GH5 Where's everyone at with V-Log vs a Rec709 Profile?

B

bengiles

Well-known member
I've gone fully GH - with a GH5 rigged for actuality and GH5S for interviews on legs. Both VLogL-enabled. Atomos Shinobi monitor with built in LUTs.

I've spent the past 3 years shooting exclusively SLog 2 and 3 on an FS7, but just wondering if I should question a few orthodoxies I've become perhaps overly comfortable with.

One is VLogL vs a flexible baked-in 709 profile. I won't get quite the DR latitude I had with the FS7, so I'm wondering if anyone else here has gone through a similar thought process and can speak from experience about the trade off between log and 709 on the GH5/S. I used to shoot quite happily on the C100/300 with their "Wide Dynamic Range" slightly flatter 709, so I'm wondering about going back to something similar.

I'm quite used to grading Log, but it is another bit of processor-slowing faff in Prem Pro, so it would save some time/bandwidth there too.

Or am I mad to ditch log on the GH5/S?

Cheers,

Ben.
 
HLG seems like the best option for the GH5/GH5S

Though there are some pretty nice corrective LUTs for both profiles.

But to answer your question CineD is a good option. Flatter than the other profiles but no where near as flat as log. If you want more of a no grading at all look the REC 709 profile is good or maybe natural for a bit more dynamic range.

As long as you expose right you'll be good to go. Log is really only needed if you want that post flexibility.
 
+1 for HLG as it has a known gamut and response curve that is easy to convert to REC709 in an NLE. V-LOGL on the other hand is an unknown quantity and the conversion to REC709 is difficult as the LUT's in NLE's are for the V-LOG and not V-LOGL. In a lot of shooting scenarios HLG just looks like a flat profile and can even be graded without LUT's using Lift, Gamma, gain and saturation.
 
I know I need to do some testing, but these are useful steers - thanks.

How do you monitor and expose HLG? A lot of my work is fairly fast moving obs doc, so having the flexibility to get exposure "in the ballpark" and then manipulate later has been very useful on the FS7. As a general rule, I used to slightly overexpose by around a stop on the FS7 and could always bring exposure back happily in post.

Ben.
 
Totally happy with either one, prefer HLG for my own personal work but for collaboration with others V-LogL is the most requested. All of the GH5 shooters I know are using it.
 
Now that I discovered how great HLG is, I never use any other profile. You just need to ETTR. Paul Leeming got Avery extensive instruction as well as a how to video on his fb group. Check it out.
 
Thanks - I'll give Paul's LUTs a try.

Looks like zebras at 90% for HLG - will do some testing with that. Haven't used zebras for years - but they're a useful exposure tool for fast-paced shooting. Nice one - thanks to everyone for the steers.

Ben.

mdifilm said:
Now that I discovered how great HLG is, I never use any other profile. You just need to ETTR. Paul Leeming got Avery extensive instruction as well as a how to video on his fb group. Check it out.
Click to expand...
 
Last edited:
bengiles said:
Thanks - I'll give Paul's LUTs a try.

Looks like zebras at 90% for HLG - will do some testing with that. Haven't used zebras for years - but they're a useful exposure tool for fast-paced shooting. Nice one - thanks to everyone for the steers.

Ben.
Click to expand...

Zebras are definitely helpful for run and gun, though I personally like to set my zebra exposure for skin rather than the hot spots of the image. Highlights blowing out a bit is bearable but bad skin is not imho. Depends on the shot of course. HLG definately gives you a bit of wiggle room, at least a lot more than the other profiles.
 
icarusfilm said:
Hlg compresses highlights heavily you don’t want any skin tones anywhere near that.
Click to expand...

So does V-LOG L. Anyway if you are shooting with 10bit 4.2.2 400mbps there is enough flexibility in the codec to pull the exposure in post without it breaking up. Also ETTR doesn't necessarily mean you are using skin as the highlight area for ETTR referencing. ETTR is a just another tool in your exposure controls and stating it's just 'terrible' without explanation is not very helpful.
 
ETTR is a terrible idea period. It creates non uniform exposure between shots where sometimes one shot may be EV 0 and the next could be EV +2 depending on what is in the shot. Sensors perform differently based on the exposure of middle gray and how many stops are above or below that middle gray.

ETTR may be easier to shoot and may result in less noise sometimes but it is way too unpredictable and can sometimes push the highlights into too thin of a range of values. It appears to have less noise because higher ETTR shots can be crushed down masking a lot of the noise and because you are padding more of the 12 stops into the shadows you have more shadow detail. That comes at the sacrifice of highlight detail which is what 99% of viewers care about. They care about the bright stuff in the shot and not what is in the shadows. Our eyes are drawn to bright things and we need to provide more detail to the highlights.
 
Shirozina said:
Please explain?
Click to expand...

First of all, it is not that I don't want to be helpful but this ground has been thoroughly plowed. I'm not going to change your mind and hopefully you don't change anyone's either because ETTR for HLG is a terrible idea.

Here's is some background:

SMPTE white paper said:
HLG is a transfer function defined by ARIB STD-B67 can capture up to 17.5 stops of DR. HLG uses a logarithmic curve for the upper half of signal values which allows for a larger dynamic range. HLG defines a nonlinear transfer function in which the lower half of the signal values use a gamma curve and the upper half of the signal values use a logarithmic curve. Sometimes there is a tendency to try and raise the diffuse white point too high and the pictures look washed out and too bright. The eye sees change in low light levels much more than in high light levels, so we need to give more bits to the lower light levels than we do to bright areas to improve how we see the blacks.
Click to expand...

Alister Chapman said:
..a problem is that a very large highlight range is squeezed into a very small recording range. As a guide you should expose diffuse 90% white (a white card or white piece of paper) at no more than 75%. Skin tones should be around 55 to 60%. You should not expose HLG as brightly as you do Rec-709.
Click to expand...

Now for my $0.02, exposing to the right squeezes skin tones into an increasingly small recording range. This may be acceptable if the scene specular highlights are very bright and you ETTR for them, the skin tones will be safely away from the small recording range, but then again maybe even too much so. On the other hand, if the scene specular highlights are not very bright then you move the all important midtones well into the smaller recording range. In either instance, your skin tones are being moved all over the place. The colourist will have to grade every scene differently because in some the skin tones are at 75%, others at 30%. The skin tones become thin when recorded into the small parts of the recording range. We want the skin tones recorded into the meatiest part of the recording range, the highlights have to fend for themselves. If you're just shooting nature scenes of trees and foliage and such, you can do as you wish. But if you are shooting paid talent, you better not fook with the lady's skin. and the colourist needs to grade from a consistent palette scene to scene as well.

HLG as a transfer function supports about 17.5 stops of DR, your camera a good amount less. It's incumbent to put the stops that your camera does have in the middle of that function where the data points are, not at the ends where there are fewer.

To some extent this applies to other log gammas, V-LogL being the exception because it doesn't record past 80 IRE, thus protecting you if you ETTR, it insures you won't record skin tones into a narrow recording range, you'll be clipping them instead!

https://www.smpte.org/sites/default/files/section-files/HDR.pdf
http://www.xdcam-user.com/2017/07/what-is-hlg-and-what-is-it-supposed-to-be-used-for/

Thomas Smet and icarusfilm got in before me with better, concise explanations! :thumbup:
 
sgttom said:
HLG seems like the best option for the GH5/GH5S

Though there are some pretty nice corrective LUTs for both profiles.

But to answer your question CineD is a good option. Flatter than the other profiles but no where near as flat as log. If you want more of a no grading at all look the REC 709 profile is good or maybe natural for a bit more dynamic range.

As long as you expose right you'll be good to go. Log is really only needed if you want that post flexibility.
Click to expand...


Yeah, this is pretty much the route I've taken too. Natural for straight out of camera/zero grading projects, Cine D and a light grade for more expensive projects.
 
Tom Roper said:
First of all, it is not that I don't want to be helpful but this ground has been thoroughly plowed.
Click to expand...

If it has, I’ve been struggling to find it. I know I’m late to the GH5 party, but all I’ve found is a sea of disinformation. I guess that’s the price of going for a prosumer product. Good to see some informed debate here. Thanks for the links.


Ben.
 
The best way to expose is simply- correctly especially when there are faces in the shot and are the focus of that shot. Exposing correctly will give you the max color fidelity of skin tones, the only times where exposing off script is useful is for landscapes when protecting highlights.


QUOTE=Thomas Smet;1986794869]ETTR is a terrible idea period. It creates non uniform exposure between shots where sometimes one shot may be EV 0 and the next could be EV +2 depending on what is in the shot. Sensors perform differently based on the exposure of middle gray and how many stops are above or below that middle gray.

ETTR may be easier to shoot and may result in less noise sometimes but it is way too unpredictable and can sometimes push the highlights into too thin of a range of values. It appears to have less noise because higher ETTR shots can be crushed down masking a lot of the noise and because you are padding more of the 12 stops into the shadows you have more shadow detail. That comes at the sacrifice of highlight detail which is what 99% of viewers care about. They care about the bright stuff in the shot and not what is in the shadows. Our eyes are drawn to bright things and we need to provide more detail to the highlights.[/QUOTE]
 
bengiles said:
If it has, I’ve been struggling to find it. I know I’m late to the GH5 party, but all I’ve found is a sea of disinformation. I guess that’s the price of going for a prosumer product. Good to see some informed debate here. Thanks for the links.


Ben.
Click to expand...

The GH5 HLG party was never the place to learn what HLG was all about. It was developed by NHK and BBC to solve network distribution complications from HDR, notably the proprietary Dolby Vision, and HDR10 which was an open source spinoff.

But...it's here, it's in cameras, and in the case of the GH5 there is even a version that plays straight to your HDR compatible tv.

The concept of shooting HLG as a prequel to converting it to BT709 and ST2084, through luts, through color management conversions, RCM, ACES and direct manipulations is the 1849 gold rush all over. Once you manage to get the HDR light to come on in your new 4k HDR television, you have struck the motherlode! But it's really only the beginning. HLG had some uses as an acquisition format in the GH5 because it was the "other" log format to V-LogL which was a paid upgrade. And it was discovered it did some things better potentially but has some traps, ETTR is one of them. ETTR and HLG will guarantee you have inconsistent exposure while occasionally nailing it right, if the circumstances agree. Your best friend is to learn the waveform monitor and shoot manual exposure if you want to play in the HLG sandbox, or prepare yourself for inconsistent exposure if you don't. And there is a whole other body of learning ahead if you want to convert it to BT709, ST2084 with consistent results. Luts with HDR are a really bad idea too. RCM with tone mapping or ACES idt's are the direct path to conversions without clipping data. If you get a cube lut that works, it won't work everywhere, but only within the narrow confines of a particular exposure and color temperature. This is the problem of converting from a larger color space to a smaller one, there will be out of bounds values, and is the very reason behind tone mapping and dynamic metadata. Cameras with 10 bit and log gammas can do HDR properly now, but the machinations to make HDR play compatible on LCD, OLED televisions is work in progress.

http://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP309.pdf
 
You must log in or register to reply here.
Back
Top