newvoicemod

0:00

S… Speaker 1 (newvoicemod)

And again,

0:00

S… Speaker 1 (newvoicemod)

it works really well.

0:02

S… Speaker 1 (newvoicemod)

Conclusion.

0:03

S… Speaker 1 (newvoicemod)

So yeah,

0:04

S… Speaker 1 (newvoicemod)

this platform is actually very powerful.

0:07

S… Speaker 1 (newvoicemod)

From text or image,

0:08

S… Speaker 1 (newvoicemod)

to a full 3D model,

0:10

S… Speaker 1 (newvoicemod)

to a rigged and animated character,

0:13

S… Speaker 1 (newvoicemod)

all in just a few minutes.

0:14

S… Speaker 1 (newvoicemod)

That's a huge shift in how 3D content can be created.

0:18

S… Speaker 1 (newvoicemod)

It saves time,

0:19

S… Speaker 1 (newvoicemod)

simplifies the workflow,

0:21

S… Speaker 1 (newvoicemod)

and makes this kind of work much more accessible.

0:23

S… Speaker 1 (newvoicemod)

So yeah,

0:24

S… Speaker 1 (newvoicemod)

this is definitely a game changer.

0:27

S… Speaker 1 (newvoicemod)

You can find the link to the tool in the description below if you want to try it yourself.

0:31

S… Speaker 1 (newvoicemod)

Thanks for watching,

0:32

S… Speaker 1 (newvoicemod)

and I'll see you in the next video.

0:34

S… Speaker 1 (newvoicemod)

Welcome to the video.

0:35

S… Speaker 1 (newvoicemod)

Today,

0:36

S… Speaker 1 (newvoicemod)

I'm going to show you a tool that allows you to generate high -quality 3D

0:41

S… Speaker 1 (newvoicemod)

models from either text or images and use them directly in

0:45

S… Speaker 1 (newvoicemod)

your projects.

0:46

S… Speaker 1 (newvoicemod)

Now,

0:47

S… Speaker 1 (newvoicemod)

before we jump into testing it,

0:49

S… Speaker 1 (newvoicemod)

let me quickly explain what this tool actually does.

0:53

S… Speaker 1 (newvoicemod)

This tool is called Triple3D,

0:56

S… Speaker 1 (newvoicemod)

and it's an AI 3D model generator.

0:59

S… Speaker 1 (newvoicemod)

It allows you to create 3D models in two main ways,

1:03

S… Speaker 1 (newvoicemod)

text to 3D image to 3D.

1:06

S… Speaker 1 (newvoicemod)

So you can either describe what you want or upload

1:10

S… Speaker 1 (newvoicemod)

an image, and it will turn that into a full 3D model.

1:13

S… Speaker 1 (newvoicemod)

And these aren't just basic shapes.

1:15

S… Speaker 1 (newvoicemod)

The outputs are actually structured and usable.

1:18

S… Speaker 1 (newvoicemod)

One of the key features is intelligent segmentation.

1:23

S… Speaker 1 (newvoicemod)

This means the model is split into different editable parts instead of

1:27

S… Speaker 1 (newvoicemod)

being one single block,

1:29

S… Speaker 1 (newvoicemod)

so you can actually select,

1:30

S… Speaker 1 (newvoicemod)

modify,

1:31

S… Speaker 1 (newvoicemod)

and refine specific areas of the model,

1:34

S… Speaker 1 (newvoicemod)

which gives you much more control.

1:36

S… Speaker 1 (newvoicemod)

Then there's AI texturing.

1:38

S… Speaker 1 (newvoicemod)

With just one click,

1:39

S… Speaker 1 (newvoicemod)

you can apply high -resolution 4KPBR ready textures

1:44

S… Speaker 1 (newvoicemod)

to your model.

1:45

S… Speaker 1 (newvoicemod)

And there's also a magic brush,

1:47

S… Speaker 1 (newvoicemod)

which lets you repaint or adjust specific areas with precision without

1:51

S… Speaker 1 (newvoicemod)

affecting the whole model.

1:52

S… Speaker 1 (newvoicemod)

Another powerful feature is automatic rigging and animation.

1:56

S… Speaker 1 (newvoicemod)

So instead of just getting a static model,

1:58

S… Speaker 1 (newvoicemod)

the tool can generate a skeleton,

2:00

S… Speaker 1 (newvoicemod)

apply skin weights,

2:02

S… Speaker 1 (newvoicemod)

and make the character ready for animation.

2:04

S… Speaker 1 (newvoicemod)

This usually takes a lot of manual work,

2:07

S… Speaker 1 (newvoicemod)

but here it's done automatically.

2:09

S… Speaker 1 (newvoicemod)

And on top of that,

2:10

S… Speaker 1 (newvoicemod)

there's something really important.

2:12

S… Speaker 1 (newvoicemod)

The DCC bridge.

2:15

S… Speaker 1 (newvoicemod)

This means the tool connects directly with professional software like Blender

2:19

S… Speaker 1 (newvoicemod)

Maya, 3DS,

2:21

S… Speaker 1 (newvoicemod)

Max Unity,

2:22

S… Speaker 1 (newvoicemod)

Unreal Engine,

2:23

S… Speaker 1 (newvoicemod)

Godot,

2:24

S… Speaker 1 (newvoicemod)

and Cocos.

2:25

S… Speaker 1 (newvoicemod)

So you can easily move your models into your existing workflow without

2:29

S… Speaker 1 (newvoicemod)

extra steps.

2:30

S… Speaker 1 (newvoicemod)

Overall,

2:31

S… Speaker 1 (newvoicemod)

this tool turns hours of manual 3D work into something

2:35

S… Speaker 1 (newvoicemod)

you can do in minutes and integrates directly into real

2:39

S… Speaker 1 (newvoicemod)

production pipelines.

2:40

S… Speaker 1 (newvoicemod)

Now let's test it.

2:43

S… Speaker 1 (newvoicemod)

Demo 1.

2:44

S… Speaker 1 (newvoicemod)

Image to 3D.

2:45

S… Speaker 1 (newvoicemod)

Let's start by uploading an image of Ezreal from League of Legends

2:50

S… Speaker 1 (newvoicemod)

and see how it turns him into a 3D model.

2:52

S… Speaker 1 (newvoicemod)

While it's generating,

2:54

S… Speaker 1 (newvoicemod)

you can already see a preview forming,

2:56

S… Speaker 1 (newvoicemod)

kind of like a 3D outline of the character.

2:59

S… Speaker 1 (newvoicemod)

This is actually really cool because you can start visualizing

3:03

S… Speaker 1 (newvoicemod)

the result before it's finished,

3:05

S… Speaker 1 (newvoicemod)

and now the final model is ready.

3:08

S… Speaker 1 (newvoicemod)

And honestly,

3:10

S… Speaker 1 (newvoicemod)

this looks really good.

3:12

S… Speaker 1 (newvoicemod)

The proportions,

3:12

S… Speaker 1 (newvoicemod)

the structure.

3:13

S… Speaker 1 (newvoicemod)

It's very impressive for something generated automatically.

3:17

S… Speaker 1 (newvoicemod)

Demo 2.

3:19

S… Speaker 1 (newvoicemod)

Text to 3D object.

3:21

S… Speaker 1 (newvoicemod)

Now let's try something different.

3:24

S… Speaker 1 (newvoicemod)

Instead of an image,

3:26

S… Speaker 1 (newvoicemod)

we'll use a prompt.

3:27

S… Speaker 1 (newvoicemod)

Let's generate an ornate carved wooden throne chair.

3:31

S… Speaker 1 (newvoicemod)

Same process.

3:32

S… Speaker 1 (newvoicemod)

It builds the base shape first,

3:34

S… Speaker 1 (newvoicemod)

then refines the details.

3:36

S… Speaker 1 (newvoicemod)

And now we have the result.

3:38

S… Speaker 1 (newvoicemod)

And this is actually very well done.

3:42

S… Speaker 1 (newvoicemod)

The details are clean,

3:43

S… Speaker 1 (newvoicemod)

and it looks like something you could actually use as a real asset.

3:47

S… Speaker 1 (newvoicemod)

Demo 3.

3:48

S… Speaker 1 (newvoicemod)

Organic model.

3:50

S… Speaker 1 (newvoicemod)

Now let's test something more complex.

3:52

S… Speaker 1 (newvoicemod)

A flying hummingbird with iridescent blue,

3:55

S… Speaker 1 (newvoicemod)

purple, and green plumage.

3:56

S… Speaker 1 (newvoicemod)

This is a good test because it's more organic and detailed.

3:59

S… Speaker 1 (newvoicemod)

And here are the results.

4:00

S… Speaker 1 (newvoicemod)

And yeah,

4:01

S… Speaker 1 (newvoicemod)

this is impressive.

4:03

S… Speaker 1 (newvoicemod)

The shape,

4:04

S… Speaker 1 (newvoicemod)

the flow.

4:05

S… Speaker 1 (newvoicemod)

It works really well.

4:07

S… Speaker 1 (newvoicemod)

At this point,

4:08

S… Speaker 1 (newvoicemod)

I can say this tool genuinely delivers.

4:12

S… Speaker 1 (newvoicemod)

Rigging and animation.

4:13

S… Speaker 1 (newvoicemod)

Humanoid.

4:14

S… Speaker 1 (newvoicemod)

Now let's move to rigging.

4:16

S… Speaker 1 (newvoicemod)

So for rigging,

4:17

S… Speaker 1 (newvoicemod)

you go into this section,

4:19

S… Speaker 1 (newvoicemod)

and you can choose presets depending on your character.

4:21

S… Speaker 1 (newvoicemod)

There's one for humanoids,

4:23

S… Speaker 1 (newvoicemod)

and another one for animals.

4:24

S… Speaker 1 (newvoicemod)

So I used the humanoid preset,

4:27

S… Speaker 1 (newvoicemod)

and the rigging was successful.

4:29

S… Speaker 1 (newvoicemod)

What's interesting is that our character wasn't even in a proper T -pose,

4:33

S… Speaker 1 (newvoicemod)

which is usually recommended.

4:35

S… Speaker 1 (newvoicemod)

But it still worked.

4:37

S… Speaker 1 (newvoicemod)

Now let's preview an animation.

4:39

S… Speaker 1 (newvoicemod)

Let's try running.

4:40

S… Speaker 1 (newvoicemod)

And as you can see,

4:41

S… Speaker 1 (newvoicemod)

it's running really well.

4:42

S… Speaker 1 (newvoicemod)

Which is kind of crazy,

4:44

S… Speaker 1 (newvoicemod)

because just a few minutes ago,

4:45

S… Speaker 1 (newvoicemod)

this was just an image.

4:46

S… Speaker 1 (newvoicemod)

And now,

4:47

S… Speaker 1 (newvoicemod)

it's a fully animated 3D character.

4:50

S… Speaker 1 (newvoicemod)

That's honestly amazing.

4:52

S… Speaker 1 (newvoicemod)

Rigging.

4:53

S… Speaker 1 (newvoicemod)

Animal test.

4:55

S… Speaker 1 (newvoicemod)

Now let's try the bird.

4:56

S… Speaker 1 (newvoicemod)

There's a preset for animals,

4:58

S… Speaker 1 (newvoicemod)

so let's use that.

5:00

S… Speaker 1 (newvoicemod)

And the rigging worked?

5:00

S… Speaker 1 (newvoicemod)

As you can see,

5:01

S… Speaker 1 (newvoicemod)

the structure is correct,

5:02

S… Speaker 1 (newvoicemod)

but there's one limitation here.

5:04

S… Speaker 1 (newvoicemod)

There aren't any preset animations available for birds,

5:07

S… Speaker 1 (newvoicemod)

so we can't preview movement directly.

5:09

S… Speaker 1 (newvoicemod)

It's a bit of a downside,

5:11

S… Speaker 1 (newvoicemod)

but the rig itself is usable if you export it.

5:14

S… Speaker 1 (newvoicemod)

Final test,

5:16

S… Speaker 1 (newvoicemod)

T -Pose character.

5:18

S… Speaker 1 (newvoicemod)

Now for a final test,

5:20

S… Speaker 1 (newvoicemod)

let's do things properly.

5:22

S… Speaker 1 (newvoicemod)

We're going to generate a character in a T -Pose to

5:26

S… Speaker 1 (newvoicemod)

see how much better the results can be.

5:28

S… Speaker 1 (newvoicemod)

We'll go with something simple,

5:30

S… Speaker 1 (newvoicemod)

just a warrior.

5:32

S… Speaker 1 (newvoicemod)

So I'm generating it directly from text.

5:34

S… Speaker 1 (newvoicemod)

Now we wait.

5:35

S… Speaker 1 (newvoicemod)

And as it's generating,

5:37

S… Speaker 1 (newvoicemod)

we can already see the preview forming in a proper T -pose.

5:40

S… Speaker 1 (newvoicemod)

We just need to wait a bit more for the full model.

5:42

S… Speaker 1 (newvoicemod)

And yeah,

5:43

S… Speaker 1 (newvoicemod)

it takes a few minutes,

5:44

S… Speaker 1 (newvoicemod)

but the results are worth it.

5:46

S… Speaker 1 (newvoicemod)

All right,

5:47

S… Speaker 1 (newvoicemod)

now it's ready.

5:48

S… Speaker 1 (newvoicemod)

And here's our warrior.

5:50

S… Speaker 1 (newvoicemod)

As you can see,

5:51

S… Speaker 1 (newvoicemod)

it looks really good.

5:52

S… Speaker 1 (newvoicemod)

Clean structure,

5:53

S… Speaker 1 (newvoicemod)

good proportions,

5:54

S… Speaker 1 (newvoicemod)

solid base model.

5:56

S… Speaker 1 (newvoicemod)

Now let's rig it.

5:58

S… Speaker 1 (newvoicemod)

Alright,

5:59

S… Speaker 1 (newvoicemod)

rigging is done.

6:00

S… Speaker 1 (newvoicemod)

Let's test some animations.

6:01

S… Speaker 1 (newvoicemod)

Let's try the afraid animation.

6:03

S… Speaker 1 (newvoicemod)

Okay,

6:04

S… Speaker 1 (newvoicemod)

this looks good.

6:05

S… Speaker 1 (newvoicemod)

Now let's try boxing.

6:06

S… Speaker 1 (newvoicemod)

Yeah,

6:07

S… Speaker 1 (newvoicemod)

this is interesting.

6:09

S… Speaker 1 (newvoicemod)

The pose estimation capability is the thing that really makes you stop

6:13

S… Speaker 1 (newvoicemod)

and think.

6:14

S… Speaker 1 (newvoicemod)

The project is partly inspired by research out of Carnegie

6:18

S… Speaker 1 (newvoicemod)

Mellon University on something called dense pose from Wi -Fi.

6:22

S… Speaker 1 (newvoicemod)

The idea that you can reconstruct human body pose from

6:26

S… Speaker 1 (newvoicemod)

radio signals alone.

6:28

S… Speaker 1 (newvoicemod)

RuVue implements this using an architecture called YFlow,

6:32

S… Speaker 1 (newvoicemod)

which takes the processed CSI signals and maps them

6:36

S… Speaker 1 (newvoicemod)

to 17 key points on the human skeleton,

6:39

S… Speaker 1 (newvoicemod)

the same set of points used in standard computer vision pose estimation benchmarks.

6:44

S… Speaker 1 (newvoicemod)

Shoulders,

6:45

S… Speaker 1 (newvoicemod)

elbows, wrists,

6:46

S… Speaker 1 (newvoicemod)

hips,

6:47

S… Speaker 1 (newvoicemod)

knees, ankles,

6:48

S… Speaker 1 (newvoicemod)

head.

6:49

S… Speaker 1 (newvoicemod)

When trained with camera ground truth,

6:51

S… Speaker 1 (newvoicemod)

the system reaches 92 .9 % accuracy on the PC

6:56

S… Speaker 1 (newvoicemod)

-K at 20 benchmark,

6:57

S… Speaker 1 (newvoicemod)

which is a standard measure of pose estimation quality.

7:00

S… Speaker 1 (newvoicemod)

Without camera supervision,

7:02

S… Speaker 1 (newvoicemod)

training purely from Wi -Fi signals with no labeled data,

7:05

S… Speaker 1 (newvoicemod)

it still learns a meaningful body model in about 84 seconds on

7:10

S… Speaker 1 (newvoicemod)

a laptop.

7:10

S… Speaker 1 (newvoicemod)

The self -supervised training approach is one of the more

7:14

S… Speaker 1 (newvoicemod)

clever parts of the system.

7:16

S… Speaker 1 (newvoicemod)

because it means you don't need to go collect a specialized data set or run a

7:20

S… Speaker 1 (newvoicemod)

camera alongside the sensor forever.

7:22

S… Speaker 1 (newvoicemod)

You just let it watch people move around,

7:25

S… Speaker 1 (newvoicemod)

and it figures things out on its own.

7:27

S… Speaker 1 (newvoicemod)

One thing worth highlighting is the privacy architecture,

7:31

S… Speaker 1 (newvoicemod)

because I think it's actually a core part of what makes this interesting rather

7:35

S… Speaker 1 (newvoicemod)

than alarming.

7:36

S… Speaker 1 (newvoicemod)

RueView produces no images.

7:39

S… Speaker 1 (newvoicemod)

There's no video,

7:41

S… Speaker 1 (newvoicemod)

no photograph,

7:42

S… Speaker 1 (newvoicemod)

no visual record of any kind.

7:45

S… Speaker 1 (newvoicemod)

What it produces is structured data.

7:48

S… Speaker 1 (newvoicemod)

A position,

7:49

S… Speaker 1 (newvoicemod)

a breathing rate,

7:50

S… Speaker 1 (newvoicemod)

a skeleton of points in space.

7:52

S… Speaker 1 (newvoicemod)

That data lives on your local hardware.

7:55

S… Speaker 1 (newvoicemod)

Nothing is sent to a cloud server.

7:57

S… Speaker 1 (newvoicemod)

There's no account to create,

7:59

S… Speaker 1 (newvoicemod)

no terms of service,

8:00

S… Speaker 1 (newvoicemod)

no company receiving your information.

8:03

S… Speaker 1 (newvoicemod)

The entire system runs on devices you own,

8:06

S… Speaker 1 (newvoicemod)

in your home,

8:07

S… Speaker 1 (newvoicemod)

under your control.

8:08

S… Speaker 1 (newvoicemod)

This matters for a lot of reasons beyond privacy.

8:12

S… Speaker 1 (newvoicemod)

It means the system works without an internet connection.

8:15

S… Speaker 1 (newvoicemod)

It means it works if the company that made it disappears tomorrow.

8:19

S… Speaker 1 (newvoicemod)

And it means you can actually read the code and understand

8:23

S… Speaker 1 (newvoicemod)

exactly what it's doing,

8:24

S… Speaker 1 (newvoicemod)

because it's all right there in the repository.

8:27

S… Speaker 1 (newvoicemod)

The use cases this opens up are genuinely compelling.

8:32

S… Speaker 1 (newvoicemod)

Eldercare is the obvious one.

8:34

S… Speaker 1 (newvoicemod)

Being able to know that someone is okay,

8:37

S… Speaker 1 (newvoicemod)

that they're breathing,

8:38

S… Speaker 1 (newvoicemod)

that they haven't fallen,

8:39

S… Speaker 1 (newvoicemod)

without putting a camera in their bedroom or asking them to wear anything.

8:43

S… Speaker 1 (newvoicemod)

Sleep monitoring without a wearable.

8:46

S… Speaker 1 (newvoicemod)

Home security that doesn't rely on line -of -sight cameras.

8:50

S… Speaker 1 (newvoicemod)

Smart home automation that actually knows where you are and

8:54

S… Speaker 1 (newvoicemod)

what you're doing rather than guessing from motion sensors with a five -second

8:58

S… Speaker 1 (newvoicemod)

delay.

8:59

S… Speaker 1 (newvoicemod)

There are also more exotic applications documented in the project.

9:03

S… Speaker 1 (newvoicemod)

Disaster response,

9:05

S… Speaker 1 (newvoicemod)

searching for survivors through rubble,

9:07

S… Speaker 1 (newvoicemod)

monitoring industrial spaces for safety compliance,

9:10

S… Speaker 1 (newvoicemod)

even contactless blood pressure estimation using higher

9:14

S… Speaker 1 (newvoicemod)

frequency radar hardware.

9:16

S… Speaker 1 (newvoicemod)

The repository has examples and documentation

9:20

S… Speaker 1 (newvoicemod)

covering a lot of this,

9:21

S… Speaker 1 (newvoicemod)

and the architecture is designed to be extensible so you can build new

9:25

S… Speaker 1 (newvoicemod)

sensing capabilities on top of the existing pipeline.

9:29

S… Speaker 1 (newvoicemod)

If you want to try it yourself,

9:31

S… Speaker 1 (newvoicemod)

the fastest path is the Docker image,

9:33

S… Speaker 1 (newvoicemod)

one command,

9:34

S… Speaker 1 (newvoicemod)

and you have a running system.

9:36

S… Speaker 1 (newvoicemod)

If you want to go further,

9:38

S… Speaker 1 (newvoicemod)

the repository has a guided installer that walks you through hardware

9:42

S… Speaker 1 (newvoicemod)

detection, profile selection,

9:44

S… Speaker 1 (newvoicemod)

and getting everything set up.

9:46

S… Speaker 1 (newvoicemod)

The full documentation covers the API,

9:49

S… Speaker 1 (newvoicemod)

the WebSocket interface for real -time streaming,

9:51

S… Speaker 1 (newvoicemod)

how to flash and provision the ESP32 firmware,

9:55

S… Speaker 1 (newvoicemod)

and how to train your own models.

9:57

S… Speaker 1 (newvoicemod)

The GitHub repository is linked in the description.

10:00

S… Speaker 1 (newvoicemod)

It's called Ruvue,

10:01

S… Speaker 1 (newvoicemod)

it's by Ruvnet,

10:02

S… Speaker 1 (newvoicemod)

and it is absolutely worth an hour of your time to explore.

10:06

S… Speaker 1 (newvoicemod)

Whether you're a developer who wants to build something on top of it,

10:09

S… Speaker 1 (newvoicemod)

or just someone who finds it fascinating that this is possible at all.

10:13

S… Speaker 1 (newvoicemod)

Because honestly,

10:15

S… Speaker 1 (newvoicemod)

the fact that a $9 chip and some open source software can

10:19

S… Speaker 1 (newvoicemod)

do this with signals that are already in your home.

10:22

S… Speaker 1 (newvoicemod)

That's the kind of thing that changes how you think about what's possible.

10:26

S… Speaker 1 (newvoicemod)

Your Wi -Fi router is doing something right now that you have no idea about.

10:29

S… Speaker 1 (newvoicemod)

It's flooding every room in your home with radio waves.

10:33

S… Speaker 1 (newvoicemod)

Those waves pass through your walls,

10:35

S… Speaker 1 (newvoicemod)

your furniture,

10:36

S… Speaker 1 (newvoicemod)

your body,

10:37

S… Speaker 1 (newvoicemod)

and they bounce back every single second.

10:41

S… Speaker 1 (newvoicemod)

And for decades,

10:43

S… Speaker 1 (newvoicemod)

all of that information has just been thrown away.

10:46

S… Speaker 1 (newvoicemod)

Until now.

10:48

S… Speaker 1 (newvoicemod)

There's an open source project on GitHub called RueView,

10:51

S… Speaker 1 (newvoicemod)

and what it does is genuinely one of the most interesting things I've

10:55

S… Speaker 1 (newvoicemod)

come across in a long time.

10:57

S… Speaker 1 (newvoicemod)

It takes those radio waves that your router is already producing and

11:01

S… Speaker 1 (newvoicemod)

turns them into a sensing system.

11:03

S… Speaker 1 (newvoicemod)

It can detect people through walls.

11:05

S… Speaker 1 (newvoicemod)

It can measure your breathing rate and your heart rate without touching you.

11:09

S… Speaker 1 (newvoicemod)

It can track movement,

11:11

S… Speaker 1 (newvoicemod)

recognize when someone falls,

11:13

S… Speaker 1 (newvoicemod)

and even estimate body pose,

11:15

S… Speaker 1 (newvoicemod)

where your arms are,

11:16

S… Speaker 1 (newvoicemod)

where your legs are.

11:17

S… Speaker 1 (newvoicemod)

All from Wi -Fi signals alone.

11:19

S… Speaker 1 (newvoicemod)

No cameras,

11:21

S… Speaker 1 (newvoicemod)

no wearables,

11:22

S… Speaker 1 (newvoicemod)

nothing you need to put on or install in a way you'd ever notice.

11:25

S… Speaker 1 (newvoicemod)

Let me show you how it works.

11:28

S… Speaker 1 (newvoicemod)

So the first thing you're going to notice when you land on the repository is that

11:32

S… Speaker 1 (newvoicemod)

this project is enormous.

11:34

S… Speaker 1 (newvoicemod)

This isn't a weekend experiment someone threw together.

11:37

S… Speaker 1 (newvoicemod)

There are over 400 pull requests,

11:40

S… Speaker 1 (newvoicemod)

79 architectural decision records documenting every

11:44

S… Speaker 1 (newvoicemod)

major technical choice,

11:45

S… Speaker 1 (newvoicemod)

a full Rust rewrite of the signal processing pipeline,

11:48

S… Speaker 1 (newvoicemod)

ESP32 firmware,

11:51

S… Speaker 1 (newvoicemod)

Docker images,

11:52

S… Speaker 1 (newvoicemod)

a desktop app in progress,

11:54

S… Speaker 1 (newvoicemod)

and over 1 ,000 tests.

11:57

S… Speaker 1 (newvoicemod)

This is a seriously well -engineered piece of software,

12:00

S… Speaker 1 (newvoicemod)

and it's all open source under the MIT license,

12:03

S… Speaker 1 (newvoicemod)

meaning you can use it,

12:05

S… Speaker 1 (newvoicemod)

modify it,

12:06

S… Speaker 1 (newvoicemod)

build on it,

12:07

S… Speaker 1 (newvoicemod)

do whatever you want with it.

12:09

S… Speaker 1 (newvoicemod)

The project is built around a concept called Channel State

12:13

S… Speaker 1 (newvoicemod)

Information,

12:13

S… Speaker 1 (newvoicemod)

or CSI.

12:15

S… Speaker 1 (newvoicemod)

Your router and your devices are constantly exchanging Wi -Fi signals,

12:20

S… Speaker 1 (newvoicemod)

and buried inside those signals is extremely detailed information

12:24

S… Speaker 1 (newvoicemod)

about how the radio environment of a room is behaving,

12:27

S… Speaker 1 (newvoicemod)

which frequencies are strong,

12:29

S… Speaker 1 (newvoicemod)

which are weak,

12:29

S… Speaker 1 (newvoicemod)

how the signal is bouncing around.

12:31

S… Speaker 1 (newvoicemod)

Most devices just use this to improve connection quality

12:35

S… Speaker 1 (newvoicemod)

and throw the rest away.

12:37

S… Speaker 1 (newvoicemod)

RooView captures it and reads it like a map of everything happening

12:41

S… Speaker 1 (newvoicemod)

in the space.

12:43

S… Speaker 1 (newvoicemod)

Now, the hardware side of this is where it gets interesting.

12:46

S… Speaker 1 (newvoicemod)

To actually capture CSI data,

12:49

S… Speaker 1 (newvoicemod)

you need a device that exposes it,

12:51

S… Speaker 1 (newvoicemod)

and the one RuView is built around is the ESP32

12:56

S… Speaker 1 (newvoicemod)

-S3,

12:57

S… Speaker 1 (newvoicemod)

a tiny microcontroller that you can get for about $9.

13:01

S… Speaker 1 (newvoicemod)

The project has custom firmware for it that streams CSI

13:05

S… Speaker 1 (newvoicemod)

data over UDP at around 20 to 28 frames per second.

13:11

S… Speaker 1 (newvoicemod)

You flash the firmware onto the chip,

13:13

S… Speaker 1 (newvoicemod)

connect it to your Wi -Fi network,

13:15

S… Speaker 1 (newvoicemod)

point it at your sensing server,

13:16

S… Speaker 1 (newvoicemod)

and you're collecting data.

13:18

S… Speaker 1 (newvoicemod)

If you don't have any hardware and just want to try the software,

13:22

S… Speaker 1 (newvoicemod)

there's a Docker image that runs the whole thing with simulated data.

13:26

S… Speaker 1 (newvoicemod)

You pull it,

13:27

S… Speaker 1 (newvoicemod)

run it,

13:27

S… Speaker 1 (newvoicemod)

and open localhost colon 3000,

13:29

S… Speaker 1 (newvoicemod)

and you've got a live demo running in under a minute.

13:32

S… Speaker 1 (newvoicemod)

That's actually a great way to explore what the system is doing before

13:37

S… Speaker 1 (newvoicemod)

you commit to any hardware.

13:39

S… Speaker 1 (newvoicemod)

Once the data is flowing,

13:41

S… Speaker 1 (newvoicemod)

the signal processing pipeline takes over,

13:43

S… Speaker 1 (newvoicemod)

and this is where the real engineering lives.

13:46

S… Speaker 1 (newvoicemod)

Raw CSI data is messy.

13:48

S… Speaker 1 (newvoicemod)

It's full of noise,

13:49

S… Speaker 1 (newvoicemod)

phase errors,

13:50

S… Speaker 1 (newvoicemod)

reflections from objects that aren't people.

13:53

S… Speaker 1 (newvoicemod)

RueView runs it through a series of research -grade algorithms,

13:57

S… Speaker 1 (newvoicemod)

things like spot -five phase correction,

13:59

S… Speaker 1 (newvoicemod)

Hample outlier filtering,

14:00

S… Speaker 1 (newvoicemod)

and Fresnel zone modeling,

14:02

S… Speaker 1 (newvoicemod)

to clean the signal and isolate what's actually caused by human

14:06

S… Speaker 1 (newvoicemod)

presence in movement.

14:08

S… Speaker 1 (newvoicemod)

After cleaning,

14:09

S… Speaker 1 (newvoicemod)

the system separates out different frequency bands.

14:12

S… Speaker 1 (newvoicemod)

Breathing happens at 0 .1 to 0 .5 hertz.

14:16

S… Speaker 1 (newvoicemod)

Heart rate sits between 0 .8 and 2 .0

14:20

S… Speaker 1 (newvoicemod)

hertz.

14:20

S… Speaker 1 (newvoicemod)

Movement produces much higher frequency changes.

14:24

S… Speaker 1 (newvoicemod)

The pipeline isolates each of these independently and extracts

14:28

S… Speaker 1 (newvoicemod)

them as separate signals,

14:29

S… Speaker 1 (newvoicemod)

which is how you get breathing rate and heart rate as clean numerical readings

14:33

S… Speaker 1 (newvoicemod)

rather than just noise.

14:35

S… Speaker 1 (newvoicemod)

The presence detection piece is probably the most immediately practical

14:40

S… Speaker 1 (newvoicemod)

capability.

14:40

S… Speaker 1 (newvoicemod)

The system learns what your room looks like when it's empty,

14:44

S… Speaker 1 (newvoicemod)

the static reflections from your walls and furniture,

14:47

S… Speaker 1 (newvoicemod)

and then continuously subtracts that baseline from every incoming

14:51

S… Speaker 1 (newvoicemod)

frame.

14:52

S… Speaker 1 (newvoicemod)

What's left is only the signal caused by something that changed.

14:56

S… Speaker 1 (newvoicemod)

A person walking in.

14:58

S… Speaker 1 (newvoicemod)

Someone breathing in the corner.

15:00

S… Speaker 1 (newvoicemod)

A fall.

15:02

S… Speaker 1 (newvoicemod)

This works through walls because Wi -Fi passes through most building materials

15:06

S… Speaker 1 (newvoicemod)

without significant attenuation.

15:08

S… Speaker 1 (newvoicemod)

You can put the sensor in one room and detect presence in the

15:12

S… Speaker 1 (newvoicemod)

next room over.

15:13

S… Speaker 1 (newvoicemod)

And the system doesn't just detect whether someone is there.

15:16

S… Speaker 1 (newvoicemod)

It tracks them.

15:18

S… Speaker 1 (newvoicemod)

With multiple nodes placed around a space,

15:20

S… Speaker 1 (newvoicemod)

it can tell you where each person is,

15:22

S… Speaker 1 (newvoicemod)

how many people are present,

15:24

S… Speaker 1 (newvoicemod)

and follow them as they move.

15:26

S… Speaker 1 (newvoicemod)

The multi -node fusion works by combining the signal from every sensor

15:30

S… Speaker 1 (newvoicemod)

simultaneously,

15:31

S… Speaker 1 (newvoicemod)

and the more overlap you have between their fields of view,

15:34

S… Speaker 1 (newvoicemod)

the more precise the tracking gets.

15:37

S… Speaker 1 (newvoicemod)

Imagine turning just a text prompt into a full 3D character

15:41

S… Speaker 1 (newvoicemod)

in minutes.

15:42

S… Speaker 1 (newvoicemod)

This AI tool does exactly that.

15:45

S… Speaker 1 (newvoicemod)

Just type something like alien,

15:47

S… Speaker 1 (newvoicemod)

and it instantly builds a full 3D model with structure,

15:51

S… Speaker 1 (newvoicemod)

depth,

15:52

S… Speaker 1 (newvoicemod)

and detail.

15:53

S… Speaker 1 (newvoicemod)

No modeling,

15:54

S… Speaker 1 (newvoicemod)

no sculpting,

15:55

S… Speaker 1 (newvoicemod)

just text to production -ready 3D assets.

15:58

S… Speaker 1 (newvoicemod)

AI is changing 3D forever.

16:01

S… Speaker 1 (newvoicemod)

This AI can turn a simple image into a fully animated 3D character.

16:05

S… Speaker 1 (newvoicemod)

First,

16:07

S… Speaker 1 (newvoicemod)

you upload an image,

16:08

S… Speaker 1 (newvoicemod)

and it rebuilds it as a 3D model,

16:11

S… Speaker 1 (newvoicemod)

automatically.

16:12

S… Speaker 1 (newvoicemod)

But it doesn't stop there.

16:14

S… Speaker 1 (newvoicemod)

It can also rig and animate it.

16:17

S… Speaker 1 (newvoicemod)

So one image becomes a playable 3D character in

16:21

S… Speaker 1 (newvoicemod)

minutes.

16:21

S… Speaker 1 (newvoicemod)

Now let's try something else.

16:23

S… Speaker 1 (newvoicemod)

Let's go with something very simple.

16:25

S… Speaker 1 (newvoicemod)

Just one word.

16:26

S… Speaker 1 (newvoicemod)

Alien.

16:27

S… Speaker 1 (newvoicemod)

So it's generating.

16:28

S… Speaker 1 (newvoicemod)

And already,

16:30

S… Speaker 1 (newvoicemod)

from the preview,

16:31

S… Speaker 1 (newvoicemod)

we can see it's forming something that clearly looks like an alien.

16:34

S… Speaker 1 (newvoicemod)

Which is interesting,

16:36

S… Speaker 1 (newvoicemod)

because this is just a single word prompt.

16:38

S… Speaker 1 (newvoicemod)

No extra detail.

16:40

S… Speaker 1 (newvoicemod)

Alright,

16:42

S… Speaker 1 (newvoicemod)

now the result is ready.

16:43

S… Speaker 1 (newvoicemod)

And here it is.

16:45

S… Speaker 1 (newvoicemod)

A blue alien.

16:46

S… Speaker 1 (newvoicemod)

This is actually pretty cool.

16:48

S… Speaker 1 (newvoicemod)

I like this.

16:50

S… Speaker 1 (newvoicemod)

Now let's take it a step further and rig this model.

16:54

S… Speaker 1 (newvoicemod)

Alright,

16:55

S… Speaker 1 (newvoicemod)

the rigging is done.

16:56

S… Speaker 1 (newvoicemod)

Let's try some animations.

16:59

S… Speaker 1 (newvoicemod)

Let's start with something like an angry animation.

17:01

S… Speaker 1 (newvoicemod)

And yeah,

17:03

S… Speaker 1 (newvoicemod)

this actually looks good.

17:04

S… Speaker 1 (newvoicemod)

Now let's try something more fun.

17:06

S… Speaker 1 (newvoicemod)

Maybe running,

17:07

S… Speaker 1 (newvoicemod)

or even dancing.

17:08

S… Speaker 1 (newvoicemod)

Let's try dancing.

17:10

S… Speaker 1 (newvoicemod)

Yeah,

17:11

S… Speaker 1 (newvoicemod)

this is actually funny.

17:13

S… Speaker 1 (newvoicemod)

I like this.

17:14

S… Speaker 1 (newvoicemod)

It's kind of crazy when you think about it.

17:16

S… Speaker 1 (newvoicemod)

Just one word,

17:17

S… Speaker 1 (newvoicemod)

and now we have a fully rigged and animated 3D character.

17:20

S… Speaker 1 (newvoicemod)

That's actually very interesting.

17:22

S… Speaker 1 (newvoicemod)

This means the tool connects directly with professional software,

17:25

S… Speaker 1 (newvoicemod)

like Blender,

17:27

S… Speaker 1 (newvoicemod)

Maya,

17:27

S… Speaker 1 (newvoicemod)

3DS Max,

17:29

S… Speaker 1 (newvoicemod)

Unity,

17:29

S… Speaker 1 (newvoicemod)

Unreal Engine,

17:31

S… Speaker 1 (newvoicemod)

Godot,

17:31

S… Speaker 1 (newvoicemod)

and Cocos.

תקציר

שאל את אל על התעתיק הזה.