GH4 GH5 V LOG Proper Exposure

[Thomas Smet]

800 is what one find in the net - but you are right: I have not seen that published by Panasonic. By the way, that is something that is not great too. We have the same situation with Sonys FS7 - where a lot of people talk about a native ISO of 2000 in the CINE EI mode, but a lot of usere report that this is wrong.

It would be great if the industry would publish those figures really - since there the dynamic range should achive a maximum.

I think there have been plenty of tests like the one at DXOMARK to indicate the myth of dynamic range is better at a higher ISO is just not true. That may be true for some cinema cameras but it doesn't seem to really ever be true in the DSLR world.

 

[Lpowell]

800 is what one find in the net - but you are right: I have not seen that published by Panasonic. By the way, that is something that is not great too. We have the same situation with Sonys FS7 - where a lot of people talk about a native ISO of 2000 in the CINE EI mode, but a lot of usere report that this is wrong.

It would be great if the industry would publish those figures really - since there the dynamic range should achive a maximum.

There is no such thing as "native ISO" for digital cameras, ISO is just an arbitrary luminance gain factor. Camera manufacturers manipulate ISO behind the scenes to produce consistent exposure levels among camera models set to the same lens aperture and shutter speed. Until recently, cameras used Rec 709 built-in profiles and ISO was fairly well standardized. But the ISO of a LOG tone curve is not directly comparable to Rec 709 ISO, it depends on where the midpoint of the LOG scale is positioned. With the GH4, Panasonic manipulated the V-log L tone curve to make its numerical ISO about twice the ISO of the built-in Rec 709 tone curves.

 

[joe12south]

I'm generally a fan of ETTR, and it is a great way to go if you have no reference point, but it can definitely be abused and misunderstood to the point it severely degrades PQ.

I see this most often when a shooter tries to protect too much – usually a sky that is so bright that saving it causes the actual subject matter to be recorded in the shadows. Because just about every part of the image pipeline is designed to sacrifice the shadows, when you lift them you're left with a noisy, muddy, compressed footage.

It's a more nuanced technique to explain, but I tend to use a hybrid approach: Properly expose the subject (w color chart if I can, skin tone if I can't) and then adjust a stop or two to try to protect highlights if I can.

 

[visceralpsyche]

But how exatly do you implement the "expose to the right"? One way would be to increase the luminance compared to the approach above, using for example the new waveform monitor in the GH5 (or the Shogun together with a GH4). You can increase the luminance until the signal start to clip at the 81%. No idea where your are compared to the approach where you use a grey/white card - but you use the maximum of the luminance range that is available. So what approach do you follow in detail?

Set your zebras to 80% IRE in the case of the GH5's V-Log implementation, and you will accurately know where the exact clipping point is. Stop down until those zebras just disappear. Voila. ETTR done As always, you need to be judicious. No point in ETTRing to keep street lights from clipping - you'd have no image left. But the zebras show you precisely what is clipping, so you can make a decision on whether you want to clip it or not, rather than guessing by some arbitrary overexposure value with the zebras set to skin tone or something.

More info about how to ETTR here.

Cheers,

Paul

 

[ozmorphasis]

Barry speaks to ETTR in the White Paper, & it's worth a read.

As someone who hands off most their footage I'm not a fan of ETTR as it creates at lot of work for the DIT, &/or post. For consistent exposure, especially skin, I use either False Color, or Zebras with a 18% Grey/90% White card.

http://www.xdcam-user.com/2014/07/l...rd-pop-up-grey-card-and-90-white-card-review/

Bern, I'm interested in finding out a bit more about your approach when exposing for skin. Obviously, there is a range of skin out there, but when using false color or zebras, what are you generally shooting for? How are you doing this when shooting in a rec709 profile on say the GH5, and how do you approach this when shooting in Vlog on the GH5?

ETTR is simple enough and has its advantage, but I too have to hand off footage a lot of the time, but some clients ask for the flattest (log-like) footage, while others want a ready to edit look. I'd love to hear about your approach for both when using false color or zebras.

Part of my question also stems from having finally gotten more reliable exposure tools to work with (smallhd monitor and the GH5's new exposure tools).

Thanks in advance!

O

 

[robbin]

So is vlog-l better on gh5. Tested vlog-l on gh4 and it didntn give me better highlight laditude. Would be great if gh5 could give 1 stop better highlight range

 

[wschmid]

I think there have been plenty of tests like the one at DXOMARK to indicate the myth of dynamic range is better at a higher ISO is just not true. That may be true for some cinema cameras but it doesn't seem to really ever be true in the DSLR world.

Nobody said that the dynamic range will be better at a higher ISO?

 

[wschmid]

There is no such thing as "native ISO" for digital cameras, ISO is just an arbitrary luminance gain factor. Camera manufacturers manipulate ISO behind the scenes to produce consistent exposure levels among camera models set to the same lens aperture and shutter speed. Until recently, cameras used Rec 709 built-in profiles and ISO was fairly well standardized. But the ISO of a LOG tone curve is not directly comparable to Rec 709 ISO, it depends on where the midpoint of the LOG scale is positioned. With the GH4, Panasonic manipulated the V-log L tone curve to make its numerical ISO about twice the ISO of the built-in Rec 709 tone curves.

The native or generic ISO has always be defined in relation to the minimum noise level - was at least my understanding. True that there seems to exist different numbers for log or rec709 curves - but that does not mean that we will not have an iso figure where the noise signal ratio should become a minimum.

Beside those more theoretical aspects I would really like to know how you expose v log l if you assume that the iso
is not important. Because if that would be true, the the use of gray cards and 50% reflection points become less important. Then you adjust with either iso or exposure the luminance, and as long as you have enough light or do not go into clipping everything is fine.

I want just to point out that ETTR is the direct consequense of this approach. But expose to a white or gray card follows another approach what is not in line with ETTR really.

 

[wschmid]

Set your zebras to 80% IRE in the case of the GH5's V-Log implementation, and you will accurately know where the exact clipping point is. Stop down until those zebras just disappear. Voila. ETTR done As always, you need to be judicious. No point in ETTRing to keep street lights from clipping - you'd have no image left. But the zebras show you precisely what is clipping, so you can make a decision on whether you want to clip it or not, rather than guessing by some arbitrary overexposure value with the zebras set to skin tone or something.

More info about how to ETTR here.

Cheers,

Paul

Right, beside the suggested use if the zebra I prefer to use the waveform monitor for that, since that tool gives me more information then the zebra. Quite clear that lamps in a night setting will clip since your dynamic range is limited anyway.

So far so good - and my impression is that this is a very good way to utilize the dynamic range of the camera in the best possible way, since you bring it in line with the dynamic range of your szene. The only questionmark that I see is that you can end up with a wrong luminance in some cases (for example, a person in a darker room in front of a bright window may end up too dark). That are cases where the use of gray cards can be a good idea - to get the exposure of this person right.

But that is fine.

 

[joe12south]

Bern, I'm interested in finding out a bit more about your approach when exposing for skin. Obviously, there is a range of skin out there, but when using false color or zebras, what are you generally shooting for? How are you doing this when shooting in a rec709 profile on say the GH5, and how do you approach this when shooting in Vlog on the GH5?

ETTR is simple enough and has its advantage, but I too have to hand off footage a lot of the time, but some clients ask for the flattest (log-like) footage, while others want a ready to edit look. I'd love to hear about your approach for both when using false color or zebras.

Part of my question also stems from having finally gotten more reliable exposure tools to work with (smallhd monitor and the GH5's new exposure tools).

Thanks in advance!

O

I think what he is saying is that he uses a grey card under the same lighting as the subject. Expose for that known value and the skin will be where it is supposed to be. If I misunderstood, please correct.

 

[Barry_Green]

Set your zebras to 80% IRE in the case of the GH5's V-Log implementation, and you will accurately know where the exact clipping point is. Stop down until those zebras just disappear. Voila. ETTR done

Only if you don't care about the color fidelity of your brightest objects!

If you were to try that technique on a shot where caucasian skin is the brightest object in the scene (a dark-haired person in a dark scene perhaps) you'll probably find that your skin tones have shifted all to yellow. 80 IRE on VLOG-L is absolute clip -- it's higher, relatively, than 105 zebras are on a normal gamma.

Remember that the IRE is a cumulative formula, made from the individual brightness of the R, G, and B channels. If an object is primarily red or primarily blue, then that red or color channel stands a very high prospect of being clipped when the IRE brightness is that high.

Your technique would probably work better if you set the zebras to 75, so at least whatever the hottest object is has a prayer of not having at least one if not two color channels clipped.

 

[ozmorphasis]

I think what he is saying is that he uses a grey card under the same lighting as the subject. Expose for that known value and the skin will be where it is supposed to be. If I misunderstood, please correct.

Yup, got that part. I'm more curious about actual values and ranges when using false color and zebras for the grey card, and specifically how to approach that with Vlog as opposed to a standard profile.

ETTR has you basically just worrying about the brightest part of the image, but the other approach has you using zebras for a grey card/white card and false color as well...I'm assuming?

 

[visceralpsyche]

Only if you don't care about the color fidelity of your brightest objects!

If you were to try that technique on a shot where caucasian skin is the brightest object in the scene (a dark-haired person in a dark scene perhaps) you'll probably find that your skin tones have shifted all to yellow. 80 IRE on VLOG-L is absolute clip -- it's higher, relatively, than 105 zebras are on a normal gamma.

Remember that the IRE is a cumulative formula, made from the individual brightness of the R, G, and B channels. If an object is primarily red or primarily blue, then that red or color channel stands a very high prospect of being clipped when the IRE brightness is that high.

Your technique would probably work better if you set the zebras to 75, so at least whatever the hottest object is has a prayer of not having at least one if not two color channels clipped.

Yes, this is a fair comment I've often thought about using 75% IRE as my suggested clipping point for V-Log. In fact I recommend something similar for my Cine-D settings on the GH5, using 100% IRE as the zebra point to give a bit of headroom for saturated bright colours, for the same reason.

I think I might recommend 75% IRE for my V-Log settings to give a similar amount of headroom, since not everyone is aware of the skintone issue at near clipping.

Cheers,

Paul

 

[wschmid]

To the question of the base iso... what do you find in the net? A lot of statements - most see the base iso for both the GH4 and the GH5 at ISO 400.

The GH4 native ISO is 800 according to Panasonic. It’s definitely the best option when shooting V-log. I think a native ISO of 400-640 would be more appropriate for 4K in the other picture settings.

http://www.mkeproductionrental.com/gh4-video-tips-tricks/

The base ISO for VLog-L is ISO 400.

http://cheesycam.com/panasonic-gh4-vlog-l-samples-before-and-after/

While using the V-Log L profile the new base ISO is 400, not 200, so make sure to bring some ND filters along of you are shooting daytime exteriors.

https://www.eduardoangel.com/2015/09/01/panasonic-gh4-vlog-tutorial/

Panasonic GH4 has ISO 400 in V Log L

http://www.4kshooters.net/2015/06/19/gh4-v-log-l-vs-sony-a7s-s-log2-plus-panasonic-g7-review/

Nick states that the GH4 VLOG L base ISO is 400 and 800 looks great

http://eriknaso.com/2015/07/08/gh4-vlog-l-sample-grade-thanks-to-nick-driftwood-for-clip/

As a result, the GH5’s base ISO is 800 (a figure misattributed to the GH4 in an interview with one of Panasonic’s Japanese executives).

https://suggestionofmotion.com/blog/nab-2015-panasonic-gh5/

Bemerkungen zur Base ISO 400

Die Base-ISO der Kamera liegt in V-Log bei 400. Diese Erkenntnis fällt bei der GH5 besonders leicht, da die Kamera auch mit Gain-Werten arbeiten kann und 0 db Gain exakt den 400 ISO entspricht. Interessant ist dabei vor allem, dass dies auch in den anderen Bildprofilen gilt, in denen man auch Einstellungen unter ISO 400 findet.

Ein kurzer Check bestätigte: ISO 400 entspricht auch im Standard-Profil weiterhin 0dB, während ISO 200 mit -6dB angegeben werden. ISO 200 entspricht also einem Negativ-Gain und somit einer Absenkung unter die Base-ISO, zumindest wenn die angezeigten Zahlen stimmen sollten (wovon wir stark ausgehen). Und das würde wiederum bedeuten, dass man auch in den anderen Profilen für die beste Dynamik immer ISO 400 wählen sollte, selbst wenn ISO 200 vorhanden ist. Und es ist stark anzunehmen, dass dies auch für den Fotobereich zutrifft. Diesen Beweis überlassen wir aber den einschlägigen Foto-Test-Gazetten…

https://www.slashcam.de/artikel/Tes...g-Shutter---Bemerkungen-zur-Base-I.html#Bemer

 

[Thomas Smet]

To the question of the base iso... what do you find in the net? A lot of statements - most see the base iso for both the GH4 and the GH5 at ISO 400.



http://www.mkeproductionrental.com/gh4-video-tips-tricks/


http://cheesycam.com/panasonic-gh4-vlog-l-samples-before-and-after/


https://www.eduardoangel.com/2015/09/01/panasonic-gh4-vlog-tutorial/


http://www.4kshooters.net/2015/06/19/gh4-v-log-l-vs-sony-a7s-s-log2-plus-panasonic-g7-review/


http://eriknaso.com/2015/07/08/gh4-vlog-l-sample-grade-thanks-to-nick-driftwood-for-clip/


https://suggestionofmotion.com/blog/nab-2015-panasonic-gh5/


https://www.slashcam.de/artikel/Tes...g-Shutter---Bemerkungen-zur-Base-I.html#Bemer

That is because it is the base ISO. Base ISO is not native ISO. Base just means the lowest it can go. You can only go down to ISO400 if you shoot V-log. That is how it was designed and a fact of the camera. It isn't some belief of the internet but a cold hard fact. Those articles are pointing out that V-log does start at ISO400 just like the Sony starts at ISO1600 when shooting log. It isn't an assumption of what is the native ISO but the only way you can shoot log at the lowest allowed ISO level. Some people have assumed the GH4 has a native ISO of 400 because of V-log but that is just a guess. Base just means base.

 

[emweb]

Question about proper exposure for GH4 and GH5 V-log using an external waveform monitor.
Do you expose for 42 ire while monitoring with the V-log to 709 lut active?

 

[Bern Caughey]

Entirely depends on whether the Waveform represents the underlying LOG, or the LUT.

 

[Cary Knoop]

Hey all,

I've been searching around for an answer to this, but I was wondering if anyone could shed light on the proper exposure for V Log on the GH5. I used V Log on the GH4 and remember that stopping up 1-2 generally provided better results than 0. Is this true for the GH5, too? I have not been able to get my hands on another copy of V Log yet, so I am unable to do any tests.

Just expose right, not under and not over, this will give you the best picture.

The only exception I am willing to take is in case there is too little light, in that case there is not much of an alternative to over expose the video.

 

[joe12south]

Just expose right, not under and not over, this will give you the best picture.

We're always looking for tricks...some hidden knowledge beyond the obvious...that is going to change the fundamental reality that these cameras have little room for exposure error. I've spent so much time and effort over the years chasing a magic bullet, starting over with each new camera, ultimately to come back to this basic truth: light for your subject, expose for that light, then do what you can to control the rest of the scene.

And that's just for a baseline, "technically" correct exposure. But the exposure you really want is the one that helps tell your story. Here's a still from a recent screen test:

Look at that histogram out of context, and that's a bad exposure. Look at that histogram in the context of the story, and that's exactly the right exposure.

 

[pixelinfected]

right exposure is your right exposure
gh5 when you read in DB instead of Iso, you can see that 400 iso is 0 db, that mean is native sensibility of that camera, not gain more or less...
different style of shooting, different contrast of picture, different right way to exposure. You can use false color to check that is right exposure on some part of picture, if you want, but is not a right science.
i think to a Andrew laszlo that volontary underexpose a film when shoot Rambo (first blood), to force later in developing to have a picture more hard like war documentary...
We have tool to choose right exposure, then we can create more with light...