go from right angles to curves
and then to making those things that are machine like start to look organic, which happens
by combining them together. You know, Glenn Vilppu's teaching, the figure as spheres,
the figure as blocks, the figures as - the figures as spheres and blocks combined. Stan: Yeah. Marshall: The figure as cylinders, that is
essentially applying perspective to a human body and it's the foundation of what I think
almost all perspective teaching should be. Drew Struzan was the one who pointed out to
me when he saw my technical illustrations, he said "you know how to apply this perspective
to machines, but the same thing applies to a head, spirals of hair, hands, arms, shoulders"
and that was what opened - opened up the practicality of knowing perspective when you're doing organic
things. Stan: Yeah, is that the whole course? Marshall: No, the Proko course is divided
into the first half, is how perspective works. Stan: Okay. Marshall: It is theory. There's - there are
practical aspect to it but it's mainly if you skipped over the first half, you could
still master perspective but you might not want to skip over the first half. Stan: Okay. Marshall: Because we are going to lay a foundation
of how you look at something through an eye and if you can see it from above, see it from
the side, see it from below, see it from different angles and why it goes through a picture plane
and when the camera gets closer and the camera gets further back, it changes and there is
no perspective course that has ever given more attention to this - this point of view
of proximity, how close you are. We're making a big deal out of that because concept artists
can use that, storyboard artists can use that, anybody can use that. It's the difference
between a wide-angle view that where you're close-up and a telephoto view where everything
is compressed down. There'll be a lot of the theory and understanding how it works and
then the second half of the course is just the practical knowledge of learning how to
draw all these forms and spin them around and turn them into what you want to invent. Stan: Is that gonna be a little more intuitive? Marshall: Well let's put it this way, if you
paid attention in the first half of the course - Stan: Yeah. Marshall: - in the second half of the course,
you're still gonna work your brain real hard, but you're also going to move toward "now
I can sit down and sketch". Like when Kim Jung Gi does that cup that you're looking
down on. He tips it one way and tips it another way, he's giving the simplest scaffolding
for now where my eye is and that it becomes very practical and very useful in the second
half of the course. And I don't introduce the depth measuring systems, the depth measuring
tricks until three quarters, 5/6 through the course. And I've done this before and teaching
it live, and students, when they get those depth measuring systems later, now they stick
forever because they know why this is happening and so, "oh gosh! Now, I've got it. I'm off
and running for practical drawing". Stan: So is that the whole summary of the
course? Marshall: Yeah, that's the summary. Stan: Um, is that pretty much what your lectures
are like because the - how much is it? $11? $12? Marshall: Oh, no. My lectures - I did - the
last time I taught perspective at the junior college was in 1994 and I was teaching one
night a week and I spent 20 hours a week preparing those lectures even though I'd done it many
times and I went in there and did these lectures and we recorded them with an eight-millimeter
home video camera. And those are available to you now and say, "I want you to teach me
perspective Marshall." I did the best job I'd ever done up until that point, for 12
bucks you can get those at my website. They're me and a chalkboard and so they are not impressive
but I completely rewrote my approach to perspective for Proko. Stan: Okay, so it's not the same thing? Marshall: It is not the same train of thought,
it's - I rewrote it so that if we've got the advantage of doing these as Proko productions,
that it's a whole different way to do it and I feel like once I'm done with that, okay
that's the last word that I've got to say on it and then I'm - I don't need to ever
teach it again, all right? You pay a few hundred bucks and you're gonna get more on perspective
than you could ever get in a school that was devoted to it for semesters. Stan: Cool. Marshall: Yeah. Stan: Is there anything else that they need
to know about learning how to draw in perspective? Marshall: No, we start from the beginning
but there's - Stan: No, no no like from not counting that
course. Considering that they're not gonna take your course, how they can - how can they
learn perspective? Marshall: Nothing else they need to know but
something that students need to do. Stan: Cool. Marshall: And something that I have not done
most of my life it is to practice control of lines, longer lines. That is to be able
to put the pin down, touch the paper and control this line freehand. It is not necessary but
it's certainly valuable. You can use straight edges but I do not recommend that you learn
perspective with T squares and triangles and straight edges, at least not exclusively.
That's instrumental perspective. It's what I'll teach you in the first half of the Proko
course, it's valuable to know but the problem with it is that I've had students in my animal
drawing course and other courses, they've taken perspective but the only kind of perspective
they've taken is with a T-square, a triangle doing things through the picture plane, plotting
all those points and when it comes to trying to draw a mannequin and put a rubber band
around it, they don't have any clue how to do it. And so, freehand practical perspective
of spinning a cube around into different positions and drawing it freehand and letting it not
be accurate, just letting it be good enough. Stan: Letting it look accurate. Marshall: Look accurate enough - Stan: To help tell the story. Yeah. Marshall: Yeah. Stan: Yeah, there's a lot of drawings and
sketching styles that I can see that the perspective is not accurate intentionally. I mean, a lot
of animations have that where the you know, things are curved and converging in the wrong
way and it looks cool. Marshall: In the Warner Brothers cartoon,
still have Elmer Fudd turn to the camera and as he said something, he'll emphasize and
his head gets bigger as if he got six feet closer to the camera, it's amazing and those
- those are examples of I'm messing around with proximity effect even in a way that it
is impossible. Stan: Yeah. Marshall: That's the - that's Ernest Watson's
approach, learn it and then just mess around. Hey, there's an analogy; you learn to count
rhythm as a musician with a metronome, for example. Stan: Okay. Marshall: No professional musician needs a
metronome any more. You internalize the rhythms enough to where you will vary them; It's a
human heartbeat, it's a person getting more excited, slowing down a little bit but all
of that comes out of the musicians feelings and instincts no longer out of the need to
conform to the metronome. Stan: It's a good metaphor. It's the whole
thing about juggling too many balls. Marshall: How do you mean? Stan: You - you have way too many things to
- to juggle when you're learning a complicated thing like drawing and so, in order to do
all of them at the same time you need to practice each one enough to where it becomes intuitive. Marshall: Yeah. Stan: And then you're not thinking about it. Marshall: Yeah. Stan: It's just your subconscious that does
it for you and you're just thinking about the story you're trying to tell or the character
that you're doing. Marshall: Yeah. I'll go back to something
that R. B. Hale mentioned in "Drawing Lessons from the Great Masters" about rendering, about
light on form. That you would spin an evening with white paper and turning it into a cylinder
and looking at how the light plays around and getting some primitive forms and noticing
how light moves around it. And he said something to the effect, that if you are going to spend
your life as an artist, you should spend one evening doing this. And perspective takes
at least 20 to 30 hours of work to get your head around it. To master it, at least a hundred
hours. Maybe two or 300 hours but let's say - Stan: Ten thousand hours? Marshall: Pardon? Stan: 10,000 hours? Marshall: No, no no. It does not take 10,000
hours to learn perspective, to master perspective, but at least a hundred hours and maybe two
or 300. Now, think about this uh, two or three years from now, that time is gonna go by anyway
and are you gonna continue to avoid learning it because it's tough? I think that the immersion
in it early on from the people who are willing to say "I want to learn this and enjoy it",
that's a great secret of perspective. If you can treat these three axes that it's all based
on; height, width and depth, think of them as poles all at right angles to each other
from different positions. If you can treat learning those the way you approach a crossword
puzzle. Crossword puzzles take energy, people spend hours doing them and they improve their
vocabulary a little bit or their knowledge of cultural trivia, but this is to do it so
that you have command of something that most artists avoid and if you immerse yourself
in this enough to say, "I'm gonna get a grip on it", then by the time you're done, you
are in a category of very few artists who know this and can put it to use in their work. Stan: Yeah. I mean, when you put it that way
where 20 to 30 hours gives you a good grasp of it. Marshall: Mm-hmm. Stan: That doesn't seem so bad. Marshall: Right. And if you got a few friends
that are doing it with you, then it's - turn it into games. One of the weaknesses of the
1994 course that I did is that it doesn't have any assignments in it and the Proko course
we're gonna do, is gonna have a lot of assignments in it. There's gonna be assignment in every
every lesson. Stan: Yeah. Marshall: Yeah but you - if you're inventive
you say "I don't want to do perspective because I'm creative and perspective is so left brained".
Well, if you're creative, here's how you can prove your creativity, take those lectures
that don't have any assignments and make up your own assignments. Be creative about how
I would master this by creating some games. Stan: Wooh. I like that. Boom! Marshall: Okay, we're done. Stan: Cool. Are you ready? Marshall: I'm ready. What do I gotta do? Stan: All right. So let's start off, draw
me a bunch of boxes. They do not have to be cubes. Marshall: Uh-huh. Stan: Make it to be elongated. Marshall: Okay. Stan: They can be cubes, if you want them
to be cubes. Marshall: Okay. Stan: No construction lines. Marshall: No construction lines? Stan: I'm sorry. Are - it does - we haven't
done that yet. It'll get confused. Marshall: That makes it harder. Stan: It makes it harder. Marshall: And I can't do the lines you think
you told me. I can't do the lines that are on the other side of the cube. Stan: No transparent boxes. Marshall: Transparent boxes are the only way
to learn how to draw boxes accurately but I'm not learning here. I'm trying to say "forget
about that - Stan: You're testing your intuitive - Marshall: "I'm testing my -" Okay. Stan: Intuitive ability to sketch boxes. Marshall: Okay. Stan: If you're - if you're sketching, I mean
I think you do this - you - you do draw through the box and you draw on the other side. Marshall: Always, always. Stan: But I think most people don't, they
draw the surface of what it will end up being. If it's - if it's sketching in ink. Anyway,
the whole point is that this is an early stage, this is beta stage of our App and it does
one very very specific thing, it's got a lot of limitations. I just don't want you to break
it because I know what - I know what will break it. Marshall: I have the power to break the AI? Stan: Well not break it, it just won't work. Marshall: Okay. Stan: There's no point. It like you can draw
a circle and it won't know what you're trying to do here, it'll be like "Uh, error", so
don't do that - don't draw a circle. Marshall: Okay. But the idea is there will
never be more - there'll never be more than nine lines. Stan: Yeah, nine lines. Isn't that so? Marshall: So, that makes it sound easier.
If all we gotta place is nine lines, yeah. Stan: Those nine lines have to work together
perfectly. Marshall: All right. Stan: Um, now the App does allow you to use
either one, two or three point perspective. Marshall: Okay. Stan: Since you're a pro, let's do three. Marshall: I just tend to draw in three point.
Even when I'm thinking in one point perspective, I am aware that I'm making the line go away
and I try to make sure it doesn't diverge as it goes away. Stan: Yeah. Marshall: Which is a primary sin, is that
if the line is supposed to be going away, it shouldn't be diverging if it's supposed
to be a right angle line. Stan: Yeah. Marshall: Okay. Stan: All right, so draw me a bunch of boxes
Marshall. Marshall: All right, I will. Mmm, I can tell
I ran into trouble there. Uh, that's our front. Oh gosh, doing it in pencil means that there
can be forgiveness. Dog gonnit, I want to test these things before AI does by doing
the third dimension through there since I didn't plot the vanishing points. There is
no guarantee, they're right. The AI is gonna say "yeah you made them taper. You made them
taper according to your whim and not according to the truth that we computers know". And
after we fussed with it to the point where I should be embarrassed, maybe I'm not embarrassed
because when you're fighting a war, all you care about is winning and there's no room
for embarrassment. The embarrassment comes after when you look at the video of it and
see how badly you fought that war. "Aren't you embarrassed? Yeah, I am now. When I was
in there, you don't know how hard it was". Oh well, I can tell some of these are off
and I could - if I could do my construction lines, it would help but this is- Stan: Yeah, I know you can draw a perfect
box if you're given enough time and have construction lines. Marshall: Enough time and construction lines,
yeah. Stan: We're just testing the App here, it's
- it'd be good - be good if you make some mistake so that we can see that it's correcting
you. Marshall: Oh yeah, that's right. I meant every
mistake. Stan: Yeah, thank you for throwing in some
mistakes in there. Marshall: Yeah, well you know perfection gets
boring. Why am I nervous? Stan: Cuz you're about to be analyzed by a
robot. Oh boy. These are good. I don't think you're gonna be that far off Marshall. Marshall: But it doesn't tell, it just shows
- it doesn't give us a percentage, it just shows you your - Stan: It just overlays the correct box. Marshall: And I assume it does it by finding
three lines that converge and figuring they didn't hit that vanishing point or they did.
It figures if it's got two lines that it knows are trying to converge it's gonna find a point
and then it's gonna say that third line, let's see if it converges there. Stan: Something like that. Marshall: And it's got to look at the context
to know whether they're trying to go away, whether those lines are going away to a vanishing
point or whether they're coming towards you. Stan: Yeah. First it has to detect the lines. Marshall: It has to detect lines and make
logic out of them. Stan: I'm not gonna go into how it works. Marshall: Yeah, you might have your competitors
seeing it. Stan: No no no, that's not why. Umm, it's
a long conversation but yes, basically it looks for convergence. It - it creates vanishing
points based on your lines and then it draws a box with those vanishing points. Marshall: Can I ask a technical question? Stan: Sure. Marshall: This might be - this might be - when
it detects vanishing points in two point perspective for example, when it detects them, does it
then figure out that there is a horizon line connecting those two vanishing points and
that horizon line should be at right-angles Stan: No. Marshall: To the third axis? Stan: 'Cause it doesn't have to be. Marshall: Well, it does if it's gonna be correct? Stan: Why can't the Box be tipped over a little
bit? Marshall: It can but if it's tipped over,
it's gonna carry the horizon line with it. Stan: Oh. Right. Marshall: Remember the Y line - Stan: Not the horizon line. Is it still called
a horizon line if its not the horizon? Marshall: We call it, I call it a carried
horizon. I think I - Stan: A carried horizon. Marshall: - took reference, so a carried horizon.
In other words, if you're gonna do a tumbling box, this is one of the most common things
is that "I learned one, two and three point perspective on a horizon line, what happens
when the thing tips around?" Well, the horizon line is still important but it's just you
carry it with it. Every box that tumbles, to master it, you imagine that it's carried
its horizon with it. I think that's hard to explain. Stan: To be honest, I don't remember. Marshall: Okay. Stan: I know that we were attempting to do
that. Marshall: Mm-hmm. Stan: And I don't remember if this version
has that in it. Well, when it's not one horizon, it's three horizons. Marshall: Yes, that's right. We'd have three
horizons. Stan: It would have three horizons, yeah. Marshall: If it doesn't have it in it, it's
gonna be more forgiving for my errors. Stan: Of where the vanishing points should
go. Marshall: Yeah, because as long as you're
aiming toward a vanishing point, you're gonna miss a little bit, it's all got it pretty
close. As soon as it brings in the other criteria of those two vanishing points, they'd better
be on a line that's at right angles with your center. Stan: We can check. Marshall: Okay. Stan: When we look at the results cuz we'll
see the vanishing points, that it - that it plotted. Marshall: Yeah. Stan: And we'll be able to see if it's parallel
or perpendicular to the - the horiz - uhh, what is it called? Marshall: Perpendicular to the Y line. Stan: The Y line. Marshall: Yeah. In training students now,
that last round of students that I trained in perspective, they got so good. I have them
do 20 cubes where you actually - you spin the cube around and you find the carried horizon
and you make sure that that horizon is at right angles to your Center Y line and it
heightens your sensitivity to when you're sketching to know how wrong it is. "Oh, they
both aim at vanishing points yeah, but the vanishing points are arbitrary", and as soon
as you've - When - when I look at these ones that I did, I can tell they're off and I can
tell they're off mainly because my carried horizon is not at right angles to my Y line
but - Stan: Okay. Marshall: To get that correct, you've got
to do some fussing with it to say, "okay, that looks a little better or else you've
just got to actually take the time to draw the entire horizon line, run the vanishing
points back, run the Y line up and make sure that they're at right angles. But that's all
a lot of gobbledygook to a person who does not already know what it means. Stan: Yeah, that's true. Marshall: Yeah. Stan: Yeah. I'm - we're sorry. Marshall: But anyone - I know that people
who know perspective really well would listen to that and say "I know what exactly what's
it - what it means". If we include it in the podcast at all, It means that a bright and
curious student could say, "I'm gonna slow that down and understand everything in there
and that's gonna be my guide to understanding how - Stan: Yeah. Marshall: Why you draw all those carried horizons. Stan: Oh man, well I am excited. We're gonna
go scan these. Marshall: Okay. Stan: And run them through. Marshall: Okay. Stan: And we'll be right back. Marshall: We're going to have the day of judgment. Stan: All right. Ad: This episode's sponsored by Proko drawing
lessons. If you want to learn how to draw, if you want to improve your knowledge of figure
drawing and figure Anatomy, look no further than proko.com. There are hundreds of free
lessons on our YouTube channel but the premium courses over at proko.com/store have more
detailed videos and a lot of assignment demonstrations. Our Anatomy course also includes 3D models
of all the bones and muscles that you can rotate around and study from right in your
browser. It has PDF eBooks for each lesson so you can review the information whenever
you want to come back to it. And if you're looking to save some money we have several
discounted package deals. Head over to proko.com and start learning. Stan: Wow! Results are in Marshall. Marshall: Wow! Stan: Marshall the results are here. All right,
I'm gonna share this folder with you Marshall. You can scroll down to cube number six.
That one looks like it's just a perfect overlay, so you - you made like no mistakes. Marshall: Yeah, I got real close on that one. Stan: See that one's like - Marshall: Yeah. Stan: Almost no mistake on that one. Marshall: Yeah. OK, we're good. I'm glad for
that. Stan: Uum. Marshall: What's the one that's the worst
off is 23, this one that I tried force perspective on. Oh, I see what went wrong and the blue
lines are the correction. Stan: Yeah. Marshall: And I know exactly what happened
there, it's that line on the right. Stan: Oh that one is really far off. Marshall: Yeah, I - I made it converge with
the front corner line too soon and it said "look, if you're gonna make those converge
there, the one over on the left has to converge there also and so it forced those other lines
into conformity. If I had not gotten that line on the right so off, it might have allowed
me to put the vanishing point, the third Point, vanishing point down there way further away
which is what I was intending to do. I just made it taper too much is what it comes down
to. Stan: Yeah, that one is interesting. This
one I feel like it actually messed up because why did it also shrink the size of the box? Marshall: Because if the vanishing point is
that close according to the two vertical lines on the right, it's finding like a close vanishing
point. So it's saying, that means that vertical line on the left has got to hit at the same
point. Stan: Yeah. Marshall: And so - Stan: But why did it make it less tall? Marshall: I think that it discerned what the
proportions were that I was aiming for and it said "we can't make it a long long skinny
thing". Stan: Okay. Marshall: I think. Stan: I'm not sure it actually, but maybe. Marshall: But I do see - Stan: That's interesting. Marshall: I do see it's logic. Stan: I haven't seen this res - this kind
of result before. Marshall: Yeah. Stan: It is correct. Did the box that it drew,
is that a correct box to you? Marshall: Uh, it's hard to tell because of
the illusion that's happening. No, no, you know why it's not a correct box? Stan: No. Marshall: It doesn't have that Y line perpendicular
to the horizon line thing going on. It doesn't have that carried horizon awareness but I
might be - Stan: It does. Which one's off? Marshall: If you were to run a line through
the center of the box, down to a vanishing point. I don't think it would be at right
angles to the X and Z, the left and right and front and back lines. Uh, but that's too
- people listening to this are just going to be driven crazy. Go down to 22.7. Stan: No, you're right. Marshall: Go down to 22.7 and I think you'll
see something that is more telling. Stan: No, you're right. Marshall: Look at how the correction creates
more convergence as it down but it does not change the size. Doesn't change the height
of the block, you know why? Stan: Mm-hmm. Marshall: Because that - I got that line on
the right, the vertical line on the right more correct. So it didn't have to distort
everything as much to fix it. Yeah, this is really in the early stages, isn't it? Stan: Oh, man. So basically the point, it's
- it's just correcting your convergence. Marshall: Oh, yeah. Look at box 18 and look
at how that corrected cube - Stan: Oh that's way off. Marshall: Is skewed. Stan: No, that messed up, something happened.
That's - Marshall: That is - that is purely. Here's
why I said this obviously in the early stages, the next thing that has to be figured into
this is that your Y line, the one we call vertical has to be at right angles. Stan: Oh no. I'm looking at the steps it took. Marshall: Yeah. Stan: For some reason it detected two lines
where it should have detected one and they completely threw off the whole thing. That
one it was not the correction stage that went wrong, it was the line detection stage. Marshall: Okay. Well yeah, you can tell that? Stan: Yeah. Marshall: Wow! Some of the forced ones that
I did did good. Go to number 12 that was one that I was afraid was gonna be
really distorted. Stan: Yeah. That one where you did very well. Marshall: Because it's a wide-angle lens and
yet it seems to like it. Stan: Yeah but look at it, it corrected you
correctly I think like at the very top edge you can see that you kind of curved it and
you dipped it a little bit and it corrected that and then on the bottom right edge, it
looks like you're giving a little lip, like you're curving that plane out. Marshall: Yeah, yeah. Stan: And it's - Marshall: It straightened it. Stan: Slightened, slice that off. Marshall: Oh, number eleven came really close
too. Stan: Yeah, most of these seem to have worked. Marshall: Okay. Stan: Right? Marshall: Yes and I didn't mean to be critical
of the thing you've put so much energy into. Stan: That's what we're doing. Marshall: Yes, it's - we have to take baby
steps and the extreme version of this, you figure before there was radio and podcasts,
there were Telegraph's and they had to do just dots and dashes to get stuff out. When
this comes to its extreme version, what it means is that you would put this down and
it would know everything about it. It would tell you how far the focal length was, how
close you were to it, what angle cropped the image at that size, you would know everything
about it. We're headed to the point that you were talking about in the AI podcast. Stan: Mm-hmm. Marshall: Which is that you would do a dozen
of these and it would say, here's where you were off and here's what you need to work
on. You need to work on convergence. Let's get that word together, so we're gonna do
a whole bunch of these. Now, you've done convergence, you've got them perfect but they're converging
at vanishing points that don't make any sense because they're not on a horizon line and
so now we're gonna work on that. You got to do a whole bunch of Y perpendicular Y lines
Okay now that you've got that ,you don't seem to be in control of your focal
length and how close you are to this. So we're gonna do a whole bunch of these things where
you're gonna move the vanishing points in, move the vanishing points out. So, it's immediately
in touch with every error you make and it diagnoses exactly what kind of exercises you
need to do and it might even get so advanced that it can predict when it's worked with
you for a while, you need to do 20 of these exercises. Now, these ones you only need to
do six of these exercises. Stan: Oh, wow! Marshall: I mean it could get that smart,
right? That it can - it can estimate. Stan: It could. Stan: The AI art teacher could be - start
to create curriculums based on your previous history of learning stuff, yeah. Marshall: The way your traffic map on your
smartphone could tell you you've got the option of the shorter route the - that is - the shorter
route that is longer, the longer route that is faster and the route that doesn't make
any toll uh, tolls. And it figures it all out and gives you your - Stan: Yeah. Marshall: - Recommendations. Stan: Right now, the only part of this that
is actually AI is the line detection. I want to show you something. Marshall: Okay. Stan: What we did - so the AI part of it works
so well I'm like, I was amazed cuz I could barely see the lines that someone drew and
it detected them. What we did was we drew a box, very thin ink and then we crumpled
up the paper. Then we unfolded it and got all these shadows and like these edges all
over the place and it detected what are the lines and water just folds on the paper. Marshall: Was that luck or did you teach it
- Stan: No. Marshall: To discern between - Stan: You could do that 20 times and it'll
work. No we didn't teach it to discern folded paper. Marshall: Okay. Stan: We fed it a bunch of images with noise
and lines that shouldn't be detected, lines that should be. Marshall: Okay. Stan: Ah, and it just - we just trained it
with this data. Marshall: Okay. Stan: For example, if you draw on a grid paper
or on paper with lines, with like you know horizontal lines - Marshall: Mm-hmm. Stan: - It'll remove those lines and it'll
count only the lines that you drew. Marshall: Okay. Stan: Because that part of it, that's the
AI, it just - it's good at that. Marshall: So you discovering this some parts
of it are really smart. Stan: Yeah. Marshall: Some parts it just does great. Stan: Yeah, the parts that are not AI are
the parts that calculate the vanishing points that's just - we're just using math to figure
that stuff out but - So Marshall let me show you this works, it's able to detect. Marshall: The lines of the background versus
those. Wow! It got shadow over there and it can still figure it out. Stan: It will get rid of that. Marshall: It just looks for the highest contrast
or - Stan: Look at that image. Marshall: Wow! Stan: You could barely see the lines Marshall: That's amazing. Stan: It gets rid of all those shadows in
it. It just Marshall: And it did this without you having
to teach it? Stan: That one is - Well we trained it with
noise and all sorts of stuff, umm - Marshall: To look for the highest contrast
of something that goes consist. Stan: No high contrast anything, we just said,
"here's a - here's an image" and we told it where the lines are. That's in an - we gave
it like a few thousand of those. Marshall: Mmhh. Stan: And all it knows is this is an image,
these are the results. It decides how it gets - how it should get to the results. Marshall: Okay. Stan: We don't tell - tell it to look for
high contrast areas and look for this sort of stuff, we literally just give it data. Marshall: And then it - it - Stan: It decides - Marshall: It learns to discern. Stan: It learns - it gives itself reasons. Marshall: Okay. Stan: It learns its own reasons. We don't
even know why it does this stuff. Marshall: Wow! Stan: Comes up with its own reasons. Marshall: It appears to be multiplying. [Laughs] Stan: Yeah. It's pretty fun to see that kind
of stuff work. Marshall: Okay. Stan: Anyway, we're gonna be improving this
thing all the time umm, and I think by the time this episode comes out umm, you guys
will be able to test it out on your own. Marshall: Great and - and when they test it
on their own you are always improving it? Does it give you more data? Stan: Not all - I mean yeah, we're improving
it but not like on a - we're not uploading the improvements on like on a daily basis. Marshall: Right but the all - every - all
the tests that people do - Stan: Oh yeah. Marshall: All the time are gonna make everything
work better. Stan: Yes, absolutely. Marshall: Okay. Stan: The whole point of releasing this right
now in it's early stages is for all this beta testing so that we can see more of what real
people are actually doing and what problems we need to solve. Marshall: Mm-hmm. Stan: What is throwing it off, what's causing
it to have errors. It's gonna break a lot, like when you guys go on there and you - and
you try it out, it's gonna probably give you funny results a lot of it, a lot of the time
because it is beta but it's fun. It's fun to just like throw something in and see what
happens. box.proko.com is where they can go to test this out. Marshall: All right. Stan: And have fun. Hey guys this is future
Stan. I did a beta test round with a small group of people from my newsletter and the
results were mixed. Some people were able to upload a bunch of boxes and got great corrections
all around, some very exciting results and some people uploaded five boxes and got errors
on all five. I know you probably want to try it out for yourself but the app needs work.
We're gonna improve it and when we launch the next version, it'll be at box.proko.com.
Okay bye. Marshall: This reminds me that when automobiles
were first available, I'm told, everybody that would go for Sunday drives and they had
to hand crank these things to get them started. The tires would blow out, they had to fix
the tires themselves while they were out there. It was just a hassle but people went out for
Sunday drives even more than they do now and if it had not been through all of those iterations
of developing the automobile, we would not have cars that we have now that we hardly
even think about them because we get in and we turn on ignition and we go. Stan: Yeah. Marshall: So this is these early stages that
are going to lead to later stages where feedback of objective criteria for learning your skills
as an artist is going to be as efficient as anybody ever imagined. Stan: That's a good pitch. Marshall: It's great. I'm excited about it. Stan: Yeah me too. Marshall: It means that the next generation
of artists could be greater than anyone in the Renaissance imagined possible. Well, if
you think about what Disney animators were doing in the 1930s, from all of the research
that they were doing to figure out how to make this work. If you had shown that to Renaissance
artists, Baroque artists, I think they would have just fainted at what level drawing was
gonna come up to. Stan: Yeah. Marshall: And I think that that's - we - we
might do the same if we were to see 20 30 years from now what kind of skills young drafts
people are going to have in their training by the time they're teenagers. Stan: I think it's all about increasing that
feedback loop. Marshall: Okay. Well this was - I enjoyed
this more than I thought I was going to. Stan: Oh yeah? Marshall: Yeah. Stan: Nice. Marshall: It's sort of the celebration of
the left brain. Stan: You were afraid that it was gonna expose
your weaknesses but instead you exposed its weaknesses. Marshall: Oh well and that led to - Stan: Look at you. Marshall: Yeah. Stan: You perspective teacher. Marshall: Paul Bunyan triumphed over the machine
for one round. Stan: Yeah. Nice, well thank you for your
help. Marshall: It was enjoying. Stan: I'm going to improve it. Marshall: Okay. Stan: And thank you guys for listening. Go
learn some perspective, leave five stars on iTunes and what's the - Marshall: What's the comments? Stan: Comment. Marshall: What do you wish AI could do for
you? Stan: Ooh man! That's gonna lead to some really
weird conversations. Marshall: It could or it could be something
good, we'll see. We'll see whether our - if you had any faith in humanity. Stan: Let's try to keep it productive guys. Marshall: Yeah. If your - Stan: Not the obvious things. Marshall: If you were going to be the greatest
artist that you could be, your highest potential that you feel like you've got in you, what
do you wish AI would do? Some people might be insightful and creative on this, they may
give us things nobody else ever thought of. Stan: Yeah, I hope so. Mhmm. Marshall: Yeah. Okay. Stan: Cool. Bye Marshall, I'll see you next
week. Marshall: See you next week. Close your eyes
we're about to begin. I can sing while we're waiting for Stan. I can sing while we're waiting
for Stan, 'cause we're waiting for Stan and it's good time to sing. I can sing while we're
waiting for - waiting for waiting for Stan. He can sing while we're waiting for Stan - for
Stan. He can sing while we're waiting for waiting for Stan. He can sing while we're
waiting, sing while we're waiting, sing while we're waiting for, waiting for Stan. Yeah. Stan: Have you ever performed in front of
an audience? Marshall: No. I'm not performing now. Stan: No. Marshall: 'Cause you're not recording this. Stan: I think they are. Charlie: Yeah, we're recording. Marshall: Ahaha, is that right?
