Mavic Pro 2 - image sometimes soft ???

[OldBlueHeron]

So far this camera is somewhat disappointing. I hope there is some tweaking done with firmware. The only downside between all the missing features and the camera issues, is it may be forever till we see a new firmware released.

 

[ddublu]

I also felt it was necessary to lower the contrast on the M2P.

 

[Black Diesel]

I also felt it was necessary to lower the contrast on the M2P.

What settings did you find best for sharpness/contrast/saturation?

 

[boldblue737]

the "normal" mode i find just ever so slightly less sharp than my M1P and I consider the video from the M2P to be a MUCH better look. you can easily boost the sharpness in DJI GO 4 and get a M1P look if you want. many videos on youtube by the M2P are filmed in D-LOG which is the flat 10 bit profile with NO sharpening added so it is MUCH softer than the M1P, and that is by design. As for the idea of the D-LOG setting, everything from color, to sharpening, to distortion fixes are to be added in post processing and grading. obviously some of these youtube creators are *choosing* to add minimal sharpening and so you get softer images than you have seen in the past, but that is by their choice.

 

[gnirtS]

Mavic 1 was only usable with sharpening set to +1 to disable noise reduction.
Sadly this resulted in a really badly oversharpend image by default and often required some post processing blur to reduce it.

 

[Black Diesel]

Mavic 1 was only usable with sharpening set to +1 to disable noise reduction.
Sadly this resulted in a really badly oversharpend image by default and often required some post processing blur to reduce it.

Is M2P okay to use with sharpness set to +1?

 

[Sockeye]

Sharpness and resolution are two pairs of shoes. You will not be able to recover the lost information in post, you lost by binning/line skipping. They are doing the same thing as with the first batch of M1 with the pixel binning for 1080p. Hopefully this can be fixed via firmware upgrade and does not need a hardware revision like the M1.

 

[Black Diesel]

Sharpness and resolution are two pairs of shoes. You will not be able to recover the lost information in post, you lost by binning/line skipping. They are doing the same thing as with the first batch of M1 with the pixel binning for 1080p. Hopefully this can be fixed via firmware upgrade and does not need a hardware revision like the M1.

Does this apply to the crop mode (HQ) as well? From what I've read, it seems the best way to optimize picture quality is to shoot video in crop mode. And since I don't have time for testing before my trip to the islands this week, I'm wondering if sharpness +1 is okay to use.

 

[Sockeye]

No, the crop (HQ) just reads the 4k pixel area out of the sensor and has from what I have seen by now about the same quality as the Zoom.
But what did I buy the Pro for? Yes, the advantage of the bigger sensor, which should result in a better low light performance and picture quality. This is something we are currently not getting.
The Phantom 4 Pro can do it since 2 years now. So why this fail on the M2P?

 

[gnirtS]

Is M2P okay to use with sharpness set to +1?

No idea, i use 0,0,0 in DLOG. Id much rather do all that in post.

 

[KiterTodd]

So, didn't hear an answer on if anyone tried a +1 on sharpening on the M2P or not. Does the +1 change meet expectations?

There's a nice comparison video of the M2P and M2Z and, as mentioned, the M2Z has a sharpness in line with what we are used to on the MP. The reviewer closed by saying, if you want it more cinematic you are likely fine with the M2P as-is, but other people could just increase the sharpness.

I haven't tried any of these settings yet, but that hardware talk above has me worried.

 

[gnirtS]

the M2Z has a sharpness in line with what we are used to on the MP.

You'd expect that from the same sensor.

If you go to "None" style its significantly more sharpened even on 0 than DLOG is. That's probably the way to go if you want to avoid post processing.

 

[KiterTodd]

You'd expect that from the same sensor.

If you go to "None" style its significantly more sharpened even on 0 than DLOG is. That's probably the way to go if you want to avoid post processing.

Well, the implication was that it is not a sensor issue (sharpness) but with how DJI processes the image in the Mavic2 vs in the Mavic1.

Personally, I like it sharper by default. I can always reduce the sharpening if I want. I'll try out the settings you noted and see what happens. Thanks.

 

[shotes]

Check out my videos M2P style +2,+1,+1 Dlog-M on my YouTube channel: Roamer105
https://www.youtube.com/channel/UCRdvVHNqe4dRZIDhLCKOMCA

Nice. Did you try for a specific aperture setting?

 

[Roamer105]

So, didn't hear an answer on if anyone tried a +1 on sharpening on the M2P or not. Does the +1 change meet expectations?

There's a nice comparison video of the M2P and M2Z and, as mentioned, the M2Z has a sharpness in line with what we are used to on the MP. The reviewer closed by saying, if you want it more cinematic you are likely fine with the M2P as-is, but other people could just increase the sharpness.

I haven't tried any of these settings yet, but that hardware talk above has me worried.

I think you will be happy using +1,+1,0 combined with HQ DlogM. Then post process it applying the free Ground Control GC_DlogM_Rec709 LUT and end up just with a little fine tuning and barrel distortion correction. The results are stunning!

 

[Roamer105]

Why don't you test it with a comparable template? Just use the 2nd last page of the linked PDF, print it out on a 1200 dpi Laser or display the page on an 4K Monitor and focus on it. Record with the M2P in HQ, FOV and wide angle/zoom on the M2Z.

I am looking forward of seeing your results. I saw the comparison between the P4P and the M2P. The P4P is 2 years older but resolution is far better. (Erstflug, Video, Erfahrungen - Seite 2 - DJI Mavic 2)

PDF: https://vcc.zih.tu-dresden.de/files/vc/tests/testbilder/anleitung.pdf

Thank you for the link to the test pictures. I put the PDF through Photoshop to extract the last two full resolution images to png files to view on my 65" 4K LG OLED TV, then photograph them with the Mavic 2 Pro as well as my $4K Canon DSLR gear with a 16-35mm professional quality lens. After taking the pictures, I compare them on a 4K monitor. To my surprise, the M2P camera still raw compared to the Canon raw is a tossup. In fact, the M2P image was a little bit crisper than the Canon! I then took full resolution video as well as HQ UHD in NORMAL mode and standard settings using H265. Took the clips into Resolve and extracted stills from both which I then added to the comparison in Photoshop. Even more amazed the resolution on the HQ UHD still was directly comparable to the RAW still (nearly indistinguishable) and the FOV version while slightly softer still showed nearly exactly the SAME LEVEL of detail resolution than the HQ version and in degree of softness directly compared (almost indistinguishable) with the Canon version.

No more arguing about "softness" and "lack of resolution" etc... The image was so clear and detailed it actually showed the 4K OLED pixels - which btw did not resolve on the Canon image. I think the "softness" is more a subjective reading resulting from the highly detailed 10bit color/luminosity graduation that shows what would appear as a hard corner on a MP(1) as a more textured gradual change edge that is much closer to reality and very comparable to the luminosity resolution of the Canon with the same subject. Taking comparisons with the color test image shows the BEST and most accurate representation when recording UHD FOV Normal/Standard mode. The colors on that image beat the Canon by a leap and was also better than its HQ equivalent for some reason. Maybe the Hasselblad "Hasselblad's Natural Colour Solution (HNCS)" is only active in FOV? - more likely however is that the color resolution improves through down scaling the 5K image to UHD.

 

[Sockeye]

I did the same test on my 4K Monitor and had about the same results as you, with FOV being just slightly behind the HQ mode.
But after discussing the results in a German forum, I had to learn, that the Template was originally produced to judge Full HD cameras. This means for our 4K test, the resolution of the display where the template is shown needs to be at least 8k or alternatively move the drone further away.. (and use exactly the double distance).

I will do this tonight and post my findings....

 

[Roamer105]

Here goes some test examples at F2.8 UHD, 24fps Color:NORMAL, Style: STANDARD for both video frame grab samples. I supplied the actual full resolution test image for comparison. Take a look at the concentric circles that in the center represent 4K. All three of the samples (RAW still taken 16:9 and HQ UHD resolve very cleanly and the FULL FOV just a hair less). This camera has amazing optics!

Unfortunately I did have to convert the sample images to jpg due to size for this forum. The conversion did degrade the images a little (note some jpg artifacts enhancing otherwise nearly invisible mpg blemishes). The RAW converted to jpg is pretty clean. What it does show however is that even at Style=NORMAL, there is a bit of sharpening added to the video as it does not show like that in the RAW image.

Note the left/right edge areas are a bit yellowish/darker - well, that is unfortunately my not-so-great-4K monitor's fault. I should have used my LG OLED 4K TV as with my prior tests, but ignoring that fact, the results are amazing!

 

[Cliff_622]

Something doesn't seen right. Is your chart measuring the Mavic Pro-II down 4000+ lines? That's not possible. Even Nikon, Canon and Sony full frame cameras cant even do that. Something doesn't seem right....thats way beyond the Nyquist frequency of 4K. (maximum frequency that 4k can mathematically hold)

This chart must have a different resolution metric.

 

[Roamer105]

Something doesn't seen right. Is your chart measuring the Mavic Pro-II down 4000+ lines? That's not possible. Even Nikon, Canon and Sony full frame cameras cant even do that. Something doesn't seem right....thats way beyond the Nyquist frequency of 4K. (maximum frequency that 4k can mathematically hold)

This chart must have a different resolution metric.

The full chart Is right here (jpg from original vector chart properly sized to 11 by 17 inches at 300 dpi. I printed it out to these exact measurements using my Canon Pro 10 onto 13 by 19 Pro Luster photo paper. See picture taken with wall context from my Canon (attached along with a resolution test picture taken with the Canon 7D Mark-II with a 20-700 1:4L IS lens at proper distance to fill image). Then I re-took the same image same light condition with the M2P at HQ Normal color and style = none (standard). That too is attached. Unfortunately I can not take a comparable video in FULL FOV mode or still with this size test chart because it takes the lens to close (not able to focus) in order to get the full picture.

I many ways, these pictures are better than the ones I took off the monitor (which you are correct was improperly scaled for this test). The printed test image is very clean and precise with no pixel aliasing of course.