00:00:00
Hey everybody, Adam Savage here. I'm at
00:00:01
Spectral Motion in LA with Dave Antelg
00:00:04
and we are going to talk
00:00:05
photoggramometry today. Absolutely.
00:00:07
Okay. So tell me how photoggramometry
00:00:08
works into spectral motions process.
00:00:12
Absolutely. So we were doing a a laser
00:00:15
light um scanning first. Um it's where
00:00:18
you use uh a device where it flashes you
00:00:23
with light and it's almost like a crappy
00:00:26
camera. So it's doing about 30 frames
00:00:29
per second and it's scanning you and
00:00:32
getting depth information from its
00:00:34
imagery. And so through that it creates
00:00:37
a point cloud and through that then we
00:00:40
can put flesh on it um and mesh it and
00:00:44
this is the way it currently works. Yes.
00:00:46
Okay. So I didn't I thought
00:00:47
photoggramometry was sort of assembling
00:00:50
a knowledge of depth data based on the
00:00:53
accretion of photos but you're
00:00:55
describing something slightly different
00:00:56
a little different and I believe you
00:00:58
were once scanned at a place with your
00:01:01
astronauts yes and in that it has 87
00:01:05
cameras about 100 and so my technique is
00:01:09
uh a little a little more modern because
00:01:12
technology has grown and with this
00:01:15
camera we can do burst shot and it will
00:01:17
track you as it's shooting. And so what
00:01:20
that means is this can shoot 30 frames
00:01:22
per second and it's saving the spatial
00:01:25
data. Exactly. Yes. And so what we do,
00:01:28
we have this camera right here. It's
00:01:30
called the anchor camera. And so this
00:01:31
stays as it is. And as you sit here, it
00:01:34
will record a perfect 360 of you steady.
00:01:39
As I do my wave pattern and collect data
00:01:43
from your top to your middle to your
00:01:45
bottom, we compile that up. So these two
00:01:48
cameras when you're doing that are clear
00:01:49
about where each of them is in XYZ space
00:01:52
and where they're pointed. Yes. Wow.
00:01:54
Yeah. No, it's very cool. And so you can
00:01:57
have up to depending on how many
00:01:59
revolutions you do, you could have up to
00:02:01
a thousand photos of data with your
00:02:02
face, the pores, the irises, the hair,
00:02:04
the everything. So you're able to get
00:02:06
incredibly fine point data. Yes. Yes.
00:02:10
The point cloud is is stunning. So um
00:02:13
can I was told that you might scan me? I
00:02:15
would love to. High definition. I can't
00:02:18
There's never too many scans of one's
00:02:19
own head. Absolutely. Please. We will
00:02:21
step into Oh, I love how simple this is.
00:02:26
It's very simple, you know. Absolutely
00:02:28
beautiful. So much less of a software
00:02:31
load. Exactly. On what you've gathered.
00:02:33
Exactly. And the post-processing is
00:02:36
almost nothing because we get such a
00:02:37
clear high definition. So, it's like
00:02:38
it's throwing out a lot of its data just
00:02:41
getting the data. Exactly. Oh, man. Does
00:02:44
it know that it's does it uh prioritize
00:02:48
based on the focal point? So it gets rid
00:02:50
of all this data. Absolutely. Yes. So my
00:02:52
focal point is um exactly from your nose
00:02:55
to your ear and it it just forgets about
00:02:58
everything in the back. Fascinating.
00:03:00
It's really cool. It's fun. It also is
00:03:02
that large format the sensor. It is
00:03:04
large format. You notice it's the Fuji
00:03:06
large camera. Yes. And that's why it
00:03:09
makes a great anchor camera because it
00:03:11
just takes the most clear photos of you
00:03:13
while I, you know, do my wave pattern
00:03:16
and get everything else. Gotcha. All
00:03:17
right. All right.
00:03:19
Boom. Ready?
00:03:21
In. So, you'll be nice zen mode. You can
00:03:24
do eyes closed or open, but I'm going to
00:03:26
do them open. Excellent. Okay, Adam. In
00:03:29
three, two, one.
00:04:16
And in
00:04:17
three,
00:04:19
two, one.
00:04:22
Beautiful. And cut. Wow. So, we blasted
00:04:25
you with about 500 plus photos. High
00:04:28
resolution. Amazing. 50 megapixel. Now,
00:04:31
I have a question. If you're gathering
00:04:34
uh point cloud data at that fidelity,
00:04:36
yes, then there's certainly going to be
00:04:38
artifacts from my breathing that have to
00:04:40
get averaged out. Am I right? You're
00:04:43
absolutely correct.
00:04:45
Fortunately, this program, a reality
00:04:47
capture, negates that movement. It
00:04:50
negates breathing and also this camera
00:04:53
compensates for breath. What? It's
00:04:56
insane. That's totally insane. How how
00:05:00
do you know roughly how it compensates
00:05:01
for breath? Yeah, it it it it has this
00:05:04
um AI within the camera itself. Wow. So,
00:05:07
this is the first generation with the
00:05:09
Sony that uses the AI and it actually
00:05:13
can compensate your breathing. Oh, so
00:05:16
it's seeing it's being able to pick up
00:05:18
that micro movement. Yes. In real time.
00:05:21
In real time. And change its Oh my god.
00:05:24
translated into our program. It averages
00:05:26
out, right? And it says, "Okay, well,
00:05:28
Adam is chess is expanding just a little
00:05:31
in about 30 photos. It's this way in
00:05:34
about 100 it averages out." And it's
00:05:37
Now, I know that a lot of times you're
00:05:39
making stuff You're making stuff to fit
00:05:40
on bodies. You're making hard objects to
00:05:42
fit on soft bodies. Absolutely. So, have
00:05:44
you come across any cases in which uh it
00:05:47
falls down a little bit like it does
00:05:49
it's not a perfect fit or you have
00:05:51
institutional knowledge of adjustment
00:05:52
that you need? I have institutional
00:05:54
knowledge of adjustments and also um
00:05:56
with our laser scanner we uh do scale.
00:05:58
So we'll the real process we'll uh have
00:06:01
you over to the the laser scanner just
00:06:04
quickly scan your ear because that's
00:06:06
really hard to pick up from a monitor.
00:06:08
Exactly. So we import the laser data
00:06:11
into our program perfectly match your
00:06:13
point cloud and you you become you can
00:06:17
print out a mask wear it done. Amazing.
00:06:20
Yes. Wow. Yeah. It's It's awesome. That
00:06:23
is so cool. Yeah. And uh we actually did
00:06:26
a a 3D print of Bill here and uh it
00:06:30
turned out absolutely fantastic.
00:06:32
Stunning. It works on Tuesdays, right?
00:06:34
Yeah. Talk to Photography Tuesdays.
00:06:37
Absolutely. Uh that's really How long
00:06:39
have you guys been using this
00:06:41
technology? Um so basically I come from
00:06:43
a long line of photoggramometry
00:06:44
companies. Okay. And I just I just
00:06:47
couldn't stand, you know, Mike was so
00:06:49
wonderful. He says, "You know, here's a
00:06:51
massive budget. We need a cage. We need
00:06:53
a hundred cameras." And I say, "Mike, I
00:06:55
don't think you do. You don't." And he
00:06:58
was so generous. He just let me do my
00:07:01
system. Amazing. And it just turned out
00:07:03
he's very happy. And so will you be. So,
00:07:06
how long is the process of processing
00:07:08
that scan? Absolutely. So, we'll take
00:07:09
our um data set and we'll put it into uh
00:07:11
the reality capture software. The point
00:07:13
cloud takes about 10 minutes to
00:07:15
generate. To put a highdefinition mesh
00:07:18
on you, about a half an hour. Okay.
00:07:19
Texture information about another 20
00:07:22
minutes. Oh my god. So it'll you'll bake
00:07:24
in about under an hour. And you guys can
00:07:26
send us that data. Absolutely. We'd be
00:07:28
happy to. We'll even 3D print you out.
00:07:30
Amazing. Thank you so much. Thank you so
00:07:32
much.
00:07:34
All right, Adam. So, we have your 3D
00:07:37
print of the scan that David and his
00:07:40
team down at Sexual Motion did of you. I
00:07:43
am super fascinated to see how this
00:07:45
looks. Right off the bat, there's so
00:07:48
much more
00:07:49
texture than I have seen in other scans.
00:07:53
It's like the topography of my skin and
00:07:56
my wrinkles is spectacular. It was
00:07:58
really funny to pull up the model cuz it
00:08:00
was a textured model, right? They sent
00:08:02
not only the geometry but also the
00:08:05
imagery of your skin. So, it's grabbing
00:08:07
both of those at the same time. It's
00:08:08
grabbing both of those. And you know, if
00:08:10
you remember when he was doing the
00:08:11
scanning, it was the single camera that
00:08:14
parabola motion. Yeah. Uh I was a little
00:08:16
skeptical, I have to admit, but it gets
00:08:19
it your face all the way there. Indeed.
00:08:22
No, the face is it's incredibly
00:08:24
recognizable. It's very there's a lot of
00:08:26
personality there. Um the hair falls
00:08:28
apart a little bit, but that's not the
00:08:30
core competency of this scanning
00:08:31
technology. Anyway, and we've seen
00:08:33
plenty of scans that people have done of
00:08:35
you and hair is a tough thing to scan.
00:08:38
It's tough thing to like our friends at
00:08:39
FBFX do that thing where they have a
00:08:41
very modeled way of executing hair,
00:08:43
right? You basically have to kind of
00:08:44
remodel it from scratch. I love seeing
00:08:47
this noisy hair, noisy digital, right?
00:08:51
Uh you can This is exactly how you
00:08:52
imagine like a point cloud look in a
00:08:54
computer. And I thought it was really
00:08:56
cool to see it realize. I totally do
00:08:58
too. And there's a way in which because
00:09:00
my hair hasn't been stylized, it feels
00:09:03
very much like you're looking at a clay
00:09:05
sculptor. Oh yes. Yes. Very much. Oh
00:09:08
yeah. And the way I'm looking at it from
00:09:10
different angles. Yeah. it. They did a
00:09:13
great job. They really did. Yeah. Um and
00:09:15
the whole thing is without a massive
00:09:17
array of cameras, a single camera. I
00:09:20
mean, we always the old axiom of having
00:09:23
more having more tools just reduces the
00:09:25
amount of time you need. But they're
00:09:26
adding time and technique with a very
00:09:29
modern, powerful single camera. And
00:09:31
that's a pretty cool result. That single
00:09:33
camera is amazing. And they're using
00:09:35
this to create physical bodies for
00:09:37
working on prosthetics. Right. Right.
00:09:39
So, it's a very different use case.
00:09:40
maybe building a whole costume off of
00:09:42
the thing. Absolutely. It's the modern
00:09:43
life cast. Yeah. Um I this we did this
00:09:46
before off camera, but like it's just
00:09:50
that's that's my body half scale. That's
00:09:52
my head half scale. They seem to work as
00:09:54
appalling as it is to see them like
00:09:56
that. Sorry everybody. We're going to
00:09:58
add this and the file to the collection
00:10:02
of Adam scans. Amazing. Uh that given
00:10:05
how simple that process was, I there's
00:10:08
so much amazing detail in this. I'm
00:10:10
really impressed with the technology.
00:10:11
Yeah, very cool. Thanks to the team at
00:10:12
Spectral Motion for having us, for
00:10:14
giving us this demo and for adding
00:10:17
another atom scan to our collection. So
00:10:18
much fun. Thanks, guys.
00:10:22
Thank you guys so much for watching that
00:10:23
video. I would be remiss if I didn't
00:10:24
tell you we have some excellent merch.
00:10:26
We have a five-pack of Demerit badges
00:10:27
for sale right now at tested/store.com.
00:10:30
We've got the I hung it off of level
00:10:32
demerit badge. The I built the chair
00:10:34
backwards demerit badge. The I hit my
00:10:36
thumb with a hammer demerit badge. We've
00:10:38
got the stapler in my finger demerit
00:10:40
badge. And my favorite I stuck the duct
00:10:42
tape to itself dearit badge. Get yours
00:10:45
now.
00:10:48
Tested/store.store dashtore.
00:10:51
Tested
00:10:53
tested. It's right here. There. Just
00:10:55
click there.
![The Waters That Divide [Session 1] with Dr. Stephen Wellum](https://ruhnrawpgxrzdrgaohnf.supabase.co/storage/v1/object/public/images/video/twukMzJ18-U/thumbnail.jpg)
