Is there actually an answer to the camera settings question?

[rmb]

Uh YouTube analytics across the Board shows it higher than 50% lately. No assuming. That would make an *** out of you and me.

Ok I definitely want to be careful here as I am loving all this input. In fact I will go so far as to say that of all of the forums and threads that have talked on this subject, this is rapidly coming the most controlled and thoughtful, so I want to keep this healthy and friendly discourse up and not take sides. That said regarding the youtube/no youtube viewing habits that both SeanOzz and Ziggy01 were talking about, I have to agree a bit more with SeanOzz. Not that millions of educated viewers are not watching and critically analyzing this footage on large quality TVs, but when most people "post" a video example, that youtube or Vimeo "post" is generally accessed by clicking on it and going directly to the sample. In fact less than 8% of the public (maybe higher in our circles) even have the ability to watch youtube and or Vimeo on their home HD set. Further, otherwise nice samples are frequently contaminated by poor compression and upload habits. Ok, enough of that. Now where Ziggy01 is correct is when he is speaking about the likes of me and perhaps a lot of us. I am a professional Aerial Cinematographer that has had the opportunity to work on very pro projects where my footage will be used against REDs and ARRIs. I am also a professional Editor and Color Grader and therefor watch everything on very high-end equipment. I mention this not to give my resume' but to let you know that I know that to start comparing anything shot on phantom 1 through 3, GoPro and of course our little friend Mavic with these high-end rigs is a fools errand. That said there is a such thing as best practices and on the technical side the best the camera can do, ideal shooting conditions and post production techniques to help cover up inevitable short comings. It is the search for those answers I am hoping this group can help bring to our small corner of the industry and ultimately why I started this thread. I think the test/testing conditions that I would challenge the group to do and present (and I will of course do it myself as well) is to pick a day with consistent lighting conditions (could even be indoors) point the camera at anything that has some light area and dark (usually a close tree or house-lined horizon), make sure to have some level of critical detail in the shot. Could be tree branches or bark, could be gravel. Put the bird in hover and shoot 5 seconds of the exact same scene. Now not everyone is going to have ND so either shutter speed or ISO won't be exact and I think that is important. My opinion is that ISO is most important when assessing noise so let's standardize on that. Here's what I propose;
I say lets pick a 200 ISO at what ever shutter gives you a good looking overall exposure/histogram (a nice hump in the center) and then shoot the following 6 clips. Regarding FPS, I don't think it matters if it is 24fps or 30fps.

Shoot 5 or 6 seconds of:
1) 3840x2160 D-Cinelike @ 0,0,0
2) 3840x2160 D-Cinelike @ -1,0,0
3) 3840x2160 D-Cinelike @ +1,0,0

4) 2720x1530 D-Cinelike @ 0,0,0
5) 2720x1530 D-Cinelike @ -1,0,0
6) 2720x1530 D-Cinelike @ +1,0,0

We can explore Dlog too, but since not everyone has access to grading software and Luts, that should come down the road.

Lastly I am NOT the BOSS here. If someone one believes that there would be a better (More telling) test procedure I am totally open.

Thoughts?

rb

 

[rmb]

I might be able to help a little.


Here's where I really break away from the pack.
I don't use dlog or d-cinelike either. (For shame on me!) I prefer true color (as long as I've also bumped my contrast to +1 and my saturation to +2.

Thanks SeanOzz. I have not had much experience with "true color" and will certainly try it on your recommendations. Regard the video sample. That guy really went to a lot of effort. Unfortunately everything he shot looked terrible. Either thats the way it looked or he had some very big compression issues. I watch it on a high-res monitor at 1080 and it all was so over compressed that I didn't learn as much as I know he was trying to convey. Another hazard of watching on line.

 

[SeanOzz]

Thanks SeanOzz. I have not had much experience with "true color" and will certainly try it on your recommendations. Regard the video sample. That guy really went to a lot of effort. Unfortunately everything he shot looked terrible. Either thats the way it looked or he had some very big compression issues. I watch it on a high-res monitor at 1080 and it all was so over compressed that I didn't learn as much as I know he was trying to convey. Another hazard of watching on line.

Yeah. It was an afterthought originally intended for just myself to see the differences side by side. Flying in sports mode wasn't helpful in this scenario.
One thing I'm enjoying about this group is the people here seem to actually get along and not fly off the handle to differing opinions.

The filters honestly boil down to personal preference.


Here's something that I do now that I didn't at first.

I shoot in 1080. Lol
I save 99% of my video and 4K hogs up a lot of external hard drive space among other reasons.

 

[ecoleman]

Ok I definitely want to be careful here as I am loving all this input. In fact I will go so far as to say that of all of the forums and threads that have talked on this subject, this is rapidly coming the most controlled and thoughtful, so I want to keep this healthy and friendly discourse up and not take sides. That said regarding the youtube/no youtube viewing habits that both SeanOzz and Ziggy01 were talking about, I have to agree a bit more with SeanOzz. Not that millions of educated viewers are not watching and critically analyzing this footage on large quality TVs, but when most people "post" a video example, that youtube or Vimeo "post" is generally accessed by clicking on it and going directly to the sample. In fact less than 8% of the public (maybe higher in our circles) even have the ability to watch youtube and or Vimeo on their home HD set. Further, otherwise nice samples are frequently contaminated by poor compression and upload habits. Ok, enough of that. Now where Ziggy01 is correct is when he is speaking about the likes of me and perhaps a lot of us. I am a professional Aerial Cinematographer that has had the opportunity to work on very pro projects where my footage will be used against REDs and ARRIs. I am also a professional Editor and Color Grader and therefor watch everything on very high-end equipment. I mention this not to give my resume' but to let you know that I know that to start comparing anything shot on phantom 1 through 3, GoPro and of course our little friend Mavic with these high-end rigs is a fools errand. That said there is a such thing as best practices and on the technical side the best the camera can do, ideal shooting conditions and post production techniques to help cover up inevitable short comings. It is the search for those answers I am hoping this group can help bring to our small corner of the industry and ultimately why I started this thread. I think the test/testing conditions that I would challenge the group to do and present (and I will of course do it myself as well) is to pick a day with consistent lighting conditions (could even be indoors) point the camera at anything that has some light area and dark (usually a close tree or house-lined horizon), make sure to have some level of critical detail in the shot. Could be tree branches or bark, could be gravel. Put the bird in hover and shoot 5 seconds of the exact same scene. Now not everyone is going to have ND so either shutter speed or ISO won't be exact and I think that is important. My opinion is that ISO is most important when assessing noise so let's standardize on that. Here's what I propose;
I say lets pick a 200 ISO at what ever shutter gives you a good looking overall exposure/histogram (a nice hump in the center) and then shoot the following 6 clips. Regarding FPS, I don't think it matters if it is 24fps or 30fps.

Shoot 5 or 6 seconds of:
1) 3840x2160 D-Cinelike @ 0,0,0
2) 3840x2160 D-Cinelike @ -1,0,0
3) 3840x2160 D-Cinelike @ +1,0,0

4) 2720x1530 D-Cinelike @ 0,0,0
5) 2720x1530 D-Cinelike @ -1,0,0
6) 2720x1530 D-Cinelike @ +1,0,0

We can explore Dlog too, but since not everyone has access to grading software and Luts, that should come down the road.

Lastly I am NOT the BOSS here. If someone one believes that there would be a better (More telling) test procedure I am totally open.

Thoughts?

rb

This is pretty much what I did on Monday Afternoon. I went out for the first time over the weekend after a lot of people started raving about +1 sharpening fixing the watercolour issue, but at 4k I found the footage to be very over sharpened and crunchy so I pointed the Mavic out of the window and did pretty much what you are suggesting.

Here's what I found:

1) 3840x2160 D-Cinelike @ 0,0,0
Same result as 2. Watercolor effect not so pronounced but still visible particularly in the leaves and grass.

2) 3840x2160 D-Cinelike @ -1,0,0
A lot of comression artifacts (watercolour effect) due to negative sharpening. Leaves and grass completely mushy.

3) 3840x2160 D-Cinelike @ +1,0,0
This is what I shot at the weekend and was far to sharp.

4) 2720x1530 D-Cinelike @ 0,0,0
Compression artifacts (watercolour) all but gone. Image was much more natural looking but the image seemed a bit soft. Detail that was completely lost at 4K was now visible (I was looking at a hanging wall basket at a distance in the shade. At 4K it was a blob of grey, at 2K you could make out what it was)

5) 2720x1530 D-Cinelike @ -1,0,0
Didn't bother with negative sharpening at 2k

6) 2720x1530 D-Cinelike @ +1,0,0
This is where I settled. The +1 sharpening just bumped up the sharpness enough but didn't introduce any watercolour effect which I have deduced to be caused not only but sharpness/noise reduction but also the limitation of the small sensor and compression when shooting 4K.

All tests shot at ISO 100 and 30fps

 

[rmb]

Yes! Thank you for this. I am encouraged to do the same today. I am guilty of shooting too much 4k as that what my "clients" want so I have been working hard to perfect it. That said if 2.7k is the answer (and it makes science based on what you and the others have said logically about data rates) then I will be really looking hard at this. Sure hope we can get others involved as well. There will always be opinions, but there will also always be patterns in discovery. Again thanks a million and I will shoot here as soon as the weather breaks latter today.

 

[Lt Shanks]

What people need to realise is that the Mavic has a tiny camera with a tiny sensor and whilst it may be marketed as a 4k camera, it doesn't mean it's a good 4k camera.

The Mavic has some serious limitation at 4k, one being the sensor size and the other being the bit rate and thus compression. Think 40mp smartphone camera vs 20mp full frame DSLR. The DSLR is always going to win.

It took me a while to stop shooting at 4k because I had a 4k camera so I wanted 4k footage, but once I did, everything just looked better, and honestly you really need to sit within a foot of a 55" TV to tell the difference.

Much of the watercolour effect that people see is a) caused by negative sharpening but also by the severe compression applied to the 4k file.

So, if you haven't already, drop down to 2k, give the sensor some breathing space and you should see a marked improvement in the quality of your footage. Sharpness, contrast and saturation are really going to be down to personal taste and your editing technique.

I third the notion...

 

[stratos]

Lol ease dcinelike, +1,-1,0 thanks me later

 

[Ziggy01]

This is pretty much what I did on Monday Afternoon. I went out for the first time over the weekend after a lot of people started raving about +1 sharpening fixing the watercolour issue, but at 4k I found the footage to be very over sharpened and crunchy so I pointed the Mavic out of the window and did pretty much what you are suggesting.

Here's what I found:

1) 3840x2160 D-Cinelike @ 0,0,0
Same result as 2. Watercolor effect not so pronounced but still visible particularly in the leaves and grass.

2) 3840x2160 D-Cinelike @ -1,0,0
A lot of comression artifacts (watercolour effect) due to negative sharpening. Leaves and grass completely mushy.

3) 3840x2160 D-Cinelike @ +1,0,0
This is what I shot at the weekend and was far to sharp.

4) 2720x1530 D-Cinelike @ 0,0,0
Compression artifacts (watercolour) all but gone. Image was much more natural looking but the image seemed a bit soft. Detail that was completely lost at 4K was now visible (I was looking at a hanging wall basket at a distance in the shade. At 4K it was a blob of grey, at 2K you could make out what it was)

5) 2720x1530 D-Cinelike @ -1,0,0
Didn't bother with negative sharpening at 2k

6) 2720x1530 D-Cinelike @ +1,0,0
This is where I settled. The +1 sharpening just bumped up the sharpness enough but didn't introduce any watercolour effect which I have deduced to be caused not only but sharpness/noise reduction but also the limitation of the small sensor and compression when shooting 4K.

All tests shot at ISO 100 and 30fps

Man, that is precisely the process and conclusion I arrived at.

I would suggest that it doesn't get any better than this (option #6).

Happy and waiting to be proven wrong though.

 

[SeanOzz]

Lol ease dcinelike, +1,-1,0 thanks me later

I prefer the look of true color.

 

[Ziggy01]

I prefer the look of true color.

Certainly worth a look. I've really only used the Cine and D-Log profiles. The rest of them looked oversaturated and/or gimmicky to me.

True Color is a flatter profile than Cine?

 

[SeanOzz]

Certainly worth a look. I've really only used the Cine and D-Log profiles. The rest of them looked oversaturated and/or gimmicky to me.

True Color is a flatter profile than Cine?

Tonme true color looks closer to real life but it comes down to personal taste. I find myself doing the least amount of post processing when using true color. Style settings are also personal. I tend to bump up the contrast and saturation when using it in camera but you might prefer it differently.

It's worth a look for sure.

I've always been under the impression people used d-cinelike so they could do color correction in post ? I feel it needs it to some degree.

 

[rmb]

Ok so

Lol ease dcinelike, +1,-1,0 thanks me later

Funny after exhaustive tests I am leaning toward +1-2-1 I am then color grading in Resolve and sometimes re-softening it bit based on content.

 

[rmb]

This is pretty much what I did on Monday Afternoon. I went out for the first time over the weekend after a lot of people started raving about +1 sharpening fixing the watercolour issue, but at 4k I found the footage to be very over sharpened and crunchy so I pointed the Mavic out of the window and did pretty much what you are suggesting.

Here's what I found:

1) 3840x2160 D-Cinelike @ 0,0,0
Same result as 2. Watercolor effect not so pronounced but still visible particularly in the leaves and grass.

2) 3840x2160 D-Cinelike @ -1,0,0
A lot of comression artifacts (watercolour effect) due to negative sharpening. Leaves and grass completely mushy.

3) 3840x2160 D-Cinelike @ +1,0,0
This is what I shot at the weekend and was far to sharp.

4) 2720x1530 D-Cinelike @ 0,0,0
Compression artifacts (watercolour) all but gone. Image was much more natural looking but the image seemed a bit soft. Detail that was completely lost at 4K was now visible (I was looking at a hanging wall basket at a distance in the shade. At 4K it was a blob of grey, at 2K you could make out what it was)

5) 2720x1530 D-Cinelike @ -1,0,0
Didn't bother with negative sharpening at 2k

6) 2720x1530 D-Cinelike @ +1,0,0
This is where I settled. The +1 sharpening just bumped up the sharpness enough but didn't introduce any watercolour effect which I have deduced to be caused not only but sharpness/noise reduction but also the limitation of the small sensor and compression when shooting 4K.

All tests shot at ISO 100 and 30fps

Ok so the weather broke and I finally got out there to do 2 tests. They are exactly as I said I would do and the same as what ecolman kindly did. The only mistake that I mad was that I only shot 6 seconds of each so as soon as you really get looking at the they change. That said what I would do if they go by too quickly is simply pause the video and look at them that way. Also because I did these tests in both 2720x1530 and 3840x2160 once you see the setting you like in each test you can frame by frame forward and backward one to judge the virtues of 2k vs 4k at each setting. Frankly I am getting the sneaking suspicion that the reason why there are such vast discrepancies in settings is not personal taste, but a quality control issue at DJI that is making each Mavic and sensor placement just different enough that we are all picking the settings (corrections) that work for our personal unit and shooting conditions. Here is the link to a 1920x1080 video clip with as little compression as practical. I would like to know if looking at these ungraded images and associated setting look familiar to everybody or if you are getting totally different results from the same settings.

Here's the link:

PW: mavic

PS I will upload a version that I also took outside of the testing specs that is my goto which is D-Cinelike +1,-2,-1 which for me is head and shoulders above the rest for my needs

 

[Alex S.]

interesting thread... I'm following

 

[michalip]

yep , nice stuff, following too

 

[rmb]

Thanks. I sure hope I am not wasting your time by directing you here. It appears that getting anyone to post actual tests is a fool's errand. If fact I am probably about to get slammed or kicked out of the 2k vs 4k forum as I just posted a rant out of total frustration at the incessant and entirely useless practice of "settings vomiting"

 

[horizontal]

I am sure glad they never offered me a job as an art museum director... my eyes don't possess the ability to see the differences or fine detail you all discern of the various settings. It's crazy but I think they all look good. Perhaps it is because I view it on only an iPad but if I ever post any videos you can start throwing tomatos and cabbage at my work!

It's pretty neat that many of you are able to see the differences. And I mean that sincerely.

 

[rmb]

beauty is in the eyes of the beholder my friend. Keep enjoying your own work. its all that counts in the end!!!!

rb

 

[Roy84]

hello everyone,

This is a very helpful and informative thread. i am new to this. i have my mavic for 3 weeks now and it is my very first drone or even 4k camera. i have spent hours and hours on the web and youtube looking for all kinds of information about best camera settings and learning basic camera information like sharpness,contrast ,saturation,iso and shutterspeed grain and noise. I think i understand most of it but i am also stuck at the negative or positive sharpness.when all youtubers were posting to go negative i followed 1 youtuber to go positive.
I am the biggest noob when it comes to pc and yes i say it my self. so i want to get the best possible image from my drone without having to edit it. Now i notice with the sharpness +1 in 4k a good amount of grain/noise. It looks a bit like a mist sometimes and this with 100 ISO in normal daylight. I didn't really try to film at 0 or -1 sharpness because i saw all the movie's online with those settings and they make my video look good.But then again when i see video's with +1 sharpness they do look better because somehow they have a lot less grain/noisy image without editing.
Now i did the test mentioned above in 4 and 2K at +1/0/+1 0/0/0 -1/0/0 each. I was expecting this oilpainting looks in my greens at -1 sharpness but its not. basically the +1 is sharp and detailed but noisy and in 2K better then 4K.
And the 0 or -1 is for my untrained eyes less sharp but also less noisy atleast on my screens and tv screen.

I would very much appreciate it if some of your well trained eyes and knowledge would take a look at my test results and tell me which settings you would take and comment on it but keep in mind to choose for no editing afterwards.

they are all at 30FPS, 100 ISO, 60 shutter, true collor with WB at sunny and ND4 filter.

this is the link of vimeo album with the 6 test results. and like i said i am a noob with pc so i didnt make a fancy edited video. test mavic on Vimeo

Thank you in advance and thank you for the information given to all newbie's like me.

 

[SeanOzz]

hello everyone,

This is a very helpful and informative thread. i am new to this. i have my mavic for 3 weeks now and it is my very first drone or even 4k camera. i have spent hours and hours on the web and youtube looking for all kinds of information about best camera settings and learning basic camera information like sharpness,contrast ,saturation,iso and shutterspeed grain and noise. I think i understand most of it but i am also stuck at the negative or positive sharpness.when all youtubers were posting to go negative i followed 1 youtuber to go positive.
I am the biggest noob when it comes to pc and yes i say it my self. so i want to get the best possible image from my drone without having to edit it. Now i notice with the sharpness +1 in 4k a good amount of grain/noise. It looks a bit like a mist sometimes and this with 100 ISO in normal daylight. I didn't really try to film at 0 or -1 sharpness because i saw all the movie's online with those settings and they make my video look good.But then again when i see video's with +1 sharpness they do look better because somehow they have a lot less grain/noisy image without editing.
Now i did the test mentioned above in 4 and 2K at +1/0/+1 0/0/0 -1/0/0 each. I was expecting this oilpainting looks in my greens at -1 sharpness but its not. basically the +1 is sharp and detailed but noisy and in 2K better then 4K.
And the 0 or -1 is for my untrained eyes less sharp but also less noisy atleast on my screens and tv screen.

I would very much appreciate it if some of your well trained eyes and knowledge would take a look at my test results and tell me which settings you would take and comment on it but keep in mind to choose for no editing afterwards.

they are all at 30FPS, 100 ISO, 60 shutter, true collor with WB at sunny and ND4 filter.

this is the link of vimeo album with the 6 test results. and like i said i am a noob with pc so i didnt make a fancy edited video. test mavic on Vimeo

Thank you in advance and thank you for the information given to all newbie's like me.

I am with you. For my mavic' camera if I want to not do any editing after I can not use +1. Makes my footage look like a grainy mess or did with my original mavic. (I'm still doing some test with its replacement).

To do as little post editing or none I found shooting in truecolor with my style set to -1sharpness +1 contrast and +2 saturation gave me a great look of what I was typically doing on most other set ups in post. Saves me time and headache normally.