Color mode & Dynamic Range

From various threads I've read it seems that Dlog worked reasonably well at some point in the past until it was broken in a firmware update, perhaps in all DJI products. I don't think anyone knows if this was a regression error or some attempt at improvement that went wrong, but probably all we can do now is hope that DJI takes note of all the excellent work done by users to document the problem and that the situation improves in a future firmware release.
 
  • Like
Reactions: Tomas Wangen
From a video found on youtube shot in DLog 0,-2,-2 with the P4Pro

It's bad, blown highlights at 65 :(

source here (not mine) : youtube video
original videos here : P4P - Google Drive

a9585703-174-dlog_0_-2_-2.jpg
Astonishing how some people can't see the difference between good log footage, like from an Arri Alexa, and absolute **** footage like this. Do never think that your 2/3 or 1-inch sensor with a ~90 dollar lens that records in h264, has the same capabilities as a tens of thousands dollar camera that records 4444 ProRes. Do they think that because something looks gray and barely has colors, it's good?

Such high quality footage looking a bit pale before applying a LUT has nothing to do with what's being recorded, it is only the media which has not gotten its correct gamma curve yet.
 
  • Like
Reactions: hidaven
That was just an exemple of completly blown highlights to show where the max value was recorded. Next time, think twice before being rude ;). And yes, it look like crap, even after grading
 
  • Like
Reactions: Dronefriend
That was just an exemple of completly blown highlights to show where the max value was recorded. Next time, think twice before being rude ;). And yes, it look like crap, even after grading
I didn't mean it towards you! Sorry about that. It was meant to the people who shoot footage like this.
 
Just wondering what was wrong with the gimbal on your P4 Pro Brian?

Sorry for the late response ...

After my first flight I changed the battery and attempted to launch for a second flight, but the gimbal would not complete the homing sequence on startup. Normally the camera should tilt up to about 30 degrees then tilt down to about -120 degrees then back to zero degrees and homing is complete, but in my case it would tilt up to 30 degrees then it would hang at about -105 degrees and not be able to reach the down limit and therefore not complete the homing sequence. I had to pack up at that point and go home. When I got home it worked just fine and I went out again a couple days later and once again I was able to fly the first flight without a problem but had the homing problem when trying to launch for the second flight. I powered off and moved by hand and I felt the camera/gimbal snag at the point where the homing hung up, about -105 degrees. I looked a bit closer and it looks like the ribbon cable is snagging on something so I felt it necessary to send it back rather than have it fail in-flight. I was tempted to tear it apart, but that would have risked the warranty so...


Brian
 
  • Like
Reactions: tml4191
Just my two cents on this one but lack of DR only seems to affect the H.265 codec on those clips that were shared on here. Agreed log hasn't been good for a while now but I've noticed two things; one is only H.265 clips are clamped down/ maxed out on lower value ire values. The other is H.265 files won't even show up in the ingest area of Resolve yet they import fine into PP with no problem. I'm on a MBP not on windows though. Maybe that will help some of you narrow down the issue or not but that's what puzzled me.
 
Last edited:
Hey guys -- sorry about the radio silence. I'm currently out-to-sea in a pretty remote part of Indonesia for the next two weeks, so I doubt I'll have much chance to do a proper follow-up until before Christmas. Too busy shooting actual footage (there's worse problems to have :D).

At any rate, I did some filming around sunset tonight and noticed that the none picture profile (with -2/0/0) is a little bit disappointing in the amount of dynamic range it retains. I didn't test it side by side vs. a phantom 4 or any other picture mode -- just my observations. At sunrise tomorrow I'll try shooting a bit in Dcinelike, which from my previous testing appears to lift the shadows slightly. Whether it'll make a significant difference in real use remains to be seen.

Really disappointed that DJI doesn't currently implement a real log gamma profile, as I feel that the sensor is able to discern a wider dynamic range than what ends up actually recorded. My other camera is a Sony A7RII, which has a very useful log gamma mode.

Anyway, here's a still from tonight to give you an idea of the lighting conditions I was dealing with (and how remote of a place I'm in :D)

20161208--11-HDR-Pano.jpg

For all the issues with the video from the P4P, I must admit I'm quite pleased so far with the quality of the stills it's producing.
 
  • Like
Reactions: blaster81
I didn't mean it towards you! Sorry about that. It was meant to the people who shoot footage like this.
Just my two cents on this one but lack of DR only seems to affect the H.265 codec on those clips that were shared on here. Agreed log hasn't been good for a while now but I've noticed two things; one is only H.265 clips are clamped down/ maxed out on lower value ire values. The other is H.265 files won't even show up in the ingest area of Resolve yet they import fine into PP with no problem. I'm on a MBP not on windows though. Maybe that will help some of you narrow down the issue or not but that's what puzzled me.
What clips are suffering from low DR due to h265? It's bevause of the settings used, not the codec afaik
 
What clips are suffering from low DR due to h265? It's bevause of the settings used, not the codec afaik

DJI_0028.MP4 & DJI_0029.MP4. Those clips were shot in H.265. When you bring them into PP they clip way below 100 on the scopes (as others have stated) essentially limiting the amount you can stretch those files to their max unlike the other H.264 files. When I brought the rest of the clips into PP and applied an S curve to them they behaved as normal in terms of the scopes not clipping lower than they should. It would seem that it actually could be a codec/ compression issue in conjunction with a crappy Log profile. Who knows though. I was just trying to help.
 
DJI_0028.MP4 & DJI_0029.MP4. Those clips were shot in H.265. When you bring them into PP they clip way below 100 on the scopes (as others have stated) essentially limiting the amount you can stretch those files to their max unlike the other H.264 files. When I brought the rest of the clips into PP and applied an S curve to them they behaved as normal in terms of the scopes not clipping lower than they should. It would seem that it actually could be a codec/ compression issue in conjunction with a crappy Log profile. Who knows though. I was just trying to help.
No worries man Im just trying to pinpoint the problem. I think it was the dlog profile, but I may be wrong. I hope Im not :eek:
 
I think it was the dlog profile, but I may be wrong. I hope Im not :eek:

Tomorrow it is supposed to be sunny, and I plan to do a small but complete test comparing dlog/dcinelike and none for both h264 and h265 for hopefully the same conditions / dynamic range of scene. I have tried all already but not in a controlled fashion. I have already excluded h265 / dlog combo but will include it for reference.

I am 99% confident that all profiles individually will perform the same contrast wise regardless of codec (h264/h265), but I have been proven wrong before :)
 
Tomorrow it is supposed to be sunny, and I plan to do a small but complete test comparing dlog/dcinelike and none for both h264 and h265 for hopefully the same conditions / dynamic range of scene. I have tried all already but not in a controlled fashion. I have already excluded h265 / dlog combo but will include it for reference.

I am 99% confident that all profiles individually will perform the same contrast wise regardless of codec (h264/h265), but I have been proven wrong before :)

Awesome looking forward to it.
 
  • Like
Reactions: Dronefriend
Hey guys -- sorry about the radio silence. I'm currently out-to-sea in a pretty remote part of Indonesia for the next two weeks, so I doubt I'll have much chance to do a proper follow-up until before Christmas. Too busy shooting actual footage (there's worse problems to have :D).

At any rate, I did some filming around sunset tonight and noticed that the none picture profile (with -2/0/0) is a little bit disappointing in the amount of dynamic range it retains. I didn't test it side by side vs. a phantom 4 or any other picture mode -- just my observations. At sunrise tomorrow I'll try shooting a bit in Dcinelike, which from my previous testing appears to lift the shadows slightly. Whether it'll make a significant difference in real use remains to be seen.

Really disappointed that DJI doesn't currently implement a real log gamma profile, as I feel that the sensor is able to discern a wider dynamic range than what ends up actually recorded. My other camera is a Sony A7RII, which has a very useful log gamma mode.

Anyway, here's a still from tonight to give you an idea of the lighting conditions I was dealing with (and how remote of a place I'm in :D)

View attachment 70332

For all the issues with the video from the P4P, I must admit I'm quite pleased so far with the quality of the stills it's producing.


Andrei, I posted another thread on it but I have a video uploaded using D-Cinelike and -2, -2, -2 that is MUCH better than D-Log. The video is a bit long as I was attempting to provide a little guidance to folks new to dornes and videography. But, the quality of the video is much better than D-Log.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.


Brian
 
Andrei, I posted another thread on it but I have a video uploaded using D-Cinelike and -2, -2, -2 that is MUCH better than D-Log. The video is a bit long as I was attempting to provide a little guidance to folks new to dornes and videography. But, the quality of the video is much better than D-Log.
From what you've seen thus far, if one intends to apply color grading would you recommend shooting in Cinelike -2 -2- 2 or None 0 0 0 ?
 
From what you've seen thus far, if one intends to apply color grading would you recommend shooting in Cinelike -2 -2- 2 or None 0 0 0 ?


Still too soon to say what the best settings are but D-Cinelike may wind up the preferred choice at this point. As to the custom settings that to is still not for sure. I think you want to reduce sharpening in-camera and -2 seemed to work OK for me though more testing would help firm that up. I also think you want to turn down the saturation a bit as it seems to be a little too punchy at zero. The value I used was -2 and that appears to be a reasonable value as well. The last bit, contrast, is the tough one as DJI appears to have a hard time handling contrast so the value I used, -2, may or may not be the best value. It may be that zero for contrast is the best at least with current camera firmware.


Brian
 
Andrei, I posted another thread on it but I have a video uploaded using D-Cinelike and -2, -2, -2 that is MUCH better than D-Log. The video is a bit long as I was attempting to provide a little guidance to folks new to dornes and videography. But, the quality of the video is much better than D-Log.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.


Brian

I will take a look, Brian. But you may want to try shooting with Contrast at 0 rather then -2. Like Dlog, I've found that the -2 on contrast doesn't help at all.. it's just linearly squeezing the same amount of information into a smaller range of the waveform.

Here's 4 screen-grabs comparing None to Dcinelike at -2/0/0 at midday (with a polarizing filter).


Screen Shot 2016-12-09 at 10.49.58 AM.png Screen Shot 2016-12-09 at 10.50.01 AM.png Screen Shot 2016-12-09 at 10.50.19 AM.png Screen Shot 2016-12-09 at 10.50.06 AM.png


I'm not really convinced Dcinelike is preserving more dynamic range -- it seems to be lifting the shadows, but it's not clear it's preserving any more detail you couldn't also recover from None footage. On the other hand, None is a bit too contrasty for my tastes straight out of camera, so I would have to reduce contrast in post. More contrasty than it was on the Phantom 4.
 
  • Like
Reactions: blaster81 and Bigdz
The none is a fine choice for shooting in situations where there isn't too much contrast, and you have perfect exposure. Given that neither is reliably true, using the flatter settings minimizes the chances of clipping either the blacks or whites. Digital clipping is truly ugly. If neither black nor white is clipped, it is easy to restore a full dynamic range image in post by setting the black and white points appropriately just short of clipping. You are not taking advantage of the full range of the video DAC most of the time, but the insurance value of avoiding clipping makes it worthwhile anyway.


Sent from my iPad using PhantomPilots
 
I will take a look, Brian. But you may want to try shooting with Contrast at 0 rather then -2. Like Dlog, I've found that the -2 on contrast doesn't help at all.. it's just linearly squeezing the same amount of information into a smaller range of the waveform.

Here's 4 screen-grabs comparing None to Dcinelike at -2/0/0 at midday (with a polarizing filter).


View attachment 70378 View attachment 70379 View attachment 70381 View attachment 70380


I'm not really convinced Dcinelike is preserving more dynamic range -- it seems to be lifting the shadows, but it's not clear it's preserving any more detail you couldn't also recover from None footage. On the other hand, None is a bit too contrasty for my tastes straight out of camera, so I would have to reduce contrast in post. More contrasty than it was on the Phantom 4.


I'm not convinced D-Cinelike is the best choice at this point either and I also have questions about using a non zero value for contrast. The video I posted was recorded on the 4th -- the same day I did the exposure tests and actually the flights were done before the exposure tests. So, I have more testing to do and it would be better to have a more contrasty day than the overcast day on the 4th. But, as I mentioned I have had to return my P4P due to the gimbal problem but I hope to have a replacement tomorrow -- sadly, the weather looks to be quite spotty here in SLC area for a few days beyond that.

I think to better quantify things it is better to do stationary tests as I did with the exposure tests on the 4th. Controlling things so that the only thing that changes is the camera settings should make comparisons easier. As in the exposure tests I did you want a scene with high contrast and then lock the aperture and ISO and control exposure by shutter speed alone. Start with it at the over exposed end and increase shutter speed until you are clearly crushing the blacks. Then repeat with different color profiles and values for contrast adjustment. Pick the exposure in each test that is just below clipping the highlights and adjust for a consistent look for all the tests then look for variations in shadow and highlight detail.


Brian
 
The none is a fine choice for shooting in situations where there isn't too much contrast, and you have perfect exposure. Given that neither is reliably true, using the flatter settings minimizes the chances of clipping either the blacks or whites. Digital clipping is truly ugly. If neither black nor white is clipped, it is easy to restore a full dynamic range image in post by setting the black and white points appropriately just short of clipping. You are not taking advantage of the full range of the video DAC most of the time, but the insurance value of avoiding clipping makes it worthwhile anyway.


Sent from my iPad using PhantomPilots

That would be true if the "flatter" profiles as you describe them actually recorded more information in either the highlights or the shadows, David -- that is, if they actually recorded a wider dynamic range. However, testing suggests that isn't the case with the Phantom 4 Pro.. The flatter profiles like dlog and dcinelike aren't retaining more information, they're just squeezing the same information into a smaller portion of the 8-bit recorded signal. Same exposure, same extreme shadow and highlight detail, but squeezed between, for example, 10IRE and 90IRE instead of 0IRE and 100IRE.
 
  • Like
Reactions: Dronefriend

Members online

No members online now.

Forum statistics

Threads
143,066
Messages
1,467,358
Members
104,936
Latest member
hirehackers