Is it normal for 4:3 to have more vertical view than 16:9?  Shouldn't 16:9 (1920×1080) have the same vertical view as 4:3 (1440×1080) since they are both 1080?   Also, horizontally, 1920 should show a wider view than 1440 but the view horzizontally look the same.  Please see picture.

Hi, thank you for reaching out. As I review the images that you had provided, it shows that there is no problem with the 2 photo. The image that has 16:9 aspect ratio shows more horizontal than the 4:3 which is normal while the vertical side of this 2 image must be the same or 4:3 has more vertical view than 16:9. Hope this helps. Thank you for your support!

Thanks but I think you misunderstood.  If you look at the 16:9 picture, there is less floor and windows shown vertically.  Horizontally, they show the same view edge to edge, 16:9 is just stretched compared to 4:3.

This looks like a bug.   Can you inform your developers?

Thanks.

This is how the camera works. 16:9 is just a cropped 4:3 image. If you want the widest field of view, fly 4:3.

schl0r Posted at 3-12 10:11
This is how the camera works. 16:9 is just a cropped 4:3 image. If you want the widest field of view, fly 4:3.

Actually no.  that is an incorrect implementation.  1920x1080 has more horizontal view than 1440x1080 (since 1920 > 1440).  The vertical view should be the same (not cropped) since they are both 1080.  1920 should have more view than 1440.  The cropping should happen with 1440 view on the horizontal view.

Your pics look correct.

4:3 is the best view for FPV.  There is more picture at the top and bottom.  

16:9 has the bottom and top of the picture cropped off.

Just fly 4:3!

mc54 Posted at 3-12 13:52
Your pics look correct.

4:3 is the best view for FPV.  There is more picture at the top and bottom.  

But 4:3 is lower quality as it is only 1440x1080.  16:9 is recording at higher quality of 1920x1080 but with less fov.  Are you suggesting that I should record with less quality?  1440x1080 isn't even full HD.

Higher resolution on the same camera should have higher field of view.

In this case,16:9 and 4:3 should have same fov vertically.   Horizontally, 16:9 should have wider fov than 4:3.

DJI didn't implement this correctly.

fans8e299cb6 Posted at 3-12 13:56
But 4:3 is lower quality as it is only 1440x1080.  16:9 is recording at higher quality of 1920x1080 but with less fov.  Are you suggesting that I should record with less quality?  1440x1080 isn't even full HD.

Higher resolution on the same camera should have higher field of view.

Here is 3:4 view using SuperViou to get Superview compared to standard 16:9 1080p.

Bruxx3d Posted at 3-13 06:53
https://www.youtube.com/watch?v=xGhjy9lEdbk

Here is 3:4 view using SuperViou to get Superview compared to standard 16:9 1080p.

Superview is "stretching " 1440 to 1920 so there will be degradation compared to true 1920....

If the recordings are actually 1440x1080 vs 1920x1080 then this is the only part of the implementation they got wrong. It should be 1920x1440 vs 1920x1080, or any other ratios where the horizontal is equal but vertical is cropped.

The actual physical sensors on most cameras are 4:3, including DJIs in this case. The only way to get a different ratio is to crop or to stretch. DJI crops.

There is a replacement lens from RunCam which has an even wider field of view, I was considering swapping and then flying with 16:9 to get similar FOV vertically as 4:3 stock, and a little extra FOV horizontally. I haven't done the math on the FOV, I was planning on just try and see approach. Going so wide might look significantly distorted though, hard to say which I'd prefer.

If you want the widest FOV, fly 4:3. It's nice having some of the OSD against black background and when flying vertical FOV is nearly as important as horizontal, it make sense to fly with a ratio that reflects that.

fans8e299cb6 Posted at 3-13 07:05
Superview is "stretching " 1440 to 1920 so there will be degradation compared to true 1920....

My point of the video is to show there no real difference and at the end of the day, the 3:4 rendered into Superviou give overall an as good result if not better...

Bruxx3d Posted at 3-13 08:51
My point of the video is to show there no real difference and at the end of the day, the 3:4 rendered into Superviou give overall an as good result if not better...

Good results maybe but not better.   1440 stretched to 1920 cannot be better than true 1920 resolution.   1920 will give you more details.

shredzoo7 Posted at 3-13 12:55
[view_image]

Incorrect.  the 4:3 resolution is 1440x1080 not 1920x1440.

16:9 records at 1920x1080
4:3 records at 1440x1080

16:9 has higher resolution but crops top and bottom of the screen which is why I posted this thread.  It is an incorrect implementation by DJI.  Going from 4:3 to 16:9 recording should give you 480 more lines of video resolution on the width (1920-1440=480).  This means you should get 480 lines of more FOV (field of view) on the width.  The height should be the same since they are both 1080 yet the DJI FPV Air system crops top and bottom.

fans8e299cb6 Posted at 3-13 13:49
Incorrect.  the 4:3 resolution is 1440x1080 not 1920x1440.

16:9 records at 1920x1080

There may have been issues with providing the higher 4:3 resolution without introducing latency. I think it’s safe to assume that they did the best that could be done with today’s technology while still being able to sell it for a reasonable price, the proof being nobody else being able to come up with anything that compares. If you are this picky about the image, you should be running a GoPro or Osmo that would have an even higher resolution, more color profile choices and stabilization. That is what I am doing. It would be nice if the air unit had the same quality as a GoPro, but we will not be seeing anything like that until one of the next versions of the system.

fans8e299cb6 Posted at 3-13 09:35
Good results maybe but not better.   1440 stretched to 1920 cannot be better than true 1920 resolution.   1920 will give you more details.

Rendered both at 2k, rendered again per youtube limitation, final results is close, but superview is nicer meaning better if you ask me

Matter of opinions

fans8e299cb6 Posted at 3-13 13:49
Incorrect.  the 4:3 resolution is 1440x1080 not 1920x1440.

16:9 records at 1920x1080

It's probably just a manufacturer design/definition of resolution. All I know is that when I record with my GoPro in 4:3 I get more FOV in the end result video. So I always film in 4:3 and crop down to 16:9 in post production when I need to.

Just fly it and pick whatever works best for what you want your end result to be.

shredzoo7 Posted at 3-18 09:03
It's probably just a manufacturer design/definition of resolution. All I know is that when I record with my GoPro in 4:3 I get more FOV in the end result video. So I always film in 4:3 and crop down to 16:9 in post production when I need to.

Just fly it and pick whatever works best for what you want your end result to be.

These are good suggestions and I am aware of those options.  I just think it is weak that DJI advertises this as Full HD 1080P when in reality it is not.  Also, it is a big hassle to have to re-encode video that should already be 1920x1080P with proper FOV instead of having to re-encode 1440x1080P to 1920x1080P which will undoubtedly stretch the width which is undesirable.

BTW - Are there any free software to encode 4:3 to "superview"?

fans8e299cb6 Posted at 3-18 12:19
These are good suggestions and I am aware of those options.  I just think it is weak that DJI advertises this as Full HD 1080P when in reality it is not.  Also, it is a big hassle to have to re-encode video that should already be 1920x1080P with proper FOV instead of having to re-encode 1440x1080P to 1920x1080P which will undoubtedly stretch the width which is undesirable.

BTW - Are there any free software to encode 4:3 to "superview"?

yes, type SuperViou on google

I followed the discussion since I was also confused after seeing the quality of the recorded videos. It´s a misleading product description to talk about 1080p recording, when it is just an upscaled and cropped 1440x1080 signal.
At 1440x1080 it might not make any sense to squeeze the full sensor into 1080p. Default recording bitrate of 40Mb/s at 60fps seems to be ok in general, but the image quality is quite low, which means that the sensor might be the bottleneck.

It would be great to have more controls on the sensor like min-max iso, shutter, luts, sharpening... to be able to control and record videos and manually adapt to the required environment.
Lot of the artifacts look similar to small sensor action cams, when shutter is to high and not enough light is able to hit the sensor. Currently options are 120 or 60fps, It would be great to be able to get down to even 30fps for slow flights and 120 for racing. Giving users more option would be great. I haven´t fully tested the camera yet, but definitely do in the next days.

I just want to know if such features are in the pipeline, if you are also working on higher res camera module for the system and if we would be able to get more camera control in the future?
Otherwise I will just use the caddx without internal recording for fpv and a action cam for offline recording.

One thing that needs to be fixed is that I got several corrupt mp4 files. I don´t now why, since camera was rolling after liftoff and automatically stopped after landing.  
Also during one takeoff lower half of the screen was just green, so I had to land and re-plug battery and it was fixed.

jamsven Posted at 3-22 13:10
I followed the discussion since I was also confused after seeing the quality of the recorded videos. It´s a misleading product description to talk about 1080p recording, when it is just an upscaled and cropped 1440x1080 signal.
At 1440x1080 it might not make any sense to squeeze the full sensor into 1080p. Default recording bitrate of 40Mb/s at 60fps seems to be ok in general, but the image quality is quite low, which means that the sensor might be the bottleneck.

I agree.  I think it is misleading to advertise that it supports 1080P which most people assume is 1920x1080.  While 1440x1080 is indeed "1080", it is not true Full HD resolution of 1920x1080.  The way DJI implemented it gives worst details in 16:9 than 4:3 even when the 16:9 resolution is supposed to be the higher resolution (1920x1080 vs 1440x1080).  While I appreciate the view of just "accept it and fly 4:3", I think misleading info from vendors like this need to be pointed out.  We shouldn't tolerate misleading advertising from what is supposed to be a reputable company.

Looking at the specs of the sensor, which says 4M (i read as 4 megapixel) the actual sensor resolution is probably around 2316x1746. They should have been able to scale the output to something higher. It could be there's some other bitrate limitation they are hitting in the air unit and they decided going past 1440x1080 would cause some other problem, we can only guess.

I would just like to point out though, you keep saying that 16:9 should be higher FOV, when in reality it's a 4:3 sensor, so any other aspect ratio will loose FOV.


Where this product is sitting though among competing products, being able to record anything above NTSC with the same camera as the fpv feed is an achievement. Those who care about fidelity will carry a dedicated camera, everyone else is getting HD recording as a bonus if you will.

Now that 250g might become much more significant, and given DJIs experience with high quality imaging, I wouldn't be surprised if they tried to remove the need for a 2nd dedicated camera for those wanting to get cinematic footage on a gen 2 air unit.

Also, those carrying a dedicated camera run into the exact same thing, if you want max FOV you have to use the full 4:3 sensor. 16:9 sensors are just uncommon.


No where does DJI say "records at 1080p at the maximum FOV!" Someone with experience with sensors and FOV would just intuitively understand, max FOV is always at native ratio.

The sensor explanation makes sense, thanks.  If indeed the sensor is 4:3, cropping top and bottom would be appropriate for 16:9.  I just wish DJI would have advertised it supports "1440x1080p" instead of just stating it supports "1080p".  Most people assume 1080p is full HD (ie. 1920x1080p native)......

Oh I think I see what you guys are worried about, you think it's taking a 1440x1080 video, cropping it to 1440x810, and then scaling it back up to 1920x1080?

There's no reason to think that's the case. Since the physical sensor is quite a bit higher resolution than either output; it's scaling them all down, every time. It would just waste electricity to scale down and then back up. I think the 16:9 output isn't loosing anything besides FOV.

I suppose it's probably using same bitrate for both outputs and so the 16:9 suffers slightly since there's more pixels to share the bitrate. But what that also means is that it's a toss up on which output to use. Use 4:3 if you plan to superview-ize it, use 16:9 if you don't want the extra step, but the final file will have the same ending resolution and the same source bitrate. (Assuming it's equal bitrate, I haven't checked. If one of the outputs has a higher bitrate then you'd benefit from using that aspect ratio)

To clarify, what it's almost certainly doing is starting with a 2240x1680 video feed, and either scaling it straight to 1440x1080 (4:3), or cropping it to 2240x1260 and scaling it to 1920x1080(16:9)

TechnoCore Posted at 3-26 07:40
To clarify, what it's almost certainly doing is starting with a 2688x2016 video feed, and either scaling it straight to 1440x1080 (4:3), or cropping it to 2688x1512 and scaling it to 1920x1080(16:9)

Is 2688x2016 the resolution of the sensor?  If not, what is it?

I was editing my reply while you were submitting yours, 2688x2016 is some other less common resolution that's near 4MP.

It's a 4MP sensor, says DJI. I had trouble finding what resolution standard 4MP sensors are, but turns out they are usually 2240x1680.
source
source