Mavic 2 Pro 4K HQ vs 4K FOV

peros550 · Aug 25, 2018

Anybody could do a comparison between those two options ?
What do they mean from a video quality perspective. To me it seems like the HQ mode is a zoomed version. But I am ignorant, so if somebody knows for sure and could provide some information, it would be great.

Have a look at this video where is shows those options inside DJI GO 4 app.

many thanks in advance

jjchsg · Aug 25, 2018

4K video only contain 8 Mpixels. But the sensor is 20MPixels. Therefore, I believe in FOV mode, it re-sample the 20MPixels down to 8MPixels. In HQ mode, it possibly just take the middle 8Mpixels without performing any re-sampling. I am not 100% sure, but this is the logical way to have the two modes.

peros550 · Aug 25, 2018

Have a look at this link below. This guy has a Mavic 2 Zoom , and the moment he selects 1080P @60fps the field of view gets more narrow.

video link

I can't remember if original Mavic pro worked the same way. I believe it didn't.

tomibeg · Aug 25, 2018

peros550 said:
Anybody could do a comparison between those two options ?

4K HQ mode is some kind of "pixel to pixel" readout of cropped sensor area without downsampling from "5K" full sensor area (all other modes have it to fit proper resolution). This means that in this "cropped" mode you will get the best possible 4K image from the sensor. Not sure that this "quality boost" will be so much noticeable, because nowadays domnsampling algorithms do their job well in all modern cameras. But at least you may be sure, that this "cropped" mode will not spoil your 4K image quality. Also, for some info, only in HQ mode you can use 10 bit video.
BTW, In this 4K HQ mode FOV converts 77 to 55 degrees, which is not far away from M2 Zoom's 48 degrees after optical zoom. This means that you may be able to get a more cinematic perspective too. But you still can't do lossless "dolly-zoom" effect in-camera. Only in post, and only for FullHD output (just animate scaling for 4K footage).

Here is examples with downloadable footage for you:

Mavic 2 Pro 4K HQ: Mavic 2 Pro 4K Full FOV:

mreloc · Aug 25, 2018

Excellent. These colors look fabulous to me..

gnirtS · Aug 25, 2018

Some are claiming you *only* get the 10 bit log output data using the HQ (crop) option.

eukaryote · Aug 25, 2018

gnirtS said:
Some are claiming you *only* get the 10 bit log output data using the HQ (crop) option.

Which would make sense. Binning a 20mp output to 8mp takes a lot of processing power for real time. Add onto that having to output 10 bit, which is again resource intensive, and then again into a H. 265 format which is resource intensive, it would require a lot of processing power, which DJI probably realised ate a lot of battery power. The Inspire 2 can do this easily because it has 2 large 6S batteries powering it. A relatively small 4S battery - not so successful.

GH5 can do the above as well, but it heats up a great deal to achieve this, and the battery is rapidly drained.

gnirtS · Aug 25, 2018

Pretty much my theory. We're at the limit of onboard processor and power systems again. Now h265 is SOC it leaves some scope for 10 bit log but not for every possible option.

Ultimately trying to cram all this into a tiny form factor that (i) has to be light enough to fly (ii) has to be able to cool itself and (iii) run off a single 4s battery is hard. REALLY hard.

I'm not sure people appreciate just how hard it is with the mavic form factor compared to phantom or inspire when they say "just add this..".

On the plus side, the new sensor and processing chips SHOULD provide far better output even from 8 bit data. (Although i still miss the 14 bit out of my DSLR).

eukaryote · Aug 25, 2018

gnirtS said:
On the plus side, the new sensor and processing chips SHOULD provide far better output even from 8 bit data. (Although i still miss the 14 bit out of my DSLR).

It's got a sony sensor. Shadow recovery is going to be crazy again.

I'll set it like with all my Nikon gear - Underexpose by 0.3-0.7 EV, and then recover the shadows. Highlights clipped less, i'm pretty sure we can drag out near 14 stops of effective dynamic range once we're done with the raw image

gnirtS · Aug 25, 2018

At base ISO, we have a solid 11.5-12 stops of usable DR; a bit more if you are careful with exposure and post processing.

Drone diaries: the 2018 DJI Mavic 2 Pro review

The above taken off the blog of one of the guys employed by Hasselblad who actually helped develop the camera. For a relatively small 1" sensor that isn't bad.
There maybe enough DR here to actually use ETTR and other concepts.

monk191 · Aug 26, 2018

I am actually noticing quite a difference between the FOV and HQ modes, beyond color. The FOV mode is much more artificially sharpened to make up for loss of detail. You can see it in the grass in the beginning and in the hills after passing the trees.
When looking at all the youtube reviews, some of them look very nice and detailed, and others look a bit artificial, closer to the original Mavic Pro. (Just speaking on detail here). I'm guessing that it is the difference between these two modes. I am watching on a 4K monitor.

Hopefully the downsampling algorithm is something they can improve with firmware. Otherwise, I will probably stick with HQ in most situations.

Roamer105 · Aug 26, 2018

On closer examination of the original footage (grabbing stills from the same vantage point to compare) it is quite clear that there is a significantly higher amount of clarity and precise detail when using the HQ (cropped) mode in comparison with the 5K down-converted mode. While this on the surface seems counter intuitive, I think that the down-conversion is pretty simple and is throwing away a lot of information that, in combination with the wider perspective ("zoomed out" so to speak) makes the image appear a lot softer and a bit flatter. I also noticed that there is a little bit more video noise on the full frame version. However, there is very little noise to start with on either types of recording - so no big deal.

I don't have my M2P yet (wait till Wednesday) to do my own testing, I suppose I will do much of my recordings using the HQ mode (but all set to manual especially exposure) using the Dlog mode with either neutral or light boost of sharpness and contrast to avoid having the lazy compression algorithm shave off too many fine details.

MauiMaps · Aug 26, 2018

The zoom has 2x manual zoom, but what is the % closer of the Pro from Full 77° to HQ 55°? Is there a calculation?

Zoom
FOV: about 83° (24 mm); about 48° (48 mm）
35 mm Format Equivalent: 24-48 mm
Aperture: f/2.8 (24 mm)–f/3.8 (48 mm)

Pro
FOV: about 77°
35 mm Format Equivalent: 28 mm
Aperture: f/2.8–f/11

tomibeg · Aug 26, 2018

MauiMaps said:
what is the % closer of the Pro from Full 77° to HQ 55°? Is there a calculation?

Based on many different footage comparisons, crop factor is ~x1.4 from the full FOV. So the calculation should be 28mm*1.4 = 39.2mm in HQ mode.

Very similar crop factor can be confirmed mathematically by simple dividing 5472 pixels (full width of the sensor) by 3840 (4K area width), which is x1.425 crop or 39.9mm
So the approximate FOV 35mm equivalent in HQ mode should be almost 40mm. Not so bad for "non-Zoom" version with much better sensor, right?

gnirtS · Aug 26, 2018

It'll be interesting when people actually get to test these units to see a HQ and normal 4k grading ability comparison to see if there are any significant differences in quality, dynamic range and so on.

bluntnose · Aug 26, 2018

Reading this thread tells me I’ve got a looong way to go when the M2P arrives. I can barely eke out some slight, foggy meaning from most posts on the thread,and I know if I’m going to get the videos I want from the little beasty I need to understand the photography aspect waaay better. I know it’s silly of me to post how ignorant I am, I guess what I’m trying to say is you guys are awesome. Thanks for the expert advice.

milsonian · Aug 31, 2018

I just noticed in full FOV there is pretty significant barrel distortion on the horizon. At least I'm presuming it's only in full FOV...need to test in HQ.

Example:

chunky · Sep 1, 2018

I already made a comparison between HQ and Full FOV mode and the result is shocking in my eyes since my 2 year old Phantom 4 Pro with the same/very similar sensor performs so much better.
HQ mode is taken from a 4K crop of the 5.7K sensor and read pixel by pixel. Due to debayering the quality is „less“ than real 4K but that‘s usual here.
The FullFOV mode takes the full sensor area (ok, only 16:9 instead of 3:2), but in contrast to the manual on page 38 it is not read from real 5.7K and is downsampled later, it is read pixel binned or interleaved! In the end we only get a resolution of about 2.7K :-(!
That is not what I have been offered and have bought, especially not where my 2 year old Phantom 4 Pro was able to read the complete sensor and made a very high quality downsampling from the „oversampled“ data where negative debayering effects were made „smaller“ due to the downscale!
DJI should really sort this out, FullFOV quality is really bad and could be much better as shown in the P4P two years ago!

Feel free to contact DJI support to sort out this issue.

Please be aware that I accidentally used the wrong scale for taking the shots in regards of ISO. One single chart „should“ be 5,625% of the picturewidth or usual 2,81% for 4K, I only took 2,5% so it‘s not comparable to other measurements.
But, in my test all tests were done at 2,5% so they are compareable among each other.
Since 2,5% is shown in the charts nothing is „wrong“ here.

Elmateo487 · Sep 1, 2018

Yeah, the M2P image quality is REALLY bothering me. I keep seeing example video after video after video, where everything is just way too soft. Maybe no one is doing +1 to sharpening... I dunno. But the P4P seems to be destroying it in image quality. Mostly sharpness, and detail. Colors can mostly be fixed post process.

I think ill wait until they discontinue their P4P line, and replace it with the M3P