DLOG

Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
Here is (finally) one single clip that is not (close to) a comprehensible test, but at least somewhat realistic.

Conclusion: For high-contrast scenes, dlog will give you much more latitude for post-processing and will in most cases be preferable when you plan to do toning. But: You can get away with none as-well.

Additional finding: Youtube take ages (read two hours) to encode 4k, even for 23 seconds of content..

I have included both ungraded and graded segments for none and dlog for a comparable clip (shot seconds apart).

Contrast caused very minor clipping for whites, so this is right on the edge of what is recordable unless you are prepared to loose highlights or shadows (therefore a rather good test).

I have tried my best to tone both recordings the same, but this proved impossible (in reasonable time); so this is a close-ish estimate but definitely not perfect.

However, we still find that dlog has a lots of definition in the shadow areas, while we can see banding here for none. The same effect applies when we apply saturation or contrast, and dlog is much easier to force into submission than none (again, the low resolution of shadows and low-medium tones for none causes issues).

An interesting side-effect is that in the shadow areas, we have more noise for dlog than none, but at the same time more detail/definition. This is probably caused by the banding of none which essentially hides both noise and definition. See note at the end.

Given this sky that actually has some light clouds and not a single, uniform tone, I cannot identify any banding even for dlog in the hightlights.

Overall I will say dlog wins, for high contrast at least.

I do expect (very) low contrast to be best captured with none in combination with strict expose-to-the-right.

Final note: The tone applied here is selected to expand into full usable range and high saturation to exaggerate any artifacts caused by toning.

RGB parade in raw material (dlog to the right):

upload_2017-1-22_21-4-6.png


RGB parade after toning (dlog to the right)

upload_2017-1-22_21-4-37.png


And here is the clip:



Note at the end: Youtube has introduced additional banding in shadow area also for the graded dlog, so here is a 100% crop showing the difference better.

upload_2017-1-22_21-29-15.png
 
Joined
Dec 13, 2015
Messages
54
Reaction score
47
Age
39
Thx for the test. A lot more info in the shadows for DLog. The problem with None would have been even more visible if you tried to pushed those shadows up.
For me DLog (-1,0,0) seems a good choice. I don't think there is need with DLog to lower the contrast, or even the saturation (I always pushed it back in post either way)
 
  • Like
Reactions: Tomas Wangen
Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
Thx for the test. A lot more info in the shadows for DLog. The problem with None would have been even more visible if you tried to pushed those shadows up.
For me DLog (-1,0,0) seems a good choice. I don't think there is need with DLog to lower the contrast, or even the saturation (I always pushed it back in post either way)
Yes, it seems I forgot to mention (-1, 0, 0) was used in the test.
 
Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
So, the test last week did not attract much attention :)

Still, I did finalize a two minute edit (from dlog h265) in case anyone is interested how the footage holds up in post.


I would recommend seeing it in 4k.

Youtube looses a lot during the transcoding, but on a big screen the original export is quite nice in 100mbit.
 
Last edited:
Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
Bringing a dormant thread back to life.

I just was hit with the limitation of dlog on a very low contrast scene. I got comparable results (for dlog) to what @Raptorman0909 has shown earlier, very visible banding in the sky.

This is a 50% resolution grab of the same scene shot a half hour after sunset (therefore the low contrast) of both dlog video and raw image. The raw capture indicates what I could have been able to capture had I remembered to switch to none (with ettr). Raw and dlog similarly toned (as close as I could get in reasonable time).

Note that my captures earlier in the evening (when the sun was up) does not have these issues, so again, this is just a (repeated) warning to be careful with dlog in low-contrast settings.

Also note that this was not a planned test, rather a accident with documented effects.

dlog source:
sample_dlog_source.jpg


dlog toned:
sample_dlog.jpg


raw comparison:
sample_raw.jpg
 
Last edited:
Joined
Dec 13, 2015
Messages
54
Reaction score
47
Age
39
Nice view. Not cool for the DLog (hope the next Phantom will shoot at 10bits). Can you post the original DLog from what you started ? Don't know if you also ETTR in DLog but going one stop down may be safer to avoid banding. Still need to test that even I'm not sure. I think I'll stay in DCinelike. Less Dynamic but safer and easier to colorgrade for now.
 
  • Like
Reactions: Tomas Wangen
Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
Nice view. Not cool for the DLog (hope the next Phantom will shoot at 10bits). Can you post the original DLog from what you started ? Don't know if you also ETTR in DLog but going one stop down may be safer to avoid banding. Still need to test that even I'm not sure. I think I'll stay in DCinelike. Less Dynamic but safer and easier to colorgrade for now.
I have added the untoned dlog which is not shot ETTR. As you can see, the problem is the wasted/unused tones in the highlights and shadow areas of the histogram which ends up as banding when corrected (due to 8 bit total range). Dlog is fine (and I would say normally preferable) for highcontrast scenes, but I would advise against using it for scenarios such as this.
 
Joined
Mar 29, 2016
Messages
1,727
Reaction score
504
Age
60
So, the test last week did not attract much attention :)

Still, I did finalize a two minute edit (from dlog h265) in case anyone is interested how the footage holds up in post.


I would recommend seeing it in 4k.

Youtube looses a lot during the transcoding, but on a big screen the original export is quite nice in 100mbit.
Very nice flying and video man!
 
  • Like
Reactions: Tomas Wangen
Joined
Aug 11, 2014
Messages
62
Reaction score
19
This is manually calibrated using a Gossen Digisky lightmeter for the greypoints of capture. Profile is created/saved using davinci resolve after curve was created.
Hm ok
The curve looks like one would expect, but would you mind explaining how a light meter can be used to measure this?
I was looking more at 3D LUT creator and color checkers..

Sent from my ONEPLUS A3003 using PhantomPilots mobile app
 
Joined
Dec 4, 2016
Messages
563
Reaction score
310
Age
47
Hm ok
The curve looks like one would expect, but would you mind explaining how a light meter can be used to measure this?
I was looking more at 3D LUT creator and color checkers..

Sent from my ONEPLUS A3003 using PhantomPilots mobile app
Greyscale color checkers measured with static lightsources and reflected light. This is manually correlated to the destination greyscale referred in the other thread. Accuracy is not 100% but a good indication. (The 3d lut is the result of the curve from the manual correlation)
 

Members online

Forum statistics

Threads
142,157
Messages
1,461,105
Members
103,946
Latest member
xr2