Color mode & Dynamic Range

I got the P4P, so the new update from today (01.01.0203) is also for it.

blade_strike from rcgroups just said they will add Art mode back soon.

For the other modes, no idea what they are and if they have changed the D-Log mode. Will do some more tests tomorrow (it's 1am here so not the best time for this) after testing how the drone flies (I didn't even started the motors yet)


And I thought the French never sleep! :D

BTW, glad to have another looking into this and with the additional profiles it may take a team effort to get a handle on this. I have to believe DJI will release a manual or guidance or at least update the user manual to cover these new profiles and give us an idea what there inteded to do.


Brian
 
Well someone at least please test Dlog after the update, I wonder if they've made any improvement.
 
Well someone at least please test Dlog after the update, I wonder if they've made any improvement.

If they've done a major change to the way the handle color profiles then one can only hop D-Log is fixed or at least usable -- all that will indeed need to be tested. I just today received my replacement P4P (bad gimbal on first one) and after charging the batteries I'll do the firmware update and see where we're at.

And then it's starting from scratch to test things out.


Brian
 
Well, I suppose change is good. Maybe they've implemented D-LOG to be more than just a contrast adjustment. Fingers crossed. Someone please add some high ISO testing e.g. 800 and 1600.

We can already see that the Sony sensor is pretty impressive with regard to dynamic range even when DJI doesn't make the most of it with gamma curves (hopefully the new firmware makes improvements here). And I am hoping (really hoping) they have looked for better ways to limit the noise in low light, high ISO conditions. The RX100IV is pretty amazing in low light. Hopefully we get the same in the P4P. Fingers and toes crossed!
 
  • Like
Reactions: tbarnesarc
Tom, and others, have done some very enlightening experiments. What Tom has demonstrated is clearly the dlog setting is broken, and is not doing what is supposed to do. There should be no clipping lines either in the whites and blacks, and there is. In addition to not fixing the clipping, it went ahead and expanded the dynamic range of the recorded image, leaving the actual subject scrunched down to fit in the enhanced dynamic range. It looks like a software error where the programmer put the range control in the pipeline after the initial digitizations, rather than before, where it would have worked as specified.

The other explanation is that the sensor itself isn't capable of delivering the range needed. The sensor problem can't be fixed without another sensor. The software issue, if it is that, can easily be fixed by debugging the dlog programming.

My recommendation still is to use one of the lower contrast settings that actually works to avoid clipping, and then adjust to taste in post. All the non-linear editors have far better tools for adjusting the image to taste, and Davinci Resolve supports your customized look in a lut that you can reuse on your new footage, and you can clone and modify for new situations. A look that works well at sunset and sunrise generally looks awful used on footage shot at noon, and vice versa. Shooting at night, with a heavily boosted ISO will need a different corrective look than the earlier two examples. All the active cinematographers I know have a custom collection of lute hidden in their bag of tricks, ready to be pulled out and used when the need arises. Some of the experimental looks being developed in this discussion could probably be marketed.if they met the needs of others.

This forum is a good example of the good that comes from a committed collaboration of serious cinematographers.


Sent from my iPad using PhantomPilots
 
Yeah, David, one can only hope they pull there head out and get D-Log to work as it should. But, not that the new update has added about a dozen new profiles and there's no information available to indicate what they are and what they're for it puts us in the position of having to test this all once again. Doing the necessary tests for a dozen new and half a dozen old profiles is time consuming.


Brian
 
Just tried the new modes :
- D-Log is as bad as before (clipped highlight at around 70%)
- Film A,..., Film I, are the old Vivid, B&W, Beach, ... profiles maybe with new ones but they are more like instagram effects (can't say the order but it's just a label problem in the app I think)
- No more Art profile so D-Cinelike is the only usable flat profile left
- Truecolor profile is kind of a more contrasty and darker profile than None

So for today, D-Cinelike (-1,0,0) will be the way to go for me.
 
as always needs the same old cents to make a dollar for dji...but i understand that this poor implementation is make to preserve sells of more expensive copter, the inspire 2.
all in all from a cine point of view i still prefer what i can obtain from the go pro pro res than any camera and video more i see from dji.
 
I tested four color modes to compare the dynamic range. In each case, I exposed to try to protect highlights/shadow details and avoid zebras triggering on the clouds/white sections of the ship:

View attachment 70007 View attachment 70008 View attachment 70009 View attachment 70011

I don't see much dynamic range difference between them. The Dcinelike profile seems to preserve a bit more highlight detail, but that's partially because the exposure is 2/3 of a stop lower. Art is quite similar to Dcinelike. Dlog is... well, very compressed for my tastes. And not clearly retaining more highlight or shadow detail.

I then went ahead and graded them (just a contrast boost) to try to fill up more of the waveform without losing any details. Here are the results:

View attachment 70012 View attachment 70013 View attachment 70014 View attachment 70015

My conclusions:

* Dlog preserves less detail in the highlights than the other modes! (check out the clouds in the upper right corner). This might be because I overexposed the dlog shot (even though in the histogram/waveform it doesn't look overexposed -- but note the strange clipping at IRE 65 on the right hand side of the vectorscope. Perhaps the histogram in DJI Go 4 is not accurate when dealing with dlog.
* Dcinelike and Art have similar dynamic range, but Art is less saturated and may have a slight color shift. In any event, I don't see much point in experimenting further with Art.
* As between None and Dcinelike; Dcinelike appears to preserve a bit more detail in the clouds. Could be because it's 2/3 of a stop underexposed as compared to None. Despite being 2/3 of a stop lower exposure than None, Dcinelike seems to lift the shadows, so the underexposure isn't creating any shadow/black clipping. Still, as between the two, I think even after grading None looks nicer/more contrasty/has more pop.
* I need to test None vs. Dcinelike vs Dlog all 1 stop underexposed (and at the same exposure) to reach a definite conclusion as to highlight/shadow retention.
What application are you grading in?
 
One great review and comparing of Phantom 4 Pro and Inspire 2 X5S.


In short: the P4P has got over 11 stops of dynamic range and totally crushes the P4, the X3 and all other drones in the same price league, even more expensive ones.

I'd be really suprised if we see a P5 in March. I think we'll see it next October 2017, and the P5 Pro in March 2018. This means strong Buy for the P4P.
 
There is a lot of confusion on this topic, which is going to keep us spinning our wheels and not capturing beautiful video. The Phantom 4 Pro Camera has much better specs than any of the earlier Phantoms:

Phantom 4 Pro and Mavic Camera specs :

CAMERA
Sensor 1’’ CMOS; Effective pixels: 20M
Lens FOV (Field of View) 84°, 8.8 mm / 24 mm (35 mm format equivalent),
f/2.8 - f/11.
auto focus at 1 m - ∞

Phantom 4 and Phantom 3 4k and Phantom 3 Pro

CAMERA
Sensor 1/2.3” (CMOS), Effective pixels:12.4 M
Lens FOV 94° 20 mm (35 mm format equivalent) f/2.8, focus at ∞

Several hugely important differences in the cameras;

1. The new sensor 66.6% more pixels - The captured image has automatically 66.6% more starting details.

2. The sensor, in addition to more pixels is 150 % bigger which will improve the noise performance

3. A new lens is being used, which is presumably sharper. DJI has published the MTF specs for it, but has never, to my knowledge, published the MTF specs for the older camera.

It is extremely important to note that this sensor/lens combination is the starting point for all DJI's settings ("none", "cinema", etc). No amount of adjustment of any of these will make any difference to the P3 K, P3 Pro, or P4 camera. The quality is totally dependent on the camera. As I understand what DJI is doing, the only way to get the higher performance camera is to buy it with the Mavic or P4 Pro. To my knowledge, there is no way to retrofit the new camera on any of the drones that don't have it.

Being the owner of a P34k which I love, I intend to continue flying it and will replace it only when something really bad happens. It has got 104 flights on its now. The picture it generates looks gorgeous on my 4K Sony TV. All the image imperfections that are bothersome result from Pilot Error - eg pointing the camera too close to directly into the sun, and not using the cinema settings to avoid clipping. Every shot gets individual TLC transcoding it to ProRes and color correcting in Resolve 12.5. My personal tests don't show any value in using any of the other presets over Resolve and it's powerful tools.

Get out and fly and see what works best in the real world.
 
There is a lot of confusion on this topic, which is going to keep us spinning our wheels and not capturing beautiful video. The Phantom 4 Pro Camera has much better specs than any of the earlier Phantoms:

Phantom 4 Pro and Mavic Camera specs :

CAMERA
Sensor 1’’ CMOS; Effective pixels: 20M
Lens FOV (Field of View) 84°, 8.8 mm / 24 mm (35 mm format equivalent),
f/2.8 - f/11.
auto focus at 1 m - ∞

Phantom 4 and Phantom 3 4k and Phantom 3 Pro

CAMERA
Sensor 1/2.3” (CMOS), Effective pixels:12.4 M
Lens FOV 94° 20 mm (35 mm format equivalent) f/2.8, focus at ∞

Several hugely important differences in the cameras;

1. The new sensor 66.6% more pixels - The captured image has automatically 66.6% more starting details.

2. The sensor, in addition to more pixels is 150 % bigger which will improve the noise performance

3. A new lens is being used, which is presumably sharper. DJI has published the MTF specs for it, but has never, to my knowledge, published the MTF specs for the older camera.

It is extremely important to note that this sensor/lens combination is the starting point for all DJI's settings ("none", "cinema", etc). No amount of adjustment of any of these will make any difference to the P3 K, P3 Pro, or P4 camera. The quality is totally dependent on the camera. As I understand what DJI is doing, the only way to get the higher performance camera is to buy it with the Mavic or P4 Pro. To my knowledge, there is no way to retrofit the new camera on any of the drones that don't have it.

Being the owner of a P34k which I love, I intend to continue flying it and will replace it only when something really bad happens. It has got 104 flights on its now. The picture it generates looks gorgeous on my 4K Sony TV. All the image imperfections that are bothersome result from Pilot Error - eg pointing the camera too close to directly into the sun, and not using the cinema settings to avoid clipping. Every shot gets individual TLC transcoding it to ProRes and color correcting in Resolve 12.5. My personal tests don't show any value in using any of the other presets over Resolve and it's powerful tools.

Get out and fly and see what works best in the real world.
The Mavic does not have anything in common with the P4P camera. It is basically the same as the P4 camera, with a 78 degree FOV.
 
We can say that independent of exactly how wide the captured dynamic range is, we do manage to reproduce absolute black and absolute white from recorded data in post production. So aside from a camera maximizing the width and detail shades of the dynamic range, it will never be totally satisfactory to represent reality. By default the dynamic range is captured linear, but ultimately restricted to 8 bits - or 256 shades. For example, if I want to focus on a lot of shadow details, I need to overemphasize the dark parts of the image by loosing the brightest parts (overexpose). The same is true the other way around. On the other hand, the different (D) modes effectively act like a lens focusing (like zooming in) on a part of the "real/linear" dynamic range to be photographed. D-modes are non-linear capturing modes that effectively distort (stretch some areas of the dynamic range and compress others) still fitting the captured shades it into 8 bits. Without post production (applying a corrective LUT and grading), the recorded video is inherently bad or at a minimum not representative of what had been photographed. However, by using a LUT, it is possible to expand the dynamic range into a wider spectrum than would have been recorded in linear mode.

One of the real issues with any form of compression is the amount and exactly what data to discard to produce the recorded media. The algorithms and codec parameters used in the drone(s) requires a critical compromise between the amount of input data, speed of recording (fps) and available processor power. From observations and comparisons, it is clear that the encoder is rather simplistic when it comes to shaving information. For example, reducing sharpness, contrast and/or saturation causes the magnitudes of significant picture information to be reduced to a point where it drops below some arbitrary threshold in the encoder causing it to be irretrievably lost. I found it is better to "stress" the encoder by not reducing input data, instead possibly increasing input data (such as landscape style (+1,+1,0) and use D-Cinelike or D-Log. It is easy to reduce sharpness in post production if so desired, the reverse, attempting to retrieve it does not really work - it will simply tend to enhance mpeg artifacts.

From what I've read here it does seem odd how the P4P modes work. Looks like DJI will be busy upgrading firmware for quite a while. As far as the older cams such as P3P are concerned, there has been a change with the latest firmware upgrade affecting the camera and modes. But clearly, the (D) modes on the P3P are working exactly as they should.

Roamer, what you've described is how the Dlog and Dcinelike modes should work. But as you can see from all the testing presented on this thread, that's not how they actually work as currently implemented on the P4P. Dlog and Dcinelike do not stretch out any areas of the dynamic range. They take the exact same range as None and linearly? compress it, leading to the same information being recorded in a smaller data space and thus a loss of fine gradation in the mid-tones.

EDIT: I see now that there's a firmware update which may impact this. I'm currently downloading it and will do some testing..
 
Last edited:

Recent Posts

Members online

No members online now.

Forum statistics

Threads
143,096
Messages
1,467,620
Members
104,981
Latest member
brianklenhart