GH4 Looking back at early Vlog samples

And that is the problem. You guys are using the Varicam LUT, which stinks on the VLOG-L.
The Varicam V709 VLOG LUT is not a great choice for a GH4. You would think it would be, since the VLOG-L curve is supposed to be the same on both cameras, but the Varicam LUT doesn't translate to Rec709, it instead goes to V709, which is a different color palette. Which is a problem, because the Varicam sensor is not the same as the GH4 sensor, which is not the same as the DVX200 sensor. This should not be a surprise to anyone, but -- it makes a difference. If you have different color coming off the sensor to start with, you can expect that there will be different color after the footage is run through the same LUT.

As Dusty has found, different LUTs produce better results. LUTs that were designed for the GH4 in particular, will do better than LUTs that were designed for the Varicam 35, when used with a GH4. This should be obvious when you think about it, although it's certainly understandable why people would think that the Varicam VLOG LUT would be a good choice. But it isn't. It exaggerates issues that, as Dusty has pointed out, have always been there in all profiles.

So, as Dusty said, don't use it. Use LUTs that have been designed specifically for the GH4, and you will probably be much happier with the results.
 
So, as Dusty said, don't use it. Use LUTs that have been designed specifically for the GH4, and you will probably be much happier with the results.

Which begs the question, why didn't Panasonic supply us with one? They posted one on the download page for the V-log update, but it was the Varicam LUT. Stands to reason it should work properly with that one, and you can't blame folks for making that assumption.
 
Last edited:
Which begs the question, why didn't Panasonic supply us with one? They posted one on the download page for the V-log update, but it was the Varicam LUT. Stands to reason it should work properly with that one, and you can't blame folks for making that assumption.
I don't blame them at all. I made the same assumption, and was disappointed when I figured out what it actually was. I was doubly disappointed when they released a "DVX200 Varicam V709 LUT", which ... is identical to the existing VariCam LUT, and is not adapted to the DVX200 at all. I guarantee you that people will assume that that's a proper LUT to use with the DVX200, but -- it is, in fact, a VariCam LUT, not a DVX200 LUT. The Varicam LUT was engineered for the output of the Varicam's sensor. When using it with different sensors (the GH4 and DVX200), they have different output, and as such the Varicam LUTs don't deliver spot-on results. It is close, and it will get you in the ballpark, but -- it isn't what I would recommend using.

I have requested that they engineer and release a dedicated, proper Rec709 LUT. One that is designed for the sensor of the camera that it will be used on, and that will deliver proper final color. I hope that they take the suggestion.
 
I have requested that they engineer and release a dedicated, proper Rec709 LUT. One that is designed for the sensor of the camera that it will be used on, and that will deliver proper final color. I hope that they take the suggestion.
Thank you, Barry, your efforts to communicate customer concerns to Panasonic are greatly appreciated.
 
I was doubly disappointed when they released a "DVX200 Varicam V709 LUT", which ... is identical to the existing VariCam LUT, and is not adapted to the DVX200 at all. I guarantee you that people will assume that that's a proper LUT to use with the DVX200, but -- it is, in fact, a VariCam LUT, not a DVX200 LUT. The Varicam LUT was engineered for the output of the Varicam's sensor. When using it with different sensors (the GH4 and DVX200), they have different output, and as such the Varicam LUTs don't deliver spot-on results.

i don't think, it's a question of different sensors. the v-log specification defines a specific gamut, the signal belongs to. and this can be archived by simple matrix transformations of the raw data from the sensor. the resulting v-log conformant output should always look the same, independent of the actual sensor. that's the goal behind this specfication.
 
I've been having pretty good luck with the Arri LogC to Rec709 LUT, I just have to pull the high and mid greens down a bit.
 
It's funny how this is one of those 'can't be unseen once you've seen it' type problems. Still have to check out those Neumann LUT's dusty likes.
 
If we need a dedicated LUT, then what was the point of hobbling V-LOG L to "match"?
I have no answer to that. The answers to "why" are found in the Japanese factory. I can come up with answers as to what something IS, but deciphering WHY it is that way, is a different subject entirely.
 
i don't think, it's a question of different sensors. the v-log specification defines a specific gamut, the signal belongs to
If you're referring to V-Gamut, that's something that the Varicam has, and something that the GH4 and the DVX200 don't have.

Different sensors output different signals. That's just a basic fact of sensor design. Depending on the color filter used, and the processing employed on the sensor, will determine what the output of the sensor is. And I can guarantee you that the output of a $1500 stills camera, and the output of a $55,000 digital cine camera, are not the same. In fact, it doesn't take more than five seconds to show the difference. Look at the vectorscope reading on a Varicam shooting a standardized DSC chart in VLOG, and then look at the readout of a GH4 or DVX200 pointed at the same chart. They are not the same.

The GH4 is closer to the DVX200, which is closer to the Varicam, but neither the GH4 nor the DVX200 match the Varicam at that starting point. And if you're not starting from the same point, then you can't end up at the same point, if using the same directions. You'd need a different set of directions to end up at the same point. You CAN get there (or pretty close), but you'd need proper directions.

and this can be archived by simple matrix transformations of the raw data from the sensor.
Yes, the end result could be achieved by the matrix transformations of a proper LUT. But since the sensors are giving the LUTs different starting points, then you need a specific LUT to get them to the same end point. The Varicam LUT is not a good choice for the GH4 or DVX200. It is a good choice for the Varicam, yes, but not for the less-expensive cameras.
 
Aren't you listening to the experts...there is no issue, we're all just idiots who don't know how to hit the record button. :happy:
Quite helpful. Not.

Look, there is an issue -- using the wrong LUT is causing issues. If you use a LUT that was actually designed for the GH4 (like the Neumann LUTs appear to be) then -- you get better results. If you use the wrong LUT (like the Varicam LUT) then you get worse results. It's not rocket science, it's ... obvious.

What hasn't been obvious, prior to now, is that the Varicam LUT is just the wrong choice. That is rather obvious now. So -- once you've identified what's causing the problem, then you quit doing that, and ... the problem goes away... or at least is greatly minimized.

I cannot speak to the quality of the Neumann LUTs, I haven't used them. Dustylense seems to like them. The only thing I know about them is that they appear to have been designed specifically for the GH4 and the GH4's implementation of VLOG. Is it really any wonder why they appear to be working better than a LUT that was designed for the Varicam 35?

Even if you look at Illya Friedman's work, there's a line in one post he made that spells it all out: "I did not ever use the V-Log LUT, it was unavailable at the time." Kind of spells it out, doesn't it? He who had no problems, had no problems because he avoided the problem. The problem is using a Varicam LUT on a GH4. Use it at your own peril.
 
Last edited:
Look, there is an issue -- using the wrong LUT is causing issues

The issue exists independent of any LUT. The first time I noticed it, which was the first image I shot, I didn't even use a LUT. Let me repeat that...I didn't use a LUT. All I did was normalize the gamma curve.

Discrete macro blocks are recorded as either too magenta or too cyan. It's an encoder error that the necessary expansion of a log profile exacerbates.

Can a LUT be developed that desaturates low contrast, low saturation areas in order to mask the problem ... for sure. But that doesn't mean this is a LUT issue. It's just not.
 

The issue exists independent of any LUT. The first time I noticed it, which was the first image I shot, I didn't even use a LUT. Let me repeat that...I didn't use a LUT. All I did was normalize the gamma curve.

Discrete macro blocks are recorded as either too magenta or too cyan. It's an encoder error that the necessary expansion of a log profile exacerbates.

Can a LUT be developed that desaturates low contrast, low saturation areas in order to mask the problem ... for sure. But that doesn't mean this is a LUT issue. It's just not.
Well, it depends on what you mean as an "issue". Because people have been using the GH4 for how long(?) and singing its praises of how awesome its footage is, yet -- this same issue has always existed, in every profile, from day one. It's always been there, but nobody ran into it. Or, if they did, they didn't care.

Along comes the VLOG LUT, which appears to stretch shades in certain ways, and the world implodes. Yet... if you don't stretch those shades in that way, then -- it doesn't happen. As others have shown. And as we've seen with every other profile the camera uses.

The fact remains -- if you use the wrong LUT, it will exaggerate and stretch things to the point where the issue rears its head. If you use a properly-engineered LUT that doesn't stretch things in that same fashion, the problem doesn't appear -- or is so minimal, people don't notice it or care.

Can you stretch things in post and drive up the color saturation to levels that cause the artifact to appear, even without using a LUT? Why, yes, I'm sure you can. So the answer in that case would be "don't do that." Keep it reasonable, and your footage should look good. Stretch 8-bit 4:2:0 footage too far (on any camera) and it will break. You can stretch it further on 10-bit 4:2:2-encoded footage, as should seem obvious, but -- it can still break.

But the first and foremost advice I'd have for anyone wanting to use a LUT is -- don't use the Varicam one, that's for Varicams. Use one that's engineered specifically for the camera you're using, and you'll likely get much better results.
 
So the answer in that case would be "don't do that." Keep it reasonable, and your footage should look good.

Have you actually worked with the GH4's 8-bit VLOG-L footage? Based on that comment, I'm guessing not. I wish what you are suggesting could be a workable rule of thumb. I really do. But the sad fact is that you don't have to push it far or hard to reveal the issue. You don't even have to push the contrast and saturation to "natural" levels to reveal it.
 
Stretch 8-bit 4:2:0 footage too far (on any camera) and it will break. You can stretch it further on 10-bit 4:2:2-encoded footage, as should seem obvious, but -- it can still break.
Here's an example of the difference between recording a log profile in 4:2:0 color depth versus 4:2:2. This was shot on a JVC LS300 using the J-Log1 profile, which doesn't appear to suffer noticeably from magenta/green chroma smearing. Even so, there are chroma discrepancies visible in the 420 footage that look more detailed and realistic in the 422 version.

http://www.dvxuser.com/V6/showthrea...g-quot&p=1986583267&viewfull=1#post1986583267
 
[/COLOR]
Have you actually worked with the GH4's 8-bit VLOG-L footage?
No. I don't have a GH4. I have used 8-bit VLOG-L internal recorded footage from a DVX200, but that's as close as I've come.

I am basing it on my experience with having shot some charts on the GH4, and on the DVX200, and the Varicam, and seeing the obvious differences in the front ends of the camera which make it really obvious that using the wrong LUT will definitely cause problems. Followed up by Illya having said he encountered no such problems, and that he didn't use the Varicam LUT either.

I cannot tell you what to do that will exactly avoid all problems. All I can tell you is that there are things that you can do which will exacerbate and magnify any underlying problems. And using the wrong LUT is #1 on that list.
 
Back
Top