1080p 120FPS Video completely unusable, what can be done?
12Next >
10727 49 2016-3-26
Uploading and Loding Picture ...(0/1)
o(^-^)o
graeme.bull
lvl.2
Canada
Offline

I saw another thread a couple of weeks back on this subject but I can't find it on the Phantom 4 not doing well with 120FPS. Essentially it commented that the 120FPS video is not only using just a quarter of the sensor, but the bitrate at which it saves the video is the lowest of all the video options that could be chosen. Something incredibly small like 12Mbps in comparison to the 4k 30FPS of 40Mbps.

Has anybody been able to get good video that isn't just horribly pixelated or completely lacking in detail? In comparison to the 4K video, it's bad, but the real issue is that it's unusable due to the lack of clarity in the picture. Trees turn into a mash of green and brown, water completely lacks any detail and the sky tends to just be one color with minor gradations in it. Let alone all the noise that is included it makes me think that there must be something I'm doing wrong.

I've turned off that 3D noise reduction thing and I haven't messed with any other settings other than setting the color pref to d-log. I'm not concerned about saturation, but I need details in my videos or they just look like overcompressed video.

Any tips would be awesome, unless this is either a bug or "just the way it is".. which would suck.

2016-3-26
Use props
Geebax
First Officer
Australia
Offline

I can't say, as I don't have a P4, but Are you judging the quality from the video contained on your phone/tablet or by looking at the files directly off the SD card?
2016-3-26
Use props
madaerial
lvl.2
Flight distance : 512251 ft
United States
Offline

I haven't tested this, but it will be on my list now. I was VERY surprised to see those frame rates at that resolution. Is that using the stock SD card?
2016-3-26
Use props
graeme.bull
lvl.2
Canada
Offline

It is off of the SD card. It's definitely one of the reasons I bought the P4.
2016-3-26
Use props
madaerial
lvl.2
Flight distance : 512251 ft
United States
Offline

I know I have had issues before with SD cards not being fast enough for cameras. I doubt that would be the case with this issue though, as you are using the stock card. I'm sure the lower you go with resolution, the better the results will be. May have to drop it down to 720. I will report with results when I get around to testing it.  
2016-3-26
Use props
graeme.bull
lvl.2
Canada
Offline

So nobody else has a problem with the 120fps video? Perhaps mine is defective?
2016-3-28
Use props
raph_london
lvl.1
Flight distance : 258622 ft
  • >>>
France
Offline

I have the same exact problem: the image is really crap, don't know why.
When I've switched to 1080p 120fps it has zoomed in the image and the result was really bad...
I'll hope someone will figure out

2016-3-29
Use props
viktor8
lvl.1
Flight distance : 18776 ft
Italy
Offline

Actually 1080p60 and 4K30p both have a maximum bitrate of up to 60Mbps. I haven't even found a way to edit that in the menu. Btw the problem with 1080p120 is not only the one related to the bitrate that you're claiming but also the fact that the video gets automatically slowed down when reproducing it on a computer as, almost no monitors have a refresh rate of 120Hz. So I only record in either 1080p60 or 4K30p
2016-3-29
Use props
graeme.bull
lvl.2
Canada
Offline

I've done another test with 4K as well and the quality is absolutely terrible. Whatever codec is being used with whatever algorithm just produces video with such low quality for detail. At first glance it looks good but if you look slightly close then it starts to look like an 8 bit video game with pixelation all over the place. I'm genuinely disappointed with this purchase as you can't even change out the camera that is on it for something that actually produces a good picture, like a gopro. At the very least I hope a future firmware option is available that lets us set the compression and/or codec.

Here's a small part of the 4K shot I took, noncompressed, just a screenshot in PNG (lossless):

p4-1.png
2016-3-29
Use props
mjlstudios
lvl.4
Flight distance : 1540131 ft
  • >>>
United States
Offline

I am sure glad I waited to see if the P4 was all that! I have learned to wait when new items first come out to see what the reviews really are and not what the manufactures claim to be. The photos from the 4K 120fps video that graeme.bull took are nasty. They look like the photos we got from the very first digital cameras years and years ago!
2016-3-29
Use props
GraemeB
lvl.2
Canada
Offline

It's not a still shot, it's a screenshot from a piece of a 4k 30fps shot video and the more I mess with this thing, the more I'm wondering what the return policy is. I'm severely disappointed with not only the quality of the camera but also the lack of any response from a DJI rep. If this was a "fan site" or something, I can understand they might not monitor or respond, but this is the official forum and site of the company. I would expect a response one way or another about this as I'm not the only person complaining.
2016-3-29
Use props
lido_bmt
lvl.2
Flight distance : 87772 ft
Offline

viktor8@hotmail Posted at 2016-3-30 00:54
Actually 1080p60 and 4K30p both have a maximum bitrate of up to 60Mbps. I haven't even found a way t ...

Uh, frame rate has nothing to with a monitor's refresh rate unless you're playing it back at 1:1 speed (and even if you do, the monitor will adjust for it). 1080p120 is in all likelihood going to be played back at < 60 FPS, usually 30 or 24. This is called over cranking and is extremely common.
2016-3-29
Use props
Vylkeer
lvl.1
Flight distance : 18776 ft
Italy
Offline

lido_bmt Posted at 2016-3-29 22:22
Uh, frame rate has nothing to with a monitor's refresh rate unless you're playing it back at 1:1 s ...

They are totally related to each other. I know that monitors would normally adjust themselves slowing the footage from 120p down to 60p but I've tested it on my Mac (connected to my TV via HDMI) and the video played slow as hell. While with 1080p60, I got 1:1 normal speed. So there's definitely something wrong going on with the 1080p120 recording mode, as it's impossible to get it to normal speed and it probably runs at 50% reduced speed (60p) anyways so it's better to record at 1080p60 from the beginning. Maybe in the future we'll get 1:1 normal speed when recording at 1080p120 as well.
2016-3-29
Use props
nebelung
lvl.3
Flight distance : 613615 ft
Offline

GraemeB Posted at 2016-3-30 03:11
It's not a still shot, it's a screenshot from a piece of a 4k 30fps shot video and the more I mess w ...

I think it was probably my thread you were referring to in your original post. Here's the link in case you want to look at it again: http://forum.dji.com/thread-45784-1-1.html

Yes, the 1080p120 video quality is disappointing. Some of it has to do with hardware limitations, but I believe there is room for a small improvement that could be addressed with a firmware upgrade.

I usually shoot my 4k videos in D-Log, with -1 sharpen, -1 contrast and -1 saturation, and I add whatever sharpening, color and contrast I want in post. The issue is really that 60Mbits bitrate, which is not a lot for 4k video. Too much detail and you get artifacts you can't get rid of. But you can get some nice videos that way.

When I tried the same for settings for 1080p120 on the P4, I was very disappointed. At this point I am shooting 1080p120 with +1 sharpen, 0 contrast and 0 saturation. It's better. Any more sharpening and you get too many compression artifacts. I have a video where I tried the different settings here:

When you shoot in 1080p120, the P4 creates a 1080p30 file that has a bitrate of 12.5Mbits. That seems really low (well, it is...), but keep in mind that when recording the video, the P4 writes 4 frames of video every 1/30s. It's the same amount of data written as a 1080p30 video with a bitrate of 50Mbits (the P4 records 1080p30 at 40Mbits). But it's less data being recorded than in the 1080p60 or 4k30 mode, so I am hoping it can be improved.

And yes, I would love to hear from an official DJI rep as well...
2016-3-29
Use props
lido_bmt
lvl.2
Flight distance : 87772 ft
Offline

Vylkeer Posted at 2016-3-30 04:54
They are totally related to each other. I know that monitors would normally adjust themselves slowi ...

No, they don't have anything to do with each other. Refresh rate is literally the number of times per second an image is updated on a display. Frames per second are the amount of temporal information captured by a device. One is in an input parameter (frame rate), and the other is a display parameter (refresh rate). You could have a 10000FPS device and it could be displayed on an ancient CRT monitor stuck at 30hz if the target output frame rate parameter was 24FPS.

When you playback a 120FPS clip it is almost always played back at 24, 30, or 60 FPS. This is the part that's called overcranking and has nothing to do with your refresh rate and has everything to do with the configured playback frame rate. It's also why your video at 120FPS is playing at slow speed: it's supposed to do this.



The only time your 120FPS footage would playback at 1:1 is if your monitor supported a frame rate of 120 (this is not the refresh rate!) at that resolution and your output device played it back at 120FPS. Macs without special hardware do not output at 120FPS.









2016-3-29
Use props
Geebax
First Officer
Australia
Offline

graeme.bull@sol Posted at 2016-3-30 04:25
I've done another test with 4K as well and the quality is absolutely terrible. Whatever codec is bei ...

Those two shots looked really bad, but how can you expect anyone to make a judgement on images that are 570 x 357 pixels. Mediainfo reports a bit depth of 32 bits and a filesize of 299Kb. I don't know what you have done to those shots, but they are certainly not 4K.
2016-3-29
Use props
GraemeB
lvl.2
Canada
Offline

Geebax Posted at 2016-3-29 16:13
Those two shots looked really bad, but how can you expect anyone to make a judgement on images tha ...

It's either you don't have an understanding of what is happening here, or haven't bothered to take the time to read the whole thread and pause for a second to understand what is happening. I can't figure it out as I've not seen such a weird response for the issue being presented...

Just to reiterate, I took a screengrab of a portion of the 4K shot. I certainly didn't shrink my 4K shot down to a 570 x 357 size picture as that obviously wouldn't show the problem... When I commented that it looks like an 8 bit video game, it truly does look like that. Each pixel looks like it has been duplicated to the next one over in many spaces. It's as if the 4K isn't actually captured at 4K and instead it's captured at 1K and then upsized to create truly crappy 4K and the algorithm is just doing a "near as possible pixel duplication".

Good on you though for grabbing the media info of the PNG and seeing that it is in fact 32bits of good info as it makes it clear that I didn't downscale the bits within the image.

Honestly though, if you can't be bothered to contribute then don't just come in here criticizing where you have no knowledge as it just dilutes the conversation.
2016-3-29
Use props
Geebax
First Officer
Australia
Offline

GraemeB Posted at 2016-3-30 11:18
It's either you don't have an understanding of what is happening here, or haven't bothered to take ...

No need to get your knickers in a knot, I am probably one of the few people here who can appreciate issues of image quality, as I have been doing it for 50 years. One of the main reasons you have not heard from any DJI person on this issue is because issues of image quality are often outside their area of expertise.

I did read the whole thread, however I did not pick up on the fact that you are presenting a portion of the shot. It is customary on this forum to provide a link to the original material, as there are a small number of people expert in the photography field who would look at the images to offer advice.

But it would seem that you don't want any help with this issue, or only want to hear from people who agree with your point of view, so I will leave you to it.

2016-3-29
Use props
Vylkeer
lvl.1
Flight distance : 18776 ft
Italy
Offline

lido_bmt Posted at 2016-3-30 00:21
No, they don't have anything to do with each other. Refresh rate is literally the number of times p ...

Then there's something I don't know. I've always thought that for optimal playback the Hz (I know what refresh rate and fps mean btw) had to match the fps rate. So 60p on a 60Hz display and so on. That would be the best situation ever, without any pulldown algorithms used by the display to make it look decent when for example the video runs at twice the speed of the monitor (120p on 60Hz display). Point is, shooting at 1080p120 with the P4 or even a GoPro 4 at this point, is stupid and pointless since the footage will be automatically slowed down and won't be playable at 1:1 normal speed.
2016-3-30
Use props
GraemeB
lvl.2
Canada
Offline

It's really disappointing to see only users tracking this post instead of somebody from DJI finally stepping in to note that this is being looked at. It is clear that the camera is NOT taking 4K video and is instead taking 1080 video and upscaling it to 4K with an algorithm. I hope a firmware update in time opens up the power of the sensor. I saw on another thread that it's the same as the gopro 5? If that's the case then it should be possible to take true 4K video and proper 120fps without huge compression artifacts.

As for playback, if you have a player that will recognize how many frames are in each second of video, then you shouldn't have a problem. Most don't though, oddly. All editing programs will recognize the footage though and you should be able to edit accordingly.
2016-3-30
Use props
grangerfx
lvl.4
Flight distance : 817713 ft
United States
Offline

GraemeB Posted at 2016-3-30 08:29
It's really disappointing to see only users tracking this post instead of somebody from DJI finally  ...

Um no. I need to get a 4K screen to view my videos on because they exceed the resolution of my 2560x1440 monitor. I can pause the video and see that there is clearly sub-pixel detail (small lines are sharp and perfectly anti-aliased right down to the pixel elements on the display). The only issues I see are that the bit rate can always be higher (less compression artifacts during rapid motion) and it really looks bad when you change the camera pitch rapidly. If you are seeing pixel doubling you have a software problem on your end.
2016-3-30
Use props
GraemeB
lvl.2
Canada
Offline

Just because the image size happens to be large, doesn't mean the original captured image was large.

All you have to do is glance at the screenshot above. That was taken from Premiere (latest version) without any upscaling on my end. That fact that it looks like a blotchy 8 bit video game is pretty good proof that the algorithm being used is either upscaling or just about the poorest compression algorithm on the planet as it lacks any detail at all. Many others are reporting the same problem, it's not just me. I expect as more sales go out that there will be many more high end users complaining about it.
2016-3-30
Use props
lido_bmt
lvl.2
Flight distance : 87772 ft
Offline

Vylkeer Posted at 2016-3-30 17:00
Then there's something I don't know. I've always thought that for optimal playback the Hz (I know w ...

The reason people shoot at higher framerates is so that when played back at normal frame rates (like typical 24FPS) you get very smooth slow motion. If you try and slow down 24FPS, you get a very stuttered, blurred video.
2016-3-30
Use props
GraemeB
lvl.2
Canada
Offline

Here are a couple of images to look at. If anybody needs the original footage then let me know, but I don't see why it would be required to see how clearly the sensor and image compression is working. I expect all of the P4s take this kind of footage:



drone-cam-pixelation-1.png
2016-3-30
Use props
GraemeB
lvl.2
Canada
Offline

Here is footage from 1080p. I'm unsure of how to add images and then add posts as the image doesn't show up in the preview panel here, so I'll explain the above two images in the next post:



drone-cam-pixelation-2.png
2016-3-30
Use props
GraemeB
lvl.2
Canada
Offline

The above two images are just small pieces of the original image. I haven't compressed them or changed their size at all. Again, nothing has been done to modify these images, they are just still frames taken from video. The first image came from 4k at 30fps and the second one is from 1080 at 60fps.

The top image has the areas in the red square from both images in it, the below image only has the red square from it's own image. The enhancement is 400% in Photoshop through the scale tool.

What you should be able to notice immediately is the lack of detail and clarity in the 4K. In fact, it looks a lot like the blown up 1080p footage! The 400% increased swatch shows more clearly how there is just a big muddy mess of pixels, a lot like the blown up 1080p footage. This is what is leading me to believe that the sensor is not actually capturing 4K and instead capturing at 1K and then an algorithm is applied to get to 4K.

This also corroborates many people's claims that when they downscale to 1080 it looks great. Of course it does as that is what it was actually captured at and all the pixels are just being put back where they belong.

I would love to be proven wrong, but if I am wrong then it leads to the question of what kind of compression is being applied to make the footage so crappy.
2016-3-30
Use props
StevenQX
lvl.4
Flight distance : 287474 ft
Italy
Offline

For me is absolutely puzzling that a 2016 aerial camera that cost 1400€ still has a 12mp sensor only while 2 years old (or even more) smartphones have 16mp+ sensors that produces a lot better quality  video and stills. When i read that the p4 has the same 12mp sensor that the p3 had (for me a 12mp sensor was old even in 2015 for p3) it was already enough to decide to do not buy it. Could have flew even 40 minutes but most of all is a digital aerial camera and the first things it have to do are good/great quality stills and video. Because is born for that reason.
2016-3-30
Use props
Geebax
First Officer
Australia
Offline

GraemeB Posted at 2016-3-31 02:29
It's really disappointing to see only users tracking this post instead of somebody from DJI finally  ...

'It's really disappointing to see only users tracking this post instead of somebody from DJI finally stepping in to note that this is being looked at. It is clear that the camera is NOT taking 4K video and is instead taking 1080 video and upscaling it to 4K with an algorithm. '

I was drawn back to this discussion. You will not get people from DJI commenting on this because it is not an area in which they have the expertise. Someone like Ken may obtain advice from DJI in China and post the replies, but that is as far as it goes. Once upon a time DJI experts in various fields used to come on this forum, but it just does not happen any more, probably got fed up with the abuse.

While I don't have a P4, I do have a P3P and it most certainly does shoot in 4K. and I am quite familiar with the image processing chip used in the design and it does say in it's specifications that it is capable of recording 4K video material. Therefore, I would ask the question, why would DJI downgrade the video recording characteristics on the newer product? The answer is that they would not do so.

And from a purely technical standpoint, you cannot upscale 1080 x 1920 material to 4096 x 2160 and make it look good because you cannot manufacture resolution, no matter what algorithm you use. You can however downscale from 4K to 1080 quite well, especially if you use high quality programs to do so. And people will say that 1080p looks great because the vast majority of people cannot view 4K material on anything, while most monitors these days support 1080p natively.

Having said that, I am still trying to figure out why your material does not look good, so far there have been no clues, however I have viewed original footage from a P4 and it was stunning, just noticeably better than that from my P3P, so your experience is not typical.

So, I have two questions, what is the resolution of the monitor you are viewing your material on, and what make and model is it. And secondly, can you post an original piece of the video you are talking about via Dropbox or similar.

The last thing is that you mention earlier that the examples you showed were a 'screenshot'. Most people relate to a screenshot as a section capturted from their display screen, in which case it then also is subject to the processes inside the monitor and is not representative of the image. If you took the shot another way, perhaps explain how you did it.


2016-3-30
Use props
Vylkeer
lvl.1
Flight distance : 18776 ft
Italy
Offline

lido_bmt Posted at 2016-3-30 18:44
The reason people shoot at higher framerates is so that when played back at normal frame rates (li ...

I know, that's why higher framerates like 60p or 120p are ideal for normal speed as well. Every turn of the camera look way more fluid with less stuttering and flickering, caused by slow fps rates like 24p or 30p. Is not something good for slow mos but also for normal speed playback.
2016-3-31
Use props
LordNeo
lvl.3
Flight distance : 548212 ft
Chile
Offline

Geebax Posted at 2016-3-30 21:14
'It's really disappointing to see only users tracking this post instead of somebody from DJI finall ...

I think (and expect) he is doing the "extract frame" from premiere pro to get the snap.
I'll get two snap from similar footage on 1080p (60fps) and 4k (29fps) to compare, but as far i've seen on 4K displays the difference is more than noticeable.

4k screen: https://mega.nz/#!dtUTHSob!rIXlkVKO69UCYNfdY_rFOIEM8HIjked_ncuEv8LZLZE
1080p screen: https://mega.nz/#!wk1EyCQa!V1EVONy5XwtZhrvFqBK_5lmr8OgrLNDyqEvQ_tH9-2I
When look at the same size, 1080p looks more sharp, but comparing sectors, 4k is a lot more sharp. Sorry the quality of the footage, I record every part of the flight even if isn't useable. This part is just when i did a video mode change.
2016-4-1
Use props
Geebax
First Officer
Australia
Offline

I am fortunate that my main screen is 4K capable and driven from a GPU card, so my images are not compromised by the operating system trying to make 4K fit into a smaller screen. But it is not inconceivable that the image processor in the Phantom is playing fast and loose with the images, but I would just like to see a sample of the original material.
2016-4-1
Use props
kirk2579
lvl.4
Flight distance : 37415 ft
United States
Offline

Geebax Posted at 2016-4-1 18:37
I am fortunate that my main screen is 4K capable and driven from a GPU card, so my images are not co ...

seems reasonable and easy enough...
2016-4-1
Use props
IBE刘彦鸣
New

China
Offline

4K  quality is absolutely terrible!
left:4K 25P D-log  right:DNG


4K 25P D-log
2016-4-2
Use props
atreidex
lvl.1

Offline

GraemeB Posted at 2016-3-30 08:18
It's either you don't have an understanding of what is happening here, or haven't bothered to take ...

I had the same results, and answers online or from DJI have not been easy to come by.  Mainly I wanted to know if it were just a bad unit, or something playing out across the whole P4 line.  

I've been going back and forth with DJI for 2 weeks about a fix or getting an RMA to exchange for another copy... but now I'm thinking an exchange wouldn't change anything.  

DLog is pretty much un-usable for me.  The details are complete mush, if you try to turn up sharpening, it's noisy as hell.  Best footage I can get out of the thing so far is to use 'none' for color profile, turn up sharpening to 1, and contrast down -1.  Kinda defeats the point of wanting to shoot quality footage you can grade yourself.

Here's hoping something changes, I was happier with my P3P footage (now sold).  The flight and quality of the P4 is a step forward, but video quality is a step backwards.
2016-4-4
Use props
Geebax
First Officer
Australia
Offline

I don't believe this for a moment. Every bit of footage I have seen from a P4 looks the same as my P3, possibly better due to the better lens. As for shooting dlog, not much point unless you know how to post-process it, and 90% of people don't have a clue how to.
2016-4-4
Use props
azichek
lvl.1

Canada
Offline

I am noticing the same issue and have been going back a forth with a DJI support rep for the last 3 weeks. Essentially I am having the same two issues:
1. The horrible bitrate and quality on the 1080p @120fps
2. The fact that the p4 writes the 120fps videos as 30fps

Here is a link to a video I shot 1080p @120fps. Notice the 2001 cell phone video quality. Also notice that 3/4 of the video is not viewable as the file was written as 30fps not 120.
https://drive.google.com/file/d/ ... TA/view?usp=sharing

As for bit rate, here is a shot of video I shot at 2.7K. The bit rate is ~ 45Mbit/s. To me that seems normal.
http://imgur.com/m8VJcrt

Here is a pic of the 1080p @120fps video. The bit rate is horrible at 5Mbit/s.
http://imgur.com/GYAbym2
DJI can you please either explain why you designed it this way or confirm that this is an issue that can be resolved?
2016-6-27
Use props
johannes
lvl.4
Flight distance : 771434 ft
Iceland
Offline

atreidex@gmail. Posted at 2016-4-4 22:11
I had the same results, and answers online or from DJI have not been easy to come by.  Mainly I wa ...

D-log will give you grate image, you will have more control over details, color and dinimic range in the image.
But you need to know how to work with it in Resolve or another NLE. In the end if done correctly you will have sharp image with much better colors and details in black, white and shadows.
Also if you use LUT it will only give you a quick starting point to work from.
2016-6-29
Use props
OmegaStageThr33
lvl.1

United States
Offline

It would be nice if someone from DJI could comment on this issue.
2016-6-30
Use props
Westside Osprey
lvl.3
Flight distance : 95915 ft
United States
Offline

Really guys, this is a $300 cam including lens, and you are recording to a microSD card.

If you need RAW quality you need to buy a professional camera. The one on the Inspire RAW is pretty good but it records to an SSD.

There are serious bottle necks in a cheap cam/SD recording system.

IMHO for the $$$ spent on the P3P or P4 they are a fantastic bargain.

I often use my P3P in broadcast for television footage. But only as quick B-roll. For that it is fantastic bargain.

Bear in mind that a decent Procam (camera only, no lens) is a minimum of $5K. I am thinking FS5. Mostly I shoot with FS7 about a $8K camera, but I can intercut the P3P footage.

But at these prices you just accept compromise. The biggest one is the recording media, that is what keeps the data rates so low. My FS7 records 4K at 600mb/s, that is 10 times the rate of the Phantom cams. There are some complaining about the FS7 images as well. Basically you get what you pay for. A real RAW 4K setup is well over $10K or even $100K for top of the line. And of course those are all in a different class than a Phantom $300.00 cam. Just a good Pro lens costs as much a few P4's.
2016-6-30
Use props
virtualimagedes
lvl.3
Flight distance : 364826 ft
  • >>>
United States
Offline

I have no quality issue at all and most use it like myself because when slowing down it is much smoother and looks fantastic for making a all around good video. I shoot it because most don't have 4k televisions or computer screens. It also takes up way more space on the sd card and when it comes to editing its much faster to work with. You just need Final Cut Pro X or some other type of good editing software. Make sure your using a class 10 sd card. When you upload to FCP you can just select 1080P @ 24 FPS and still get that super slow motion smooth effect. If you shoot in a low FPS it just looks horrible.  
2016-6-30
Use props
12Next >
Advanced
You need to log in before you can reply Login | Register now

Credit Rules