GH4 Looking back at early Vlog samples

Ted Ramasola

Ted Ramasola

Veteran
With all the issues raised, I looked back at a couple of samples.

I did not read in the description that they used the varicamlogtorec709 LUT and somehow the blockiness are either not there or not too obvious. I wonder what the early testers methods are? Looking at the comments, there are no new ones bringing up the issues of blockiness.

Maybe my eyes are already tired, what do you guys think?

This from Pro Photo Supply:


This one by Ilya of Hot Rod Cameras. He hasn't chimed in on the issues since the "release" yet right?

 
Hot Rod Cameras in Hollywood is hosting a V-Log L "happy hour" this Friday. Maybe Ilya will be there to address Magentagate?

hotrod.jpg
 
He was on the forefront on this development and was very active here prior to release. He is strangely quiet in all this. As a dealer, he must have been disappointed with the panny faux pas in the image app.
Surely he must know a way or methodology in processing vlog.
 
Ted Ramasola said:
He was on the forefront on this development and was very active here prior to release. He is strangely quiet in all this. As a dealer, he must have been disappointed with the panny faux pas in the image app.
Surely he must know a way or methodology in processing vlog.
Click to expand...

And as a dealer I'm sure he wants to help all the people who dived head first into a 3-foot deep kiddie pool using a backdoor implementation of a paid upgrade.

It's all a bit embarrassing as far as I'm concerned (and I'm not referring to Panasonic....)
 
I'm interested to read what the beta-testers thought about this issue, but you can't use youtube / vimeo videos as references. They're ridiculously compressed compared to the source material. You can't see the image noise that's present in the first place, let alone the magenta / green shift.
 
Ted Ramasola said:
He was on the forefront on this development and was very active here prior to release. He is strangely quiet in all this. As a dealer, he must have been disappointed with the panny faux pas in the image app.
Surely he must know a way or methodology in processing vlog.
Click to expand...

Umm... I actually have not been particularly active on DVXuser now for years, and really only appear when I am alerted to the fact.

While I appreciate the rampant speculations, I've shot a fair bit of V log and have not had the experience of others with Magenta or other color blocks. The number one issue I'm seeing is that saturation is being turned up WAY too high in the grade by the two users I saw complaining about the problem. When I politely mentioned this to one person on youtube who said "v log sucks" (and demonstrated this in a highly un-scientific video w/saturation turned up to the maximum possible in Premeire), there was no real response.

IMHO, this hasn't been an issue for me and so when I see sat turned up to 200 and complaints about the image- that's just FUD. Unless I have time to do a scientific test there's no point in getting into arguments with people doing that.

As I have had a few other things on my plate over the past couple weeks, including IBC, Celere Lenses, The Cinematography Podcast and a couple other projects there's really not been any time for me to participate in any forum based discussions about anything in a meaningful way.

In the two projects I shot with v log- I'm satisfied, so it really doesn't make sense why I should be the one to make people who (as far as I can tell) didn't buy V log be happy about their non-purchase?

There's some smart folks on-line who can do ANY KIND of testing required and could document the process. One of those people might take on the project if asked nicely.
 
I think the critical difference is most likely that the beta testers all used external 10 bit recorders, so the magenta cyan chroma smearing doesn't show up anywhere near as badly.
 
Illya Friedman said:
The number one issue I'm seeing is that saturation is being turned up WAY too high in the grade by the two users I saw complaining about the problem.
Click to expand...

I've seen this problem rear its head with no changes to color saturation at all, just the V-log LUT applied and nothing else.
 
Batutta said:
I've seen this problem rear its head with no changes to color saturation at all, just the V-log LUT applied and nothing else.
Click to expand...

I did not ever use the V-Log LUT, it was unavailable at the time. Anyone done a comparison of the reported problem with and without the LUT?

When I tried the V-log LUT for the Varicam 35 I did not think it was accurate, and haven't used it with that camera either.

We only recorded internally C4K 8-bit for the GH4 V-Log demo project.
 
Illya Friedman said:
Anyone done a comparison of the reported problem with and without the LUT?
Click to expand...
There is a simple methode to check for the Chroma smearing, inspired by what Lpowell wrote:

http://www.dvxuser.com/V6/showthrea...-LOG-L&p=1986578386&viewfull=1#post1986578386
http://www.dvxuser.com/V6/showthrea...-LOG-L&p=1986578534&viewfull=1#post1986578534

…Use a YUV based codec to check your footage

-Open your footage in Mpeg Streamclip.
-Export to Quicktime.
-Apple Component Video YUV-422/Apple Motion Jpeg A. Both operate in YUV space. Jpeg A slightly less accurate but a third in size.
-Make movie.
Open the resulting clip in Quicktime and adjust Brightness and Contrast with help of the A/V controls. You can even desaturate the clip.

Even if the problem it is not related to Chroma smearing a YUV codec does a great job showing what most of us experience as the problem …even when recorded in 10 bit.
V-log_YUV_s.jpg

Oh and btw I'm new here. Have been following the threads about the GH4 and Vlog for the last two weeks. Learned a lot already

Cheers,

bunk
 
Illya Friedman said:
Umm... I actually have not been particularly active on DVXuser now for years, and really only appear when I am alerted to the fact.

While I appreciate the rampant speculations, I've shot a fair bit of V log and have not had the experience of others with Magenta or other color blocks. The number one issue I'm seeing is that saturation is being turned up WAY too high in the grade by the two users I saw complaining about the problem. When I politely mentioned this to one person on youtube who said "v log sucks" (and demonstrated this in a highly un-scientific video w/saturation turned up to the maximum possible in Premeire), there was no real response.

IMHO, this hasn't been an issue for me and so when I see sat turned up to 200 and complaints about the image- that's just FUD. Unless I have time to do a scientific test there's no point in getting into arguments with people doing that.

As I have had a few other things on my plate over the past couple weeks, including IBC, Celere Lenses, The Cinematography Podcast and a couple other projects there's really not been any time for me to participate in any forum based discussions about anything in a meaningful way.

In the two projects I shot with v log- I'm satisfied, so it really doesn't make sense why I should be the one to make people who (as far as I can tell) didn't buy V log be happy about their non-purchase?

There's some smart folks on-line who can do ANY KIND of testing required and could document the process. One of those people might take on the project if asked nicely.
Click to expand...

I happily paid my $99, and I never went looking for this problem. It reared it's ugly head on the very first test image I shot. A lot of "smart folks" have proven that the poor encoding is not only real, it's prevalent in real world images. The whole point of shooting with a LOG curve is to have more latitude for adjustment in post. 8-bit V-LOG L is very brittle. You have to be more careful about how you push it than you do one of the "baked" profiles like Natural.

You of course have no responsibility to defend Panasonic's misstep(s), and I don't fall into the conspiracy camp, but I am bewildered as to how every single beta tester avoided the issue in every single beta shoot.
 
I'm not a pro anything, but noticed the odd colorings immediately. Took a video that happen to have my off white fridge in it, and bringing it back up, it looks like there is green mold growing on it! Nothing even close to that on it in reality. This was with the alleged correct LUT, other luts, and my own color corrections. Other profiles do not do this. Panny botched this, plain and simple. They should fix it for the PAYING customers - and leave the freeloaders dangling :D
 
joe12south said:
The whole point of shooting with a LOG curve is to have more latitude for adjustment in post. 8-bit V-LOG L is very brittle. You have to be more careful about how you push it than you do one of the "baked" profiles like Natural.

You of course have no responsibility to defend Panasonic's misstep(s), and I don't fall into the conspiracy camp, but I am bewildered as to how every single beta tester avoided the issue in every single beta shoot.
Click to expand...

I just spent about an hour going through the threads. It's a lot of speculation without testing all the relationships between which factor(s) is causing said problems. It's easy to say, "I can see this problem more when using V-log" so bam, that must the problem, right? Sorry, that logic doesn't hold water when the same artifacting is seen in other picture profiles- albeit less pronounced.

As Joe12south says- 8-Bit V-Log L is very brittle. My though reading this with all respect to Joe is frankly- no kidding, 8-bit 4:2:0 has always been f-in brittle. That 's what you sign up for when you commit to an 8-bit 4:2:0 capture and workflow. This isn't unique to Panasonic, or a news flash, it's been this way as long as I can remember.

While it is possible to do the labor to figure this out, definitively, I hardly feeling inspired to do so. If the answer is that it's predominantly user error, users don't want to hear this. If OTOH there is a problem in the implementation of V-Log, why are some people having what appears to be no usability problems?

So the end results of spending days or weeks testing and arguing about the results to deduce what's going on is either, blaming the user for capture, post processing or Panasonic for a fault for something that I and other people have already explained at length was not EASY to do, required serious time, labor and money to create.

Something I don't think anyone is remembering is that ALL the people who tested V-Log received a pre-loaded V-Log camera from Panasonic, not the V-Log upgrade by itself.

So go back to very most basic step that I didn't see anyone bring up. Did anyone test the cameras involved?

I proved back in 2009 with the GH1 that some native MFT lenses created processing problems with the camera compression system, and manual lenses did not.

Get 3x V-Log L cameras, point these cameras and lenses at a subject that reveals the problem in a very, very obvious way (you may have to create this subject) and if they all get the same results to at least rule out it's not a hardware issue effecting some cameras more than others, start with the basics and eliminate all the variables.

I don't have the clock cycles to invest in this. Frankly, I've helped this community with a number of technical issues over the years and with exceptions of a few peers who keep me honest and are grateful for the contributions, my previous efforts were largely treated with indifference or even contempt from the predominantly anonymous user base and even a couple of the moderators. So while I am flattered that Barry and others have asked me to get involved I have a dire lack of bandwidth for something like this, and don't really want wade through a million comments from DVX punters.

Has anyone thought to ask Adam Wilt or Phil Rhodes? They are more than capable of tracking down this error and if folks in the US or UK are willing to lend them a couple cameras they may be more receptive than I.
 
No one is blaming you (that I know of), you've done a lot of good around here.

BUT plenty of disappointment surrounding Panasonic's 8-bit VLOG-L for the GH4 around the webs. I guess we were all thinking since Canon and others had it working well years ago, Panny wouldn't have to reinvent the wheel... which they didn't even do just copying it from another camera in a way that probably does't fit GH4 very well.
It wasn't the amazing 2 stop improvement recorded internally that the ad-speak gave us. Now, the LUT seemed to be an improvement in the color space, but why can't that be fixed for normal profiles as well?

Anyway, not baggin on you, Illya, but disapointed about spending money because of advertising that didn't deliver anything better than maybe 1/2 stop on the high end, while loosing levels, adding noise, and generally being harder to grade and bring back.
I WISH it was user error, but so many here have way more skills than I do, I gotta believe it is not my lack of skills or knowledge about log, and more a real issue for most everyone.
 
Ilya,
I appreciate your reasoned response, and again, I understand that Panasonic's products aren't your responsibility, but to address some of your points:

- I, in fact, have 2 GH4's. They both exhibit the exact same issue. (As have many others, obviously.) I've factory reset both cameras.
- Yes, the YUV encoding error is present in all profiles. The problem is that it is unacceptably exaggerated when the V-LOG curve is expanded.
- Canon's C-LOG holds up well in 8-bit. Sony's S-LOG 3 is pretty good, too. V-LOG L is unnecessarily compromised in order to maintain compatibility with V-LOG.
- I have used a wide variety of lenses, manual and native. No difference.

I can understand not wanting to get in the middle of a random, anonymous internet forum circle-jerk...much pain for little gain...but I would think as a reseller who actually held a party to "upgrade" people's cameras, you'd have a greater than normal interest that said paid upgrade led to satisfied customers. For example, I think you'd want to know that you need to advise them to use an external recorder for consistently acceptable results.

This can be bandied back and forth, but this simple fact remains:
1. Choose V-LOG
2. Point your GH4 at a low contrast, low saturation target
3. Record internally.
4. Apply Pansonic's LUT in your preferred NLE.
5. The problem will manifest.
 
Illya Friedman said:
Has anyone thought to ask Adam Wilt or Phil Rhodes? They are more than capable of tracking down this error and if folks in the US or UK are willing to lend them a couple cameras they may be more receptive than I.
Click to expand...

i think, the technical reasons behind this issues are quite clear. Lpowell has explained them very well. that's not the point. but it's a pity, that panasonic didn't do it better. a log implementation for this specific camera -- a-vlog, that carries this 'L' behind it's name for a serious reason -- would have required a few very important design decisions done in a quite different way.

ACES proxy also doesn't use the huge gamut of its bigger sibling on small/legacy 10/12bit channels. and the same applies for the spread of used values. that's necessary to get a maximum out of limited resources. there are many well know log implementations around, where panasonic could have learned from to make it better and more efficient, but the actual product became unsatisfying in almost every aspect. -- sorry!
 
Last edited:
joe12south said:
Ilya,
I appreciate your reasoned response, and again, I understand that Panasonic's products aren't your responsibility, but to address some of your points:

- I, in fact, have 2 GH4's. They both exhibit the exact same issue. (As have many others, obviously.) I've factory reset both cameras.
- Yes, the YUV encoding error is present in all profiles. The problem is that it is unacceptably exaggerated when the V-LOG curve is expanded.
- Canon's C-LOG holds up well in 8-bit. Sony's S-LOG 3 is pretty good, too. V-LOG L is unnecessarily compromised in order to maintain compatibility with V-LOG.
- I have used a wide variety of lenses, manual and native. No difference.

I can understand not wanting to get in the middle of a random, anonymous internet forum circle-jerk...much pain for little gain...but I would think as a reseller who actually held a party to "upgrade" people's cameras, you'd have a greater than normal interest that said paid upgrade led to satisfied customers. For example, I think you'd want to know that you need to advise them to use an external recorder for consistently acceptable results.

This can be bandied back and forth, but this simple fact remains:
1. Choose V-LOG
2. Point your GH4 at a low contrast, low saturation target
3. Record internally.
4. Apply Pansonic's LUT in your preferred NLE.
5. The problem will manifest.
Click to expand...


And that is the problem. You guys are using the Varicam LUT, which stinks on the VLOG-L.

Look, the "magenta" issue and even banning are inherent to pretty much ALL profiles on the GH4.

Look at the beige wall outlined in red:

VLOG:
VLOG copy.jpg

CineD with Leeming LUT:
L109B copy.jpg

Sure, compressed here on DVXuser. But on my monitor the CineD version is actually WORSE than the version with VLOG.

Now, I think expectations on VLOG were too high. Especially when it comes to 8Bit 420. I've said it once, twice, maybe a dozen times. VLOG was and is designed to optimum use when using a 10bit recorder. And it works really really well when recording out. But I've gotten some great result shooting internally as well. It boils down to what and how you are treating the footage.

There is no conspiracy. It mostly boils down to high expectations and what you are doing to the footage.

And yes, there are easily 2 stops increased DR over the stock profiles. EASILY.
 
You must log in or register to reply here.
Back
Top