MarcoR Posted at 10-25 08:08
Direct comparison by taking frames from your video.
Regardless of Youtube compression and considering that I can only download your video in fullhd the difference is there.
In the flower spread you can see it very well.

Now we need matching lenses.

djiuser_zEKgdWRops1S Posted at 10-25 08:15
Now we need matching lenses.

Now I don't think this can be solved simply with a lens.
I thought so before, now that I've seen your results I don't.
Then of course in case I would be delighted to be proven wrong.
But this shows that the sensor can give more.

Another one, taken in RAW, developed with Lightroom mobile standard settings, think it looks pretty decent (ISO 100, even if taken through the window of my car)


bilder bei google bilder hochladen

MarcoR Posted at 10-25 08:27
Now I don't think this can be solved simply with a lens.
I thought so before, now that I've seen your results I don't.
Then of course in case I would be delighted to be proven wrong.

This test does not show that you can get more from the sensor without losing sharpness in the distance.  Maybe, like another forum member said, a smaller correction can improve the overall sharpness...

fish sticks Posted at 10-25 11:03
This test does not show that you can get more from the sensor without losing sharpness in the distance.  Maybe, like another forum member said, a smaller correction can improve the overall sharpness...

Of course, as I had previously written perhaps a +2 lens might suffice.
But personally I would use this test precisely just to see if the sensor could get a less blurry image in the center, having ascertained this only DJI could solve with a new lens system.
An applicable solution in front would be posturing.
The replaced cam from DJI will be delivered to me today.

MarcoR Posted at 10-25 20:51
Of course, as I had previously written perhaps a +2 lens might suffice.
But personally I would use this test precisely just to see if the sensor could get a less blurry image in the center, having ascertained this only DJI could solve with a new lens system.
An applicable solution in front would be posturing.

Cant wait for your tests , good luck !

BigTB Posted at 10-25 05:33
Could be a lens collimation issue.  I would contact DJI support and see what they say.

I contacted them. Here is their answer:

MarcoR Posted at 10-25 05:59
Please write in English here.
Then it could be the same misalignment defect.
Forget the retailer on AliExpress, contact DJI directly through the support area on their site.

I think the same. The lens has no tolerance with the matrix, stands a little at an angle, or is not completely screwed to the main board.

Zarg Posted at 10-25 22:43
I think the same. The lens has no tolerance with the matrix, stands a little at an angle, or is not completely screwed to the main board.

What distance were you with the cam in that video you took?
And what mode did you record in? Wide?
Is that video a crop of the original recorded?

MarcoR Posted at 10-25 20:51
Of course, as I had previously written perhaps a +2 lens might suffice.
But personally I would use this test precisely just to see if the sensor could get a less blurry image in the center, having ascertained this only DJI could solve with a new lens system.
An applicable solution in front would be posturing.

I can't wait for you to share the tests with the new camera

MarcoR Posted at 10-25 22:50
What distance were you with the cam in that video you took?
And what mode did you record in? Wide?
Is that video a crop of the original recorded?

The distance is 30 cm. The mode is standard (not wide), it can be seen that there is no distortion. Video original, no editing.

Another example i shot while traveling:

https://drive.google.com/file/d/ ... LF/view?usp=sharing

TNQ Posted at 10-26 03:11
Another example i shot while traveling:

https://drive.google.com/file/d/1Dh5zV6nc7y9wcb8s0BDyS6qKzm4wRlLF/view?usp=sharing

Only 360p for now 1080p now

StevoB Posted at 10-26 03:16
Only 360p for now 1080p now

No its raw video straight out of camera in 4K.

It probably takes until the highest resolution size, like on YT, when I insert a video....

StevoB Posted at 10-26 03:27
It probably takes until the highest resolution size, like on YT, when I insert a video....

Yep could be the case

Zarg Posted at 10-25 22:53
The distance is 30 cm. The mode is standard (not wide), it can be seen that there is no distortion. Video original, no editing.

Excuse me, yours is an A4 sheet like mine, these are your settings, how can your video be at 30 cm uncropped if this is the field of view I have at 32 cm with the same settings as yours with the zoom in the app at 1.0x??
First image in D-Cinelike, the second natural color, frames from a video.
Nice to note that in D-Cinelike the focus is really bad, in Normal much better.
Also during these tests that lasted a little over 10 minutes, the cam started to heat up and send the video to the smartphone late, lagged, you can see the screen showing something else.

The new one?

StevoB Posted at 10-26 07:22
The new one?

Yes, empirical tests indoors, tomorrow I try outdoors.
But this D-Cinelike thing destabilizes me; my previous tests were all done only in D-Cinelike.
There is an abysmal difference in focus.
Could you please verify this as well?
I have already reported this to DJI.

They have previously stated the increased sharpening is not applied to cinelike files as they assume you will do it yourself if needed

4wd Posted at 10-26 07:33
They have previously stated the increased sharpening is not applied to cinelike files as they assume you will do it yourself if needed

Okay, so as a default it's not that good.

MarcoR Posted at 10-26 07:00
Excuse me, yours is an A4 sheet like mine, these are your settings, how can your video be at 30 cm uncropped if this is the field of view I have at 32 cm with the same settings as yours with the zoom in the app at 1.0x??
First image in D-Cinelike, the second natural color, frames from a video.
Nice to note that in D-Cinelike the focus is really bad, in Normal much better.

Seems like you have a good one now ?? For my eyes it seems the paper is in focus better than before..

Did you do a factory reset ? (maybe after update?)

TNQ Posted at 10-26 07:37
Seems like you have a good one now ?? For my eyes it seems the paper is in focus better than before..

Did you do a factory reset ? (maybe after update?)

Yes, factory reset after the firmware update, and then the accel/gyro calibration.
Seems to go better considering that it is "Standard" FOV footage, so already very much internally cropped by the cam, but with the other one I had never recorded without D-Cinelike, damn!
In fact the result in D-Cinelike is almost identical.
It would be much better if DJI put adjustable sharpness, three modes would be enough.
Tomorrow I'll try again outdoors.

MarcoR Posted at 10-26 07:00
Excuse me, yours is an A4 sheet like mine, these are your settings, how can your video be at 30 cm uncropped if this is the field of view I have at 32 cm with the same settings as yours with the zoom in the app at 1.0x??
First image in D-Cinelike, the second natural color, frames from a video.
Nice to note that in D-Cinelike the focus is really bad, in Normal much better.

MarcoR А3

Zarg Posted at 10-26 07:52
MarcoR А3

Ahhhh damn, ok!
However the A3 is little bigger than A4, to get to the edges like your I would have had to put an A0 on it!

MarcoR Posted at 10-26 07:38
Yes, factory reset after the firmware update, and then the accel/gyro calibration.
Seems to go better considering that it is "Standard" FOV footage, so already very much internally cropped by the cam, but with the other one I had never recorded without D-Cinelike, damn!
In fact the result in D-Cinelike is almost identical.

Totally forgot to ask , is it a brand new box or the previous one is fixed and sent back ?

MarcoR Posted at 10-26 07:38
Yes, factory reset after the firmware update, and then the accel/gyro calibration.
Seems to go better considering that it is "Standard" FOV footage, so already very much internally cropped by the cam, but with the other one I had never recorded without D-Cinelike, damn!
In fact the result in D-Cinelike is almost identical.

Totally forgot to ask , is it a brand new box or the previous one is fixed and sent back ?

TNQ Posted at 10-26 08:20
Totally forgot to ask , is it a brand new box or the previous one is fixed and sent back ?

New, sealed with cellophane.
I don't think it can be a refurbished one, they didn't even have time to pick up some to repair and distribute, in August 2022 it still had to go out on the market.
So still 2022/08, but other serial starting with "5S4DK8R," also to the other friend always in replacement same initial part.

MarcoR Posted at 10-26 08:37
New, sealed with cellophane.
I don't think it can be a refurbished one, they didn't even have time to pick up some to repair and distribute, in August 2022 it still had to go out on the market.
So still 2022/08, but other serial starting with "5S4DK8R," also to the other friend always in replacement same initial part.

Ehi Marco! how is going the "new" camera?

MarcoR Posted at 10-26 08:37
New, sealed with cellophane.
I don't think it can be a refurbished one, they didn't even have time to pick up some to repair and distribute, in August 2022 it still had to go out on the market.
So still 2022/08, but other serial starting with "5S4DK8R," also to the other friend always in replacement same initial part.

It looks better now or... ?

............ or

They don't have a chance to repair the optics in such a short time. If only they had worked on it earlier and the new batches would have already been treated. New (cover) - old (optics) will also come to me.

In my opinion nothing has changed.
Even if the shots were taken with lower ambient brightness, the same problem is noticeable.
I should get the diopter lenses in the morning, I will do the same test again with those.
In the archive an old shot of the returned OA3, one with the new one, and two frames taken from two 4K 4:3 30 fps movies, one in D-Cinelike and the other in standard color, all in PNG format.
From what is understood the new firmware applies more sharpness only in videos without D-Cinelike, in fact it is well noticeable that in the frame with standard colors the detail is more digitally pumped, but starting from a fuzzy one out-of-focus the result is not the top.
I still think that this action cam having Ultra Wide as its native resolution, i.e. without internal digital crop, has little detail for close subjects.
My old GoPro HERO 6 manages to work better because the lenses were designed for a considerably smaller field of view.
So imho it would have to be understood if the OA3 could do better with a different lens system, since it is not a sensor problem, or if this is the best we can get, only DJI knows.
Personally I would have preferred a Wide with more detail, but being in competition with GoPro 11 they had to go overboard.
Report already forwarded to DJI as well.

https://drive.google.com/file/d/ ... 4w/view?usp=sharing

It gave. But the +2 diopter would have to be a large front protective glass, not in the lens system, in order to maintain the maximum viewing angle..... I could be wrong.

I sent mine back few days ago and DJI will repair it for free (warranty). They made a quotation though (but with discount). Anyways, the item list says: Core Board Module and repair fee. Status says: "The image only shows parts with detailed cosmetic damage.."

We will see..

Urbsi Posted at 10-27 00:45
I sent mine back few days ago and DJI will repair it for free (warranty). They made a quotation though (but with discount). Anyways, the item list says: Core Board Module and repair fee. Status says: "The image only shows parts with detailed cosmetic damage.."

We will see..

That's what I had written yesterday, a friend also got a similar machining quotation (replacement of optical unit and labor, with stated costs but obviously under warranty), but someone here told me I was writing BS, a conspiracy.
But you know, flies disturb when they buzz around but they are absolutely harmless, just ignore them.
Perhaps that listing is for their internal use, but they will send you a new one, as to my friend.

MarcoR Posted at 10-27 00:21
In my opinion nothing has changed.
Even if the shots were taken with lower ambient brightness, the same problem is noticeable.
I should get the diopter lenses in the morning, I will do the same test again with those.

Thanks for spending the time to do these tests and share them here.  Sorry to hear that the exchange did not improve the experience with the camera.

I think that the issues with sharpness or detail people are noticing with the camera are mostly due to digitally zooming and (de)warping an image from a very wide angle lens. I notice this on the A2 when narrowing the FOV, or switching stabilization modes: Steady crops more than RockSteady on the A2, and RockSteady on the OA3 seems to crop more than on A2, judging by comparison videos online.

The lens (on the A2 and OA3) is focused in the distance, so you can get clearer image up close (for example with a close focusing adapter/filter), but if you are mostly filming at short distances, a camera with autofocus might be a better solution...

And now a test with +2 diopter

djiuser_zEKgdWRops1S Posted at 10-27 01:33
And now a test with +2 diopter

+2 it seems to return less detail up close, but from a distance it doesn't get worse compared to +4, in fact I can almost read the license plates of cars.
As soon as I get the lenses I will design and print in TPU an adapter to be able to place it fixed, easily removable.
Now I am printing this to make some videos with OA3 and GoPro HERO 6 in combo.

Now they told me to they´ll send a new one. Serial no: 5S4DK8Q...