Tyler StalmanGuest
Stephen RoblesHost
David SparksHost
I always enjoy your coverage of this because I feel like that's another thing that you really kind of opened my eyes to.

Yeah, I know they actually, they quietly changed the algorithm so that it is now leaning more on just using one lens.

Like, is it LiDAR? Is it the two different lenses measuring distance? So previously there was some amount of both lenses having some awareness of the distance of objects away.

Now it basically just looks at the RGB images like, okay, I know what to do with this because machine learning is where it is and it understands 3D space somehow.

What it doesn't do better though, is the actual cutting out of objects, especially hair, like really fine details.


So maybe I'll speak to like, if you already like using it, you can add it afterwards now, which is great.

And that I would say generally just don't make the background too blurry because that's when you start to see the cutouts looking kind of fake.

Just leave it like barely on and you'll get some of that separation still a lot more than the phone would have on its own, but you're not going to have.

Like I find when you turn it up, those videos are going to age like milk in five years when you're like, what were we doing? Like the algorithm is going to get better and you'll still be looking at a video from, from way back in 2026 when it wasn't perfected yet.

I love this part of your review where you're showing cinematic mode and it looks okay.

It's like, it's struggling to figure out where is it supposed to go? And then you basically tell us the viewer, like, if you want it to look like a film, like really professional, just shoot in log two and use a LUT because the colors and the look you get here looks more professional than just getting a blurry background where certain elements are just not clear.

I always enjoy your coverage of this because I feel like that's another thing that you really kind of opened my eyes to.

Yeah, I know they actually, they quietly changed the algorithm so that it is now leaning more on just using one lens.

Like, is it LiDAR? Is it the two different lenses measuring distance? So previously there was some amount of both lenses having some awareness of the distance of objects away.

Now it basically just looks at the RGB images like, okay, I know what to do with this because machine learning is where it is and it understands 3D space somehow.

What it doesn't do better though, is the actual cutting out of objects, especially hair, like really fine details.


So maybe I'll speak to like, if you already like using it, you can add it afterwards now, which is great.

And that I would say generally just don't make the background too blurry because that's when you start to see the cutouts looking kind of fake.

Just leave it like barely on and you'll get some of that separation still a lot more than the phone would have on its own, but you're not going to have.

Like I find when you turn it up, those videos are going to age like milk in five years when you're like, what were we doing? Like the algorithm is going to get better and you'll still be looking at a video from, from way back in 2026 when it wasn't perfected yet.

I love this part of your review where you're showing cinematic mode and it looks okay.

It's like, it's struggling to figure out where is it supposed to go? And then you basically tell us the viewer, like, if you want it to look like a film, like really professional, just shoot in log two and use a LUT because the colors and the look you get here looks more professional than just getting a blurry background where certain elements are just not clear.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.