Mavic Pro-2 vs Zoom and P4P - Line Skipping + Sensor Heat?

[Vegasrobbi]

Yea if we take the Mavic Pro to Mavic Pro Platinum to Mavic 2 as a template, I doubt the P5P will be out soon. Add to that the real lack of any significant competition apart from maybe the Yuneec Typhoon H Plus that costs about $1900 and hardly anyone has even talked about. I was honestly surprised to see it at Best Buy the other day on the shelf.

Too bad guys don't recognize how good the Typhoons are; and Anfai is an excellent value compared to the M series.

 

[kilomikebravo]

="sar104, post: 566865, member: 22742"[Unremarkable. The 4k FOV is noticeably better, both in terms of sharpness and resolution.

I concur, even on a mobile the difference is discernible. Thanks SAR.

 

[mcsenerd]

Maybe I’ve just crazy eyes, but I just don’t see this noticeably “better”. To me, the 4K FOV is barely better than the resolution power shown in the 2.7k snip, and to me the MP1 is clearly able to more clearly resolve than the FOV on the M2P regardless of clearly additional sharpening. But, I’m just a chump...maybe it’s just me and my eyes.

 

[igor88]

Here are some comparisons that have been discussed, all at ISO 100 and f2.8.

(1) M1P vs M2P DNG files, with the M1P upsampled to 20 MP:

View attachment 49215
M1P DNG upsampled
View attachment 49216
M2P DNG

That shows exactly what you would expect - the 20 MP sensor clearly does better than the 12 MP sensor.

(2) M2P HQ:

View attachment 49217

Similar resolution to the M2P DNG - not surprising since it is reading out the same pixels in this cropped video mode.

(3) M1P 4k vs M2P 4k FOV

View attachment 49220
M1P 4k

View attachment 49219
M2P 4k FOV

Interesting result. The M1P 4k looks like it has much more sharpening applying, even at a -1 setting. It also has a lot more artifacts. Absolute line resolution appears to be similar, but the M2P resolution fades better to unresolved than the M1P.

(4) M2P 2.7k vs M2P 4k FOV

View attachment 49221
M2P 2.7k upsampled to 4k

View attachment 49222
M2P 4k FOV

Unremarkable. The 4k FOV is noticeably better, both in terms of sharpness and resolution.

pity that you did not compare hq vs mav1

for m2 2.7 - it must be sayed: the difference is obvious. but bitrate on 2.7 is 1/2 of 4k bitrate (50Mb vs 100Mb - so chosen by dji). it would be interesting to see 2.7 in 100Mb bitrate and then do comparison.

 

[sar104]

Maybe I’ve just crazy eyes, but I just don’t see this noticeably “better”. To me, the 4K FOV is barely better than the resolution power shown in the 2.7k snip, and to me the MP1 is clearly able to more clearly resolve than the FOV on the M2P regardless of clearly additional sharpening. But, I’m just a chump...maybe it’s just me and my eyes.

I think possibly you are just not familiar with resolution charts. The 2.7k resolution is about what you would expect in this context:



The 4k FOV is better:



But it's not as good as the full sensor DNG downsampled to 4k:



You have to pay attention to the numbers - just because you can see lines doesn't mean it was properly resolved. For example, compare the angled lines labelled 5 - they superficially look as if they were resolved in the 2.7k and 4k FOV crops, but then you notice that they are coarser than the lines labelled 3. They are resolved in the downsampled DNG, albeit with artifacts.

The full DNG resolution shows how they should look - it doesn't show serious artifacts until level 7.

 

[sar104]

pity that you did not compare hq vs mav1

for m2 2.7 - it must be sayed: the difference is obvious. but bitrate on 2.7 is 1/2 of 4k bitrate (50Mb vs 100Mb - so chosen by dji). it would be interesting to see 2.7 in 100Mb bitrate and then do comparison.

M2P HQ and Mavic 1 4k are both in the post above. Or did you want to see something different?

 

[igor88]

ups. i see it now

 

[Sockeye]

Don't get yourself fooled by the diagonal line test. The screens above are to small to see the dirty trick. Look at my comparison below. At a first glance the M2P FOW kicks-**** all others.... but then I counted the lines.... :-(

 

[sar104]

Don't get yourself fooled by the diagonal line test. The screens above are to small to see the dirty trick. Look at my comparison below. At a first glance the M2P FOW kicks-a** all others.... but then I counted the lines.... :-(View attachment 49241

That shows exactly the same results, except that you didn't include wide enough line pairs to assess either the 2.7k or 4k FOV resolution. Were you unable to view the images any larger, or read what I wrote ? The M2P FOV resolution is better than the 2.7k resolution, but it doesn't kick anything else.

 

[Sockeye]

The M2P FOV resolution is better than the 2.7k resolution, but it doesn't kick anything else.

Sorry, this is something I cannot see. Help me out, in which of the line resolution of your images does the 4K FOV show more details then the 2.7K?

 

[sar104]

Sorry, this is something I cannot see. Help me out, in which of the line resolution of your images does the 4K FOV show more details then the 2.7K?

Quantitatively, 3 and 4. The 4k FOV resolves level 3 without artifacts and almost resolves level 4. The 2.7k has artifacts at level 3 and completely misses 4. If you are unfamiliar with line pair degradation, look at the numbers themselves - the difference is obvious there.

 

[UAVNV]

this video by op went from 34k views to 21k a couple days later, checked on friday was 14k, cant find it in search here

 

[igor88]

Jittering: yea I'm pretty sure that I had it at the correct shutter speed, so I don't know. YouTube compression maybe?

Aliasing: yea that's a thing but it was worse on my MP.

Speed: I wrestled with that one but I flew these orbits by hand and they were too slow. So I sucked it up and speed up the footage. I considered cropping the cars out but the angle was wrong for that. It's something I'll work on in the future.

MLog: the whole point for me is to take on something that I haven't done before. I could have kept flying my 3DR Solo, my Karma or my MP and kept doing what I was doing. Heck, the GoPro app all but edits for you! If I were to record in H264 normal, I may as well have kept using the MP. Apparently, according to you, that's what I should have done. Well, what you should at least suspect, it's that I'm not the type to give up so...

My mental state: I'm no crazier or violent than I was before so I don't think editing has had an effect on me. I'll take it up at the VA next time I'm in for PTSD counseling and see what they think but I'm pretty sure it's a non issue. I've always used "hahaha" rather than lol as a way of being different. I wasn't aware of any correlation between it's use and insanity but again, I'll make inquiries.

Low res: because I used the 180 rule and things are always moving, I would expect constant blurring of the image. Isn't that correct? I used to use high shutter speeds because I mainly extracted pictures from video in the past. That's why my photo account is massive, while my YouTube account isn't. Everyone was always, "you should post videos" because I fly almost every day as my job runs me all over the place and I see things that I find interesting all the time. So I posted videos and everyone hated how un cinematic they were. This video was shot in HQ BTW, not that it matters. I'm not afraid of hard work and I'll keep at it until I'm happy with the results.

Music: yea that was easy but thanks for the begrudging compliment.

I know you're not really wanting to "help" me but I appreciate it anyway because it does help re enforce a lot of what I already think. So thank you.

What I was delicately trying to suggest you was - not to use Mlog. It is much to much unsharp (unclear). Also, you'll not get any more detail from it. Not to mention the noise. Just be patient and wait for next FW.
In the meantime you can do colorgrade from vid. in "normal" color, using all HQ benefits (except 10 bit).
It was meant to be helpful.

p.s.
I used this clip just to show how much aliasing you can expect in hq

 

[Sockeye]

Quantitatively, 3 and 4. The 4k FOV resolves level 3 without artifacts and almost resolves level 4. The 2.7k has artifacts at level 3 and completely misses 4. If you are unfamiliar with line pair degradation, look at the numbers themselves - the difference is obvious there.

Not too sure if the difference is significant. I took your pics and put them together with the original. In my point of view they both make it up to 3.2-3.3 and then the vodoo begins. But regarding the distance you filmed the chart, the next frame could give an opposite result...



Especially the Binning/skipping with the micro movement of the gimbal result in a strong variation comparing the frames.

But as Igor said, we can pixel-peep a day long, we have no other option than to wait for a new FW. But with the noise made, I presume the chances increase that something will change.

 

[sar104]

Not to sure if the difference is significant. I took your pics and put them together with the original. In my point of view they both make it up to 3.2-3.3 and then the vodoo beginns. But regarding the distance you filmed the chart, the next frame could give an opposite result...

View attachment 49251

Voodoo? I was assuming that you understood the underlying principle of aliasing. I don't see how you can view those two results as equivalent, but it's not my place to convince you otherwise.

In my opinion the vertical lines are not particularly useful, except for taking line profiles and then analyzing numerically, because the aliasing artifacts are too difficult to distinguish from the pattern itself.

 

[Sockeye]

Voodoo? I was assuming that you understood the underlying principle of aliasing. I don't see how you can view those two results as equivalent, but it's not my place to convince you otherwise.

In my opinion the vertical lines are not particularly useful, except for taking line profiles and then analyzing numerically, because the aliasing artifacts are too difficult to distinguish from the pattern itself.

Sorry for my sloppy wording. I would love to continue this conversation in German and I promise, I would not be rude between the lines.

But the "Voodoo" was pointing on the reason for this aliasing. Line skipping, pixel binning or this magical "raw subsampling" and "surprisingly" kicking in in about the same pixel density in 4K FOV and 2.7K... whereas 1080p has a significant distance to 2.7K

 

[WithTheBirds]

I think possibly you are just not familiar with resolution charts. The 2.7k resolution is about what you would expect in this context:

View attachment 49237

The 4k FOV is better:

View attachment 49238

But it's not as good as the full sensor DNG downsampled to 4k:

View attachment 49239

You have to pay attention to the numbers - just because you can see lines doesn't mean it was properly resolved. For example, compare the angled lines labelled 5 - they superficially look as if they were resolved in the 2.7k and 4k FOV crops, but then you notice that they are coarser than the lines labelled 3. They are resolved in the downsampled DNG, albeit with artifacts.

The full DNG resolution shows how they should look - it doesn't show serious artifacts until level 7.

View attachment 49240

Thank you for your efforts here SAR- no doubt many will find the results very informative.

It would seem that we can now safely almost completely discount certain of the previous assumptions, particularly the suggestion that the sensor performance was crippled by heat issues or that the 4K FOV output was simply an upsampled 2.7K sensor read.

It is doubtful DJI has tried to pull as swifty on us.

If in fact DJI employed the Sony IMX183CQJ sensor in the M2P a quick perusal of the published specifications reveals that the only readout mode that supports all pixel scan in 10bit depth above 25fps is "type 1/1.4" which provides 9.03mpx max at 17:9 aspect ratio (reading 71% of the sensor horizontally). All other readout modes above 25fps employ a combination of horizontal and/or vertical line addition.

FOV mode must be subject to the sensor limitations, that is line addition is, and must be, employed prior to further SOC processing.

DJI has perhaps shot themselves in the foot by not providing further explanation as the the technical issues with the camera and the design decisions. HQ mode looks like a genuine attempt to give us something extra. Unfortunately they are being criticised by those of us who might not appreciate they have done the best they might have done with FOV mode given the Sensor limitations.

 

[sar104]

Sorry for my sloppy wording. I would love to continue this conversation in German and I promise, I would not be rude between the lines.

But the "Voodoo" was pointing on the reason for this aliasing. Line skipping, pixel binning or this magical "raw subsampling" and "surprisingly" kicking in in about the same pixel density in 4K FOV and 2.7K... whereas 1080p has a significant distance to 2.7K

Apologies - I should have considered the possibility of a language barrier. Your English is better than my German though. Okay - we can check the difference that would be expected for 4k vs 2.7k upsampled back to 4k.


High-resolution view of the crop


4k bicubic downsample


Downsampled to 2.7k then upsampled to 4k

The second and third images show the theoretical maximum resolutions available at 4k with this image. Notice that the artifacts on the 2.7k only show up one level earlier than on the actual 4k, and that the last recognizable level is also just one level different.

 

[sar104]

Thank you for your efforts here SAR- no doubt many will find the results very informative.

It would seem that we can now safely almost completely discount certain of the previous assumptions, particularly the suggestion that the sensor performance was crippled by heat issues or that the 4K FOV output was simply an upsampled 2.7K sensor read.

It is doubtful DJI has tried to pull as swifty on us.

If in fact DJI employed the Sony IMX183CQJ sensor in the M2P a quick perusal of the published specifications reveals that the only readout mode that supports all pixel scan in 10bit depth above 25fps is "type 1/1.4" which provides 9.03mpx max at 17:9 aspect ratio (reading 71% of the sensor horizontally). All other readout modes above 25fps employ a combination of horizontal and/or vertical line addition.

FOV mode must be subject to the sensor limitations, that is line addition is, and must be, employed prior to further SOC processing.

DJI has perhaps shot themselves in the foot by not providing further explanation as the the technical issues with the camera and the design decisions. HQ mode looks like a genuine attempt to give us something extra. Unfortunately they are being criticised by those of us who might not appreciate they have done the best they might have done with FOV mode given the Sensor limitations.

My conclusions so far:

4k FOV is not just upsampled 2.7k, but that would be a ridiculously poor strategy - I can't believe that anyone really thought that.​

​

4k FOV is worse than the M2P 20 MP images downsampled to 4k, but that was expected - we just don't know how they are being generated.​

​

4k FOV is not too dissimilar to 4k on the Mavic Pro, but since both have more than enough resolution for 4k, that's not particularly surprising. The 10 bit output from the M2P is superior by other metrics.​

​

4k HQ is almost exactly as expected in compared to the full 20 MP images.​

​

None of the results quite lives up to the theoretical maximum shown above, but given the additional factors of real lens effects, low-pass filters etc., it's credibly close.

 

[Lon Denard]

Too bad guys don't recognize how good the Typhoons are; and Anfai is an excellent value compared to the M series.

That Typhoon looks cool but the size and set up turned me off. I don't know much about the Anafi, Heck, I've hardly heard a word about it since it launched.