DJI Air 3S - Rumours, leaks, gossip and general discussion

[PingSpike]

@Leylo1971

the antennas look like someone on here’s printed them

Rude

[Leylo1971]

@PingSpike

Rude

Sorry Rich, didn’t mean to offend any 3d printer dudes that reside on the forum, I’m actually sure they could better the antennas on the RC2 anyways😂

[Leylo1971]

Did a bit of testing with the stills cam on the drone today. Totally static with the drone on a wall…coz it was toooo fffin windy (even for me)

I watched Ian in Londons vid the other night that maybe hinted that the 50mp camera was not in fact a true 50meg lens but a quad beaur ( probably spelled wrong) . So, after gettin me nuts in a twizz after spending X amount on the 3S and going to bin my 2S( proper 1" sensor an never let me down), which I kept faith in and not wanting the Air3 because of the ‘fake’ 48 mp cam, I decided today to go do my own tests,no science here.

Tooooo windy , so propped on a wall outside Dalton church

AEB X 12MP merged in lightroom, NO changes made. just merged

AEB X 5 50mp … same as above

Just for the fun of it, I compared manual exposure bracket with Auto so heres the man…X 5 , 1 stop in between.

Like I said ,no science here, but you can defo tell the difference with the 5 auto being well the best, detail and colour wise.

Any input from you rascals out there that are Pros

[Kirky]

Hey @Leylo1971 Ady ( I’m defo no pro but my thoughts ) there very very clear pictures no two ways about it but, I’ll tell you what id like to see ( well me anyway ) id love to see a comparison with the same layout as you done above but put the air 2s into the mix just for comparison no pressure, you’ve all ways took good pictures regardless “but yes” would love to see how much of a difference there is, is that possible

[Leylo1971]

@Kirky

air 2s

Totally with you on that Rich, I actually had the same thought as I was putting the post together.

On the evidence off them snaps, disregarding the manual ones, the 50mp are way more detailed. Not in the case of a pixel being spread into 4, but actual real pixels. But I’m defo no expert, just move sliders in Lightroom.

Next weekend I think I’ll do the exact same test, on a wall, an like you said, involve the 2S. Curious as I am.

[Kirky]

I’ve always took note of your pictures that come out of the air 2s ( I’ve never owned one ) still learning myself etc but always said to myself the quality that comes out of it are fantastic so yes would be great to see mate & I think your the perfect candidate to show us all
Thanks ….

[Leylo1971]

@Kirky

perfect candidate

Haha, thanks mate. Really don’t know about that🤣

I had the same idea meself while typing that first post, 2S vs 3S

I’m only really doing this because basically, well, I was well happy with the 2S and it’s camera, but as great as the actual drone is, there’s always room for improvement. The Air 3 give the impression it was the better drone, and it is, all but for the camera. In my humble opinion the camera isn’t as good as the 2S.. No doubt I’ll be had off by the Air 3 owners over this🤔.

So I waited for the 3S. Then watching one of I in Lon vids he mentioned that it may be a quad baeur( probably still spell wrong ).

I have a bit of a conclusion on this…on the mini 3 pro, the difference in detail from the 12 to 48 mp images is only marginal. The difference between the images I posted between the the 2 specs are VERY noticeable.

Like I’ve asked, if there’s anybody with any relevance to to photography professional wise, that could give us a heads up on the images, I be more more than welcome to give them sexual favour’s…ahem, really pleased😁

[ParmoPaul]

@Leylo1971

I watched Ian in Londons vid the other night that maybe hinted that the 50mp camera was not in fact a true 50meg lens but a quad beaur ( probably spelled wrong)

I wrote this just further up the page…

DJI Air 3S - Rumours, leaks, gossip and general discussion

It’s the other way round isn’t it, 48MP sensor that outputs a 12MP image due to QB. That way you are getting 4 sensor pixels for every image pixel. And why you can force the sensor output to give 48MP image files. Edit : yep, looks like Motorola have a 50MP sensor in their phones so you end up with a 12.5MP image.

It is a 50MP sensor. Quad Bayer-ing takes four of those pixels for a colour to make a 12MP (well 12.5) image.

There are many pages on the subject if you want to read further, like this one

Quad Bayer sensors: what they are and what they are not

[Drumsagard]

On first look the 50mp does look the better quality image, it’s sharper and more detailed in the centre. If you look at the corners though at 100% the 12mp image has no noise, whereas the 50mp does and is also less sharp.

12mp

50mp

50mp man

Would be good to see this side by side test done with a well exposed single image, not an AEB.
There does seem to be a lack of sharpness at the edges of the 50mp, but it is difficult to tell as it’s a AEB. You often get problems merging AEB’s together when there’s been movement in trees, clouds etc.

[DroneGeek]

@ianinlondon

paying for the bright screen and no different to their ‘smart controller

The pro controller is solid and feels far superior to the more basic cheaper version.

I guess we could compare cars or camper vans.

A cheap basic car will still get you from A to B, but some people prefer comfort and bells and whistles.

You could take your trips in a very basic camper, but prefer something a bit better.

Oh plus you can’t load other software on the smart controller

Once you have used the pro for several flights, you wouldn’t want to go back.

In the same way, as I have no intention of flying economy again

[Leylo1971]

@ParmoPaul

There are many

Cheers mate, I’ll have a little read👍

[Leylo1971]

@DroneGeek

In the same way, as I have no intention of flying economy again

Very nicely said Ian. Exactly what I was banging on about​:+1:

[Leylo1971]

@Drumsagard

with a well exposed single image

Yeah I get what you mean John. Originally the test was to see what kind of result came from using the AEB option on the drone as opposed to me manually changing the shutter speed. I believe the auto option changes the EV by 0.7, where as I’ll change it by 1.0. I think using the 7 image option will virtually give me the same my 5 x 1.0 manually does, if you know what I mean. It’s quicker and less hassle, may make a difference with any ghosting in windy conditions if I’m thinking rightly🤔

Only thing is, the images at 4:3 are around 68mb,thats going to be a lot of space .

I’m out and about today, so maybe like you said above I’ll take some single shots to compare.

Cheers John👍

[new_wave_24]

Anyone tested out waypoints to see if they are more accurate and consistent between runs?

[Leylo1971]

@Drumsagard

Here you are John, some images what I took last night 2S vs 3S.

day 2s (1).DNG (39.0 MB)
Air 2s

day 3s (2).DNG (67.3 MB)
Air 3S

night 2s (1).DNG (38.9 MB)
Air 2S night

night 3s (2).DNG (53.1 MB)
Air 3S night

sunset 2s (1).DNG (56.2 MB)
Air 2S sunset

sunset 3s(2).DNG (38.9 MB)
Air 3S sunset

As you’ll see, there’s a mark up in detail between the 2. I did notice a greeny cast on the day time shot from the 3S . The colour depth seems a hell of a lot better on the 3S and a tad more contrasty.

All the shots were taken with iso100, EV 0.0, apart from the sunset of the 3S being 0.3 negative for some reason.

All straight out of drone.

You’ll probably be able to see many differences I can’t

[Drumsagard]

Ady, there certainly is a huge increase in detail from the 2s to the 3s. The one thing that is evident on the 50mp 3s images still, is the noise in the shadow areas, though Photoshop’s denoise AI does a decent job of sorting it. The colour, detail and tonal range is pretty impressive.

Here’s the bottom left corner at 100%.

Photoshop denoise AI version.

[Leylo1971]

@Drumsagard

the noise in the shadow areas

Glad you noticed it, because I thought I was going barmy. There’s quite a shed load isn’t there. The snaps that I took of the Singing ringing tree the other day were taken before sunrise, and when looking at them in LR, I actually thought LR had given me the DNG preview . I used the AI denoise in LR, that did a pretty good job, but then mopped up with Topaz, and didn’t turn out that bad.

I’m pretty peeved with the situation, with the 3S hoping to replace the 2S. Here’s hoping for some update that improves the noise level🤔.

All in all, I’m over the moon with the new acquisition, brilliant drone, great update. It’s just that issue letting it down.

Cheers John for the input as always👍

Edut…It’s very funny that no reviewers mention it, apart from Billy Kyle, and he’s on about video🤷

[Drumsagard]

@Leylo1971

Here’s hoping for some update that improves the noise level🤔.

Did you shoot anymore 12mp images? I didn’t notice any noise in the first 12mp ones you posted.

[Leylo1971]

@Drumsagard

12mp images

No, but it did cross me mind. The only thing was I thought to meself, why bother with the 12 Meg mode when I have a 2S and a Mini 3 P. But after reading your post earlier on, having a mundane task at work, began thinking…

1. Air 2S is 4yrs old or so, so it may balance out with there being new tech inside the 3S.
2. The Mini 3 P hasn’t a 1" sensor, so the 12 meg from the 3S should be better🤷.

I think I might have another little side by side test going on again sometime this week at night🤔

[BingBong]

Thanks. You got the file names the wrong way round on the sunset shots. Model number in the EXIF data, and the file size gives it away. Daytime comparison of the 2S and the 3S in both 12MP and 50MP would be very useful. I suspect the 12MP will have more in-built noise processing.