DLOG
- Thread starter Tom K
- Start date
Big difference there obviously.
I'm yet to shoot DLOG. I always do some degree of editing to my footage - but only minor stuff really. I use the budget priced Cyberlink Power Director 15 Live suite on the PC to edit. Colour Director is included in the suite. Do you think it would take much work with this software to get DLOG footage looking at least as good as that shot on standard settings?
Sent from my iPhone using Tapatalk
I'm yet to shoot DLOG. I always do some degree of editing to my footage - but only minor stuff really. I use the budget priced Cyberlink Power Director 15 Live suite on the PC to edit. Colour Director is included in the suite. Do you think it would take much work with this software to get DLOG footage looking at least as good as that shot on standard settings?
Sent from my iPhone using Tapatalk
If you wanna try something new, free, and software I know will work nicely with your log footage you will be testing. Download Davinci Resolve. If you are more comfortable editing in PD, then just use Davinci to color grade and correct, then use PD to edit the footage afterward.
Not very many ppl use Power Director so I'm not sure if anyone will be able to answer your question if it will work nicely for grading. Only thing to do is test it yourself
.
I I did not put a lot of time into color correcting this shot. Don't know colour director. One of the nice things about Davinvi is the ability to pick a color ( the sky in my shot) and change hue/saturation/ contrast. Pretty cool to be able to do that
I'm uploading a video that show some banding problems with D-log -- will let everyone know when it's available in 4K. Also, since YouTube has a way of smearing things even more the banding may not be as evident on YouTube so I'll provide a link for my Google Drive so you can see the rendered video as I made it and the raw file from which it was made. Again, I'll let everyone know when they're available.
Brian
Brian
A extreme curve (focused on one small luminosity band) will magnify any banding which will then be visible even after processing by YouTube.
I have posted a video on YouTube where I take a 1:28 clip and copy it three times -- the first copy is left unaltered and is the raw video with not color grading or alterations whatsoever. The second copy I apply a modest amount of contrast stretching and a little sharpening and the banding is even more noticeable. The third clip I do even more contrast stretching but even this version is not extreme. The banding is visible in the raw footage.
I uploaded this a couple hours ago but it still doesn't have the 4K version -- so far only up to 1440P.
UPDATE: Video now available in 4K.
Also, see these files for the original raw file and the rendered video before YouTube upload.
DJI_0009.MP4
P4P Banding Problem.mp4
Brian
Last edited:
Interesting, Raptorman. I'll have to test it more myself, but at the end of the day, you have to be very careful with LOG curves on an 8-bit recording like h.265/h.264. Basically, it's a tool -- you want to use it when you need to preserve more highlight and shadow detail simultaneously than you can with in a non-log profile. But if the dynamic range of your scene fits just fine in dcinelike or none, then stick with that, you won't have to stretch the contrast in post as much and should see less banding.
Incidentally, I did some dlog shooting at dusk a few days ago and noticed that, compared to None/dcinelike, the DJI Go app reports the same exact exposure as 2 stops brighter in DLOG now. In other words, if left in its default auto exposure mode, the P4P will underexpose DLOG by 2 stops (presumably to preserve more highlight detail) and lift the shadows so they still look normally exposed (or even brighter than they might in none/dcinelike).
That's exactly what you want DLOG to be doing if your goal is to preserve the largest possible dynamic range. Unfortunately, it also can produce very noisy shadow areas, which is exactly what happened to the footage I shot. But it also suggests that the base ISO when shooting SLOG isn't really ISO 100, it's more like ISO 400 (in other words, ISO 100 in DLOG is as noisy as ISO 400 normally).
I've uploaded an unedited sample clip to youtube to show what I mean. This was shot at ISO 100, 1/60, F5.6 just before sunset. h.264. Check out the noise/mushiness in the trees.
My thinking is that--just with SLOG on the Sony cameras--DLOG on the P4P should only really be used if you have enough light to shoot at base ISO (in this case ISO 100). If you have to shoot at higher ISOs, DLOG footage is going to come out very noisy.
Some of the DLOG shots I took later where I was forced to bump up the ISO to 400 to compensate for the falling light definitely look much noisier then I expected ISO 400 footage to look. Here's an example:
Incidentally, I did some dlog shooting at dusk a few days ago and noticed that, compared to None/dcinelike, the DJI Go app reports the same exact exposure as 2 stops brighter in DLOG now. In other words, if left in its default auto exposure mode, the P4P will underexpose DLOG by 2 stops (presumably to preserve more highlight detail) and lift the shadows so they still look normally exposed (or even brighter than they might in none/dcinelike).
That's exactly what you want DLOG to be doing if your goal is to preserve the largest possible dynamic range. Unfortunately, it also can produce very noisy shadow areas, which is exactly what happened to the footage I shot. But it also suggests that the base ISO when shooting SLOG isn't really ISO 100, it's more like ISO 400 (in other words, ISO 100 in DLOG is as noisy as ISO 400 normally).
I've uploaded an unedited sample clip to youtube to show what I mean. This was shot at ISO 100, 1/60, F5.6 just before sunset. h.264. Check out the noise/mushiness in the trees.
My thinking is that--just with SLOG on the Sony cameras--DLOG on the P4P should only really be used if you have enough light to shoot at base ISO (in this case ISO 100). If you have to shoot at higher ISOs, DLOG footage is going to come out very noisy.
Some of the DLOG shots I took later where I was forced to bump up the ISO to 400 to compensate for the falling light definitely look much noisier then I expected ISO 400 footage to look. Here's an example:
Last edited:
Yes, 8 bit is going to make it difficult to avoid banding if the conditions are right and D-log only makes that problem worse by providing fewer bits to code certain tonal regions. I can see how D-log can work if the conditions are right but there are a number of people much smarter than I that argue 8 bit and log isn't going to work most of the time.
One of the problems here is that you can't so easily test a drone as there are many places where they're not permitted and DJI isn't keeping things constant -- we've had two firmware updates in 5 weeks and both of them have altered the color profiles and done so in a way the DJI has not detailed in the slightest. It's like trying to heard cats!
Brian
One of the problems here is that you can't so easily test a drone as there are many places where they're not permitted and DJI isn't keeping things constant -- we've had two firmware updates in 5 weeks and both of them have altered the color profiles and done so in a way the DJI has not detailed in the slightest. It's like trying to heard cats!
Brian
I am so far excited about the new dlog. I am still yet to try out dlog with direct sunlight to light my scene however. This was taken just after the sun dropped below the mountains but still was hitting a few of the peaks. I used dlog with -1/0/0 picture settings and just touched it up in post. No sharpening was added. Shot in 4K h.265, transcoded to prores, then back into h.264 for youtube upload. Has anyone tried uploading straight prores to youtube? I know I have on Vimeo before.
- Joined
- Mar 20, 2016
- Messages
- 440
- Reaction score
- 163
- Age
- 35
I use d-log and ETTR since d-log pushes EV 0.7-1 up. I then push the material down to compensate for the lift in shadows and midtones that d-log does.
Since the grading occurs after the material has been transcoded to Prores, my goal with this is that Prores handles the new shadow levels better then h265 would if recorded straight from camera. The benefits as I see them is also that it's easier to recover bright highlights (unless blown) than recovering crushed shadows.
So ETTR while avoiding blown highlights gives me a good starting point when grading. This is my current opinion.
I would love to know the actual gamma-curve applied in d-log.
Since the grading occurs after the material has been transcoded to Prores, my goal with this is that Prores handles the new shadow levels better then h265 would if recorded straight from camera. The benefits as I see them is also that it's easier to recover bright highlights (unless blown) than recovering crushed shadows.
So ETTR while avoiding blown highlights gives me a good starting point when grading. This is my current opinion.
I would love to know the actual gamma-curve applied in d-log.
Sky looks pretty significantly blown and with very noticeable banding -- it may be that the new D-log is going to be a real challenge to avoid banding.
Brian
I took a quick look at the raw material, and I agree that banding is certainly a issue here.
I notice that it is very low contrast and also rather hazy and for these conditions, none ettr will most certainly give better working conditions.
The reallocation of tones to shadow areas are a good use of tones i a high-contrast scenario, but for low contrast you end wasting all that low end of the spectrum.
For now, I think I will stick with two modes:
* For high contrast, dlog -1 /0 / 0 (sunset / sunrise / sunny day with clear skies or low cloud-coverage)
* For low contrast, none -1 / 0 / 0 (typically overcast or hazy / foggy conditions)
This is only a teaser, but I captured this rather demanding scene today and will do a edit tomorrow. I shot mainly in dlog but included a bit of none for comparison. None had some minor clipping, but we will see how they both do. Footage not reviewed yet so I am as interested as you.
Disclaimer: preview image not shot with drone.
Disclaimer: preview image not shot with drone.
Similar threads
