Please select Into the mobile phone version | Continue to access the computer ver.
Poor 4K Quality
3881 29 2017-5-19
Uploading and Loding Picture ...(0/1)
o(^-^)o
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Hello,

I have spent some time trying to get the video in 4K better, but finally it always gets 'pixelized' after You Tube compression. But only P4P clips, not others...

Here you can find link to one video I prepare, first 50 seconds (and some parts further) are done with P4P.



4K, 30fps, h264, ND8 filter, 1/60 shutter, some parts ISO 100 + D-Cinematic and some with D-Log (ISO 500), in camera settings set as -3, -3, -3. Processing in Final Cut Pro X, color corrections and sharpening. Still really terrible look when scaled down to 1080p. Even in 4K it looks like poor 1080. And straight in FCPX it doesn't look like 4K.

The rest of the material (from Olympus EM1 Mk II) is perfectly sharp, fantastic. So it's not FCPX compression/conversion issue or YouTube issue alone... It doesn't even look like too aggresive processing (and I dragged lights and darks to +/- 40, even 50). I added 5-10 sharpening max.

Is that something known about Phantom 4? Is it so bad in 4K when there are fast moving objects and the scene is not lit so well? In fact, all scenes look ugly. I really struggle with it so far.

Any suggestion as to what settings/resolution is really best and gives great results? It seems that video is a bit disappointing, at least not what I expected. Thankfully, photos are really great!

Thanks for any suggestions! Would be really helpful, as I'm getting frustrated

Cheers,
Marcin

2017-5-19
Use props
JoiCam
Second Officer
Flight distance : 771434 ft
Iceland
Offline

Try lower the sharpness and contrast to -2 and bring it back as needed in post.  
2017-5-19
Use props
Geebax
Captain
Australia
Offline

Almost your entire clip is made up of fine tree foliage and detail, probably the worst subject to shoot with a camera that utilises MPEG compression. And the short answer is no, you cannot expect the P4 to produce better results than that. If you then put it through YouTube, the final result is always going to be worse.
2017-5-19
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

JoiCam Posted at 2017-5-19 12:50
Try lower the sharpness and contrast to -2 and bring it back as needed in post.

Yes, I had it set at -3 during shooting.
2017-5-20
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Geebax Posted at 2017-5-19 14:26
Almost your entire clip is made up of fine tree foliage and detail, probably the worst subject to shoot with a camera that utilises MPEG compression. And the short answer is no, you cannot expect the P4 to produce better results than that. If you then put it through YouTube, the final result is always going to be worse.

I was suspecting that, but was hoping it's not about the subject only.

But would it look better if recorder in 1080 straight away, rather than 4K and downscaling?

Thanks!
2017-5-20
Use props
Geebax
Captain
Australia
Offline

Sowa Posted at 2017-5-20 00:19
I was suspecting that, but was hoping it's not about the subject only.

But would it look better if recorder in 1080 straight away, rather than 4K and downscaling?

Hmm, it might look a bit better, as the bitrate of 1080p is less susceptible to compression of detail. But hey, only one way to find out.
2017-5-20
Use props
JoiCam
Second Officer
Flight distance : 771434 ft
Iceland
Offline

Sowa Posted at 2017-5-20 00:18
Yes, I had it set at -3 during shooting.

It was in your post   Sorry.
This looks like over sharpness in the details. I do lot of 4k shooting and do not having this problem, even when I use them in 1080 projects.
I usually use D Log and -2 -2 -2 in the settings and shutter set to 2x frame rate and even if I am using non color profile I do not have this problem. When I am filming trees in bright sun I get just minimal over sharpness noise effect only if I use non color profile and all set to -0, It does not happen If I have all set to -2. And I get the best outcome when I use use D Log and All set to -2.
2017-5-20
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

JoiCam Posted at 2017-5-20 02:23
It was in your post   Sorry.
This looks like over sharpness in the details. I do lot of 4k shooting and do not having this problem, even when I use them in 1080 projects.
I usually use D Log and -2 -2 -2 in the settings and shutter set to 2x frame rate and even if I am using non color profile I do not have this problem. When I am filming trees in bright sun I get just minimal over sharpness noise effect only if I use non color profile and all set to -0, It does not happen If I have all set to -2. And I get the best outcome when I use use D Log and All set to -2.

So basically I has very similar approach and it somehow failed. Well, I need to try more, or maybe I use FCPX wrong and my workflow destroys it?

Anyway, same footage made with Olympus, in same forest, full of branches and lots of details, is perfectly sharp.

Thanks!
2017-5-25
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Geebax Posted at 2017-5-20 01:56
Hmm, it might look a bit better, as the bitrate of 1080p is less susceptible to compression of detail. But hey, only one way to find out.

Yep! I have to keep going and see when it gets worse and when I have best results! Maybe one P4 is not perfectly same as another!
2017-5-25
Use props
Dobmatt
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Sowa Posted at 2017-5-25 12:32
Yep! I have to keep going and see when it gets worse and when I have best results! Maybe one P4 is not perfectly same as another!

Your sample video (nice, by the way) contains clips made by two hugely different cameras. To be honest, comparing Oly EM-1 MkII excellent performance with Phantom 4 camera doesn't make much sense. The main culprit here is the bitrate: Oly does UHD 4K video with 102 Mbps easily, while P4 camera struggle with 60 Mbps bottleneck. Another issue: grass and foliage on fast and close passages are extremely difficult to digest by compression algorithm at such low 4K bitrate. That's why your P4 footage looks so pixelated in comparison to Oly footage. Poor implementation of codec by DJI aside, further YouTube processing doesn't help either. Unfortunately not sufficient 4K video recording bitrate is given to any DJI drone camera up to Zenmuse X5R and P4P. Particularly with DJI D-Log formula, which is rather useless IMHO. If you really expect fair competition with your Oly, you must try P4Pro camera, which is absolutely amazing.

I had some success dealing with such artifacts of previous generation 4K footage in Premiere Pro. Try to export troubled clip with identical parameters but with - say - CTR 25 Mbps bitrate and highest rendering quality (not default VBR, 1 pass 10-12). This will deliver bigger output file, naturally. Later assemble new project with original Oly clips and "doctored" Phantom clips and see if there's any improvement. Try to publish you work on Vimeo, perhaps. Let us know ...
2017-5-25
Use props
Geebax
Captain
Australia
Offline

Dobmatt Posted at 2017-5-25 20:50
Your sample video (nice, by the way) contains clips made by two hugely different cameras. To be honest, comparing Oly EM-1 MkII excellent performance with Phantom 4 camera doesn't make much sense. The main culprit here is the bitrate: Oly does UHD 4K video with 102 Mbps easily, while P4 camera struggle with 60 Mbps bottleneck. Another issue: grass and foliage with fast and close passages are extremely difficult to digest by compression algorithm at such low 4K bitrate. That's why your P4 footage looks so pixelated in comparison to Oly footage. Poor implementation of codec by DJI aside, further YouTube processing doesn't help either. Unfortunately not sufficient 4K video recording bitrate is given to any DJI drone camera up to Zenmuse X5R and P4P.

I had some success dealing with such artifacts of previous generation 4K footage in Premiere Pro. Try to export troubled clip with identical parameters but with - say - CTR 25 Mbps bitrate and highest rendering quality (not default VBR, 1 pass 10-12). Later assemble your project with original Oly clips and "doctored" Phantom clips and see if there's any improvement. Let us know ...

Sowa says in the opening post that he is using a P4P, which can shoot with a bit rate of 100 Mbs, so the bitrate should compare with the Olympus, but he does say if he is using the 100 mbs option.
2017-5-25
Use props
Dobmatt
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Geebax Posted at 2017-5-25 20:58
Sowa says in the opening post that he is using a P4P, which can shoot with a bit rate of 100 Mbs, so the bitrate should compare with the Olympus, but he does say if he is using the 100 mbs option.

ooops, what was I thinking? Than I don't know ...
2017-5-25
Use props
Dobmatt
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Przepraszam za pomyłkę, Sowa. Gdzies w połowie napisałes Phantom 4 i dalej już poszło tym torem. Tak czy inaczej sprobuj tej metody, ale to mi się nie podoba w ogóle. Nic takiego nie widzę w moich klipach z P4P+. Sporo takiego smiecia jest z Zenmuse X5, ale - co ciekawe - tylko w D-Log. Wszyscy narzekali w swoim czasie na tę usterkę, w końcu DJI się wycofał z wszelkich prób naprawy i tak już zostało. Aha, sprobuj jeszcze jakies inne SD, P4P jest trochę grymasny. Sam już nie wiem co Ci doradzić. Powodzenia.
2017-5-25
Use props
Dobmatt
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Geebax Posted at 2017-5-25 20:58
Sowa says in the opening post that he is using a P4P, which can shoot with a bit rate of 100 Mbs, so the bitrate should compare with the Olympus, but he does say if he is using the 100 mbs option.

Gee, are you saying the 100 Mbps is optional within P4P setup? I know it's given, but can you choose lower bitrate?
2017-5-25
Use props
Geebax
Captain
Australia
Offline

Dobmatt Posted at 2017-5-25 21:33
Gee, are you saying the 100 Mbps is optional within P4P setup? I know it's given, but can you choose lower bitrate?

To be honest, I don't know, I don't have a P4P, but it might be optional. Perhaps someone else could answer that. I have a Sony A7R where the bitrate is dependant upon the frame rate, and perhaps the P4P is too. (?)
2017-5-25
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Dobmatt Posted at 2017-5-25 21:16
Przepraszam za pomyłkę, Sowa. Gdzies w połowie napisałes Phantom 4 i dalej już poszło tym torem. Tak czy inaczej sprobuj tej metody, ale to mi się nie podoba w ogóle. Nic takiego nie widzę w moich klipach z P4P+. Sporo takiego smiecia jest z Zenmuse X5, ale - co ciekawe - tylko w D-Log. Wszyscy narzekali w swoim czasie na tę usterkę, w końcu DJI się wycofał z wszelkich prób naprawy i tak już zostało. Aha, sprobuj jeszcze jakies inne SD, P4P jest trochę grymasny. Sam już nie wiem co Ci doradzić. Powodzenia.

Hello again! (sorry, I will reply in English, so that other may take part as well )

So I have made another material from P4P - again in 4K (100MBit, no other option), 24 fps, D-Cinematic, -2/-2/0 custom settings (I used -2 for all in others,  same bad results) and I used H265 - I was advised in other forum it should solve my problems. But it did not. Please have a look at below test clip, especially around second 20.



How is it possible? Quality gets terrible. I tried multiple approaches:
- using h265 straight in FCPX and export 4K to YouTube (both straight and via Master file)
- using h264 (converted from h265 by Brorsoft) in FCPX and export 4K to YouTube (both straight and via Master File)
- both of the above, but with 1080p export
- export to 4K Master File from FCPX and then converting that file by HandBrake or Brorsoft to lower bitrate or to 1080p (once I even used 6MBit bitrate for 1080, so even lower than requirements for YouTube - and it looked ok in the computer)

Whatever from above - it always looks the same bad in YouTube.

Dobmatt - you mentioned one method of using raw material (like CTR 25, VBR, 1 pass 10-12....) -   can you explain to me in better language? I don't get it... What are those settings and what soft? What soft should I use? Apple compressor could help?

Or maybe my P4P is just so bad? I think I will just change and start recording 24fps 1080, not 4K anymore. If it helps..

Thanks!
2018-1-13
Use props
Matthew Dobrski
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Sowa Posted at 2018-1-13 14:02
Hello again! (sorry, I will reply in English, so that other may take part as well  )

So I have made another material from P4P - again in 4K (100MBit, no other option), 24 fps, D-Cinematic, -2/-2/0 custom settings (I used -2 for all in others,  same bad results) and I used H265 - I was advised in other forum it should solve my problems. But it did not. Please have a look at below test clip, especially around second 20.

I'm afraid you're expecting too much from Phantom 4 Pro camera, filming such demanding objects like white water (rapids). This is a classic example of H.264/265 codec efficiency. Try to film a glass of boiling water (with submerged electric heater) and you'll see how H.264 is struggling to preserve details of such chaos. BTW H.265 codec is not really better in terms of resulted video quality, it's about drastically smaller file for Internet distribution.

Summarizing, we're talking about 3 separate issues here:

1. The quality of raw footage, determined mainly by recording bit rate and efficiency of compression codec implemented by DJI, either H.264 or H.265. This is given after the fact and there's nothing you can do about it.
2. The quality of rendering during media export from within video editor environment. Here you have a plethora of options to choose from, among them the bit rate as being the most important. More about this later.
3. The quality of Internet player (i.e. YouTube, Vimeo etc.). They'll give you some resolution options to choose for playback, but they will process your file again anyway, applying lower bit rate and therefore reducing file size. Another words, what you'll consider of acceptable quality on your 4K computer monitor may become of terrible quality on YouTube. And this is also beyond your control, unfortunately.

Now, back to pt.2. The more fast moving details (grass, foliage, falling water etc.) is in the frame, the harder codec is working to preserve them. Too much details means that visible skipping may occur, and this is exactly the case. Some areas of frames may appear mushy. Said that, Phantom 4 Pro camera performs splendidly, one must agree, taking 100 Mbps rate into consideration. The trick is to not make this worse during post production. Among other output parameters like frame size/rate and compression codec, rendering bit rate is critical. These are accessible in Export dialog, tons of YouTube tutorials explain these issues in detail. I'm using output video parameters identical to original material and Constant Bit Rate (CBR), rendering of about 25 value with good results. Some people may recommend even higher rate when dealing with H.264/265 codecs. Avoid Variable Bit Rate (VBR), in Adobe Premiere Pro set by default to between 10 and 12, as not capable to produce quality 4K output.   
2018-1-13
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Matthew Dobrski Posted at 2018-1-13 18:33
I'm afraid you're expecting too much from Phantom 4 Pro camera, filming such demanding objects like white water (rapids). This is a classic example of H.264/265 codec efficiency. Try to film a glass of boiling water (with submerged electric heater) and you'll see how H.264 is struggling to preserve details of such chaos. BTW H.265 codec is not really better in terms of resulted video quality, it's about drastically smaller file for Internet distribution.

Summarizing, we're talking about 3 separate issues here:

Thanks!

The only thing that is strange is that the footage gets pixelized randomly, sometimes a couple of seconds it looks fine and then it gets that mushy moments. Also, notice that not all same looking parts of the image get mushy, what's even more strange. Additionally, as I'm shooting landscape mostly, then trees and grass are quite common, so if P4P struggles so much, then I'm quite disappointed. Please have a look at below full project and see in how many different places it struggles - that is export from FCPX straight to YouTube:



Could we just summarize the workflow that you suggest for that? So... In terms of raw footage from DJI (in this case it's H265, as I said): should I use it as it is in FCPX, or should I first run it through any converter with certain settings, then use it in my project? (if you have any specific converter you use, please shar what you use). Secondly, once my project is done in FCPX and ready for publishing - should I use native YouTube export or I should export to Master file and then convert? I expect second option is better, so once I have huge Master File, what conversion parameters would be best and are used by you?

Problem is, that even though whole my raw footage is H265 100Mbit, but once I export from FCPX to Master File with highest quality, it is... 30Mbit only. So looks like I should use something like Apple Compressor to have custom export settings with full quality. I would expect to have Master File with same bitrate as original material.

Could I overcome that problem when I shot 1080p 24fps instead of 4K? Then I would have more data per frame in 1080p. It would mean 4K is really sensitive to subject.

And please be aware again, I'm FCPX user, but if I can achieve better rendering with Premiere, then maybe I should switch...

If you want, I can share that raw file with H265 to let you try get good results. I would be interested to see if anyone could make it nice.

It's better in Vimeo, but still far from what I would expect. Never had such issues with GoPro footage, even though it's smaller sensor and smaller bitrate...

Huge thanks for help!
2018-1-14
Use props
YOYOMAN
First Officer
Flight distance : 676260 ft
Réunion
Offline

What about your setting Anti Flicker? 50hz? 60hz?
Have the original's rushes the same poor quality?
2018-1-14
Use props
Matthew Dobrski
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

YOYOMAN Posted at 2018-1-14 11:28
What about your setting Anti Flicker? 50hz? 60hz?
Have the original's rushes the same poor quality?

Anti Flicker frequency option is developed mainly to minimize picture flickering in presence of fluorescent light. This is country dependent, some employs 50Hz AC grid power, other 60Hz. This setting is irrelevant in landscape filming.

http://www.school-for-champions. ... st.htm#.WlvW_ainFhE
2018-1-14
Use props
Geebax
Captain
Australia
Offline

Sowa Posted at 2018-1-14 03:30
Thanks!

The only thing that is strange is that the footage gets pixelized randomly, sometimes a couple of seconds it looks fine and then it gets that mushy moments. Also, notice that not all same looking parts of the image get mushy, what's even more strange. Additionally, as I'm shooting landscape mostly, then trees and grass are quite common, so if P4P struggles so much, then I'm quite disappointed. Please have a look at below full project and see in how many different places it struggles - that is export from FCPX straight to YouTube:

Frankly, I don't use any Apple software, I do all my work with Da Viinci Resolve 14, which has very good control over the output rendering, but when I import the material into Resolve, I do so at the first stage and choose a high quality intermediate that actually makes the footage much larger file wise.

I would suggest you might upload one of the clips you have shown us to Dropbox, in the form of the edited product, so we can look at it in its original form.
2018-1-14
Use props
Matthew Dobrski
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Sowa Posted at 2018-1-14 03:30
Thanks!

The only thing that is strange is that the footage gets pixelized randomly, sometimes a couple of seconds it looks fine and then it gets that mushy moments. Also, notice that not all same looking parts of the image get mushy, what's even more strange. Additionally, as I'm shooting landscape mostly, then trees and grass are quite common, so if P4P struggles so much, then I'm quite disappointed. Please have a look at below full project and see in how many different places it struggles - that is export from FCPX straight to YouTube:

Sowa, firstly let's find out if original (unprocessed) footage exhibits traces of this ugliness. To do that simply transfer your file to a computer drive and play back via sufficient video player. Note that not every one player supports H.265 codec, you must shop around and install the proper one.

I have limited experience dealing with 265 codec since the implementation of this new technology is still in embryo state. For all my amateur-ish needs well established H.264 is good enough to maintain decent 4K video quality. Venerable Adobe Premiere Pro CC editing software - for example - takes several hours to render few minutes long output file with 265 codec, while 264 is done in 30 min average.

Also - I must confess - YouTube fame and glory is not my thing. If I want to share my films I'll use a Vimeo paid subscription, which offers better playback parameters. Please note again: every Internet publishing place will process your downloaded file up to their standards, downgrading the playback quality more or less and introducing compression artifacts.

I don't believe that any conversion process is harmless and necessary in this case, therefore I don't use any. I simply import original files into project as they are copied directly from SD card to G-Raid storage. I film 3840x2160 size with 29.97 fps as MPEG Movie and H.264 codec. After all post production cuts, color grading and tweaks in Premiere, I'll export using exactly the same parameters. This is done by choosing 2 basic export settings at the top of dialog: Format: H.264 and Preset: Match Source - High bitrate. This decision will result in Summary description, where you'll see Output and Source parameters. These should be identical. Next, determine the most critical parameter: rendering bitrate. Scroll down the dialog, you'll see a row of tabs: Effects>Video>Audio>Multiplexer>Captions>Publish. Click Video tab, you'll see Bitrate Settings options, set Bitrate Encoding to CBR and Target Bitrate [Mbps] to - say - 25. That should be good enough to preserve most of original file quality. You may, however, experiment with bitrates as high as 240. Finally check Use Maximum Render Quality box at the very bottom of dialog and hit Export button. This will open progress bar, rendering output file to a location determined in Output Name (Save As) dialog above. Note that Bitrate Setting will affect output file size: higher values will result in bigger file.

Again, examine the quality of original footage first. If these artifacts are clearly visible, than indeed there's something wrong with your camera. If not, than with settings described above you should be able to get at least something like this:



Far from perfection, you'll see mushed fragments here and there, but it is what it is with such demanding image. The same film played on my monitor directly from hard drive doesn't exhibit any particularly annoying artifacts. Anything better than that will require Inspire 2 class camera and better than H.264/265 codec.

Powodzenia,
Maciej Dobrski.
2018-1-14
Use props
Matthew Dobrski
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Sowa Posted at 2018-1-14 03:30
Thanks!

The only thing that is strange is that the footage gets pixelized randomly, sometimes a couple of seconds it looks fine and then it gets that mushy moments. Also, notice that not all same looking parts of the image get mushy, what's even more strange. Additionally, as I'm shooting landscape mostly, then trees and grass are quite common, so if P4P struggles so much, then I'm quite disappointed. Please have a look at below full project and see in how many different places it struggles - that is export from FCPX straight to YouTube:

Please share the most obviously problematic raw footage in Dropbox, I would like to examine it and experiment in PP with rendering.
2018-1-14
Use props
kenshatz
lvl.1
Flight distance : 125492 ft
United States
Offline

Youtube always makes my videos the same. Darkness and trees always tricky to get right.
2018-1-17
Use props
randygd
lvl.3
United States
Offline

As suggested play the native file and see if you are getting the same artifacts...and get compressor. Stepping on 100mb to 30 will cause a quality hit. Also know FCPx does not give full frame full res playback in preview.What settings are you using for the 4k from the Olympus? Also some of what I am seeing looks like issues I have had with interlaced video and the field(lower v upper) reading wrong in the edit suite. But I do see  you are using 30p so it doesn't make sense and FCPx should correctly interpret whatever is on the timeline. What project settings are you using in FCPx?  I am wondering why the ND filter and only 60fps. Dump the filter and go to higher frame rate. I say this because it looks to me like frame issue with some artifacts...clear for split second then not. Again the native file will tell the story
2018-1-17
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

fansb515097a Posted at 2018-1-14 11:32
I would check to see if your auto focus is really working ,   this could explain whey some stuff look great and others soft, I do not think it is anything to do with bit rate or work flow.    This appears to be something with the camera.  

I shot this up close on the rocks to test autofocus and such and you could see it working.  I also shoot alot of videos in the Rain and the Snow and the pictures are vibrant and clear.   

Hi,

I don't think it's autofocus, as I'm filming (including those clips) in manual focus mode with infinite focus - one of the settings widely advised on P4P, and which I tested myself and proved to be a good one (never had out of focus photo/video with that).

Cheers,
Marcin
2018-1-19
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

randygd Posted at 2018-1-17 18:58
As suggested play the native file and see if you are getting the same artifacts...and get compressor. Stepping on 100mb to 30 will cause a quality hit. Also know FCPx does not give full frame full res playback in preview.What settings are you using for the 4k from the Olympus? Also some of what I am seeing looks like issues I have had with interlaced video and the field(lower v upper) reading wrong in the edit suite. But I do see  you are using 30p so it doesn't make sense and FCPx should correctly interpret whatever is on the timeline. What project settings are you using in FCPx?  I am wondering why the ND filter and only 60fps. Dump the filter and go to higher frame rate. I say this because it looks to me like frame issue with some artifacts...clear for split second then not. Again the native file will tell the story

Hi,

Again, I don't think the shutter is the issue. The clip from the beginning of the thread was first filmed by me without ND and my shutter speeds were up to 1/500. It was same result as with ND and slower shutter. The longer I look at it, the more I think it's just not the best way DJI has implemented the codes and also it's very fragile for fast motion of such detailed staff like trees and grass (especially when filming close to them). From bigger distance and with slower motion it's much better. Just a very important thing to remember

Thanks,
Marcin
2018-1-19
Use props
Sowa
lvl.1
Flight distance : 4226 ft
Poland
Offline

Matthew Dobrski Posted at 2018-1-14 15:59
Sowa, firstly let's find out if original (unprocessed) footage exhibits traces of this ugliness. To do that simply transfer your file to a computer drive and play back via sufficient video player. Note that not every one player supports H.265 codec, you must shop around and install the proper one.

I have limited experience dealing with 265 codec since the implementation of this new technology is still in embryo state. For all my amateur-ish needs well established H.264 is good enough to maintain decent 4K video quality. Venerable Adobe Premiere Pro CC editing software - for example - takes several hours to render few minutes long output file with 265 codec, while 264 is done in 30 min average.

Hi again

Please see the link to one of the shots:
https://www.dropbox.com/s/74o1qwu226qmzbt/DJI_0015.MP4?dl=0

Anyway, I think the problem for DJI are not the details of trees alone, but connected with fast motion. Then it just is too lazy with codec. That's of course something that should not happen in better camera, but something to remember when using P4P.

Additionally, please have a look at the same in Vimeo (password: beskid2017):


You are right, Vimeo is much better (of course, watched in 1080 at least). It was exported without any magic, just using embedded Vimeo export in FCPX. First scene with water is a way better, very pleasing, and later artifacts are not so visible - very comparable to what I can see in your video (very nice!).

Please let me know what you think after looking at raw footage, and huge thanks for help, that's awesome!

I think I will just continue what I do and just be more careful in certain topics when filming. And I should think about Compressor, or moving to Premiere...

Cheers,
Marcin
2018-1-19
Use props
Matthew Dobrski
Captain
Flight distance : 1831050 ft
  • >>>
Canada
Offline

Sowa Posted at 2018-1-19 13:54
Hi again

Please see the link to one of the shots:

Spoko, Marcin, there's absolutely nothing wrong with your Phantom. Taking huge amount of details (tiny branches) in your footage into consideration, compression codec performs within expectations, at least for DJI standards. The worst areas are around branches with water in background, where codec is struggling in effort to distinguish the difference between both objects. The same branches on contrasting background looks OK. This is quite normal. Similar footage taken by my X5 camera with 60Mbps will look much worse, believe me.

Just for the sake of experimentation I exported your footage in Premiere with settings described in previous post, using 25. 50 and 100Mbps renders. As expected, 25Mbps output was of acceptable quality but with visible artifacts, 50 provided much better render, and 100Mbps resulted with exactly the same size file (385MB), where I couldn't tell the difference between original and exported video.

Bottom line is that the quality of exported video is directly related to a rate of rendering. It's a compromise between size and quality.

Powodzenia,
Maciej Dobrski.
2018-1-19
Use props
Geebax
Captain
Australia
Offline

Matthew Dobrski Posted at 2018-1-19 20:44
Spoko, Marcin, there's absolutely nothing wrong with your Phantom. Taking huge amount of details (tiny branches) in your footage into consideration, compression codec performs within expectations. The worst areas are around branches with water in background, where codec is struggling in effort to distinguish the difference between both objects. The same branches on contrasting background looks right. This is quite normal for 100Mbps. Similar footage taken by my X5 camera with 60Mbps will look much worse, believe me.

Just for the sake of experimentation I exported your footage in Premiere with settings described in previous post, using 25. 50 and 100Mbps renders. As expected, 25Mbps output was of acceptable quality but with visible artifacts, 50 provided much better render, and 100Mbps resulted with exactly the same size file (385MB), where I couldn't tell the difference between original and exported video.

I agree with Matthew/Marciej. I would like to add though, the real problem is the use of MP4 as a capture mechanism, it was never really intended for this purpose, it was designed as a distribution medium. So it makes decisions at the time of encoding to do the best job of encoding the high detail, and it is then locked into that from then on. Whereas lower compression systems cope better with the capture process and leave it to the user to optimise the distribution format according to how it will be distributed.

Time for DJI to move on to CinemaDNG format. The problem is that the average consumer is barely able to cope with MPEG4, much less a lightly compressed format that makes great demands on the recording medium and the post-production process.
2018-1-19
Use props
Advanced
You need to log in before you can reply Login | Register now

Credit Rules