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

Shirozina said:
What LUT are you using ? Never seen the problems you describe unless I have underexposed footage.
Click to expand...

Upon further research, it is possible that I am underexposing my footage for v-log.
Originally used the V-LOG LUT provided by Panasonic. It is not good. The blacks are lifted and the color, particularly skin tone and certain blue tones, are just wrong. I presume this is because the LUT is expecting a V-Gamut input but the GH5 isn't really using V-Gamut like the Varicam.

I am using a LUT that I created using Lutcalc. Which is a great app. I am VERY happy with the color and contrast of the LUT I have created. I arrived at it through lots of reading and lots of trial and error.

But I am going to do some more exposure tests when I have time. Perhaps giving it another stop of exposure will solve my problem while still yielding better highlight retention than the Natural and Standard profiles. For my style of shooting, I am less interested in preserving shadow detail than I am preserving highlights.
I'll also do tests with HLG, as the method of converting it to V-LOG and then applying my LUT looks exactly the same and yet has a bit more information recorded, theoretically.
 
Also using a LUT isn’t the end of all things it gets the footage into a ballpark but fine tuning is still needed.



dvbrother said:
Upon further research, it is possible that I am underexposing my footage for v-log.
Originally used the V-LOG LUT provided by Panasonic. It is not good. The blacks are lifted and the color, particularly skin tone and certain blue tones, are just wrong. I presume this is because the LUT is expecting a V-Gamut input but the GH5 isn't really using V-Gamut like the Varicam.

I am using a LUT that I created using Lutcalc. Which is a great app. I am VERY happy with the color and contrast of the LUT I have created. I arrived at it through lots of reading and lots of trial and error.

But I am going to do some more exposure tests when I have time. Perhaps giving it another stop of exposure will solve my problem while still yielding better highlight retention than the Natural and Standard profiles. For my style of shooting, I am less interested in preserving shadow detail than I am preserving highlights.
I'll also do tests with HLG, as the method of converting it to V-LOG and then applying my LUT looks exactly the same and yet has a bit more information recorded, theoretically.
Click to expand...
 
dvbrother said:
Upon further research, it is possible that I am underexposing my footage for v-log.
Originally used the V-LOG LUT provided by Panasonic. It is not good. The blacks are lifted and the color, particularly skin tone and certain blue tones, are just wrong. I presume this is because the LUT is expecting a V-Gamut input but the GH5 isn't really using V-Gamut like the Varicam.

I am using a LUT that I created using Lutcalc. Which is a great app. I am VERY happy with the color and contrast of the LUT I have created. I arrived at it through lots of reading and lots of trial and error.

But I am going to do some more exposure tests when I have time. Perhaps giving it another stop of exposure will solve my problem while still yielding better highlight retention than the Natural and Standard profiles. For my style of shooting, I am less interested in preserving shadow detail than I am preserving highlights.
I'll also do tests with HLG, as the method of converting it to V-LOG and then applying my LUT looks exactly the same and yet has a bit more information recorded, theoretically.
Click to expand...


A picture is emerging, that potential problems exist in how the coloring app is interpreting the footage levels. Not talking about exposure here, but VIDEO levels versus FULL. Lifted blacks is a red flag. You could expose and light everything properly in-camera, and still get caught in a trap. I'm not a Premier user, so I'll describe how it works in Resolve, and how it works inside the Panasonic camera. V-Log, S-Log2/3 are all based on recording FULL levels, 10 bit, 0-1023 code value. V-LogL only records to 80% IRE, or about 815 but works the same as full levels because the mapping to 80% is the same as regular V-Log, just abbreviated.

But...HLG is recorded with VIDEO levels, 64-940 so there is a key difference right there but that's not all. The GH5/GH5s are recording the levels properly. How the coloring app interprets the file can be based on metadata, or in the absence of metadata base its selection on the file container. Assume you are recording 4k-UHD 10 bit 422 with the GH5 into an mp4 container. V-LogL gets recorded FULL, HLG gets recorded VIDEO. Since Panasonic h.264 is proprietary, most coloring apps can recognize from metadata that V-LogL is recorded FULL, and the app can bring it into the project that way, with the correct matching level. But the app often doesn't know what the levels are with HLG, or even that it is HLG.

But now you make a change, your client doesn't want or need 4K-UHD and requests you deliver in 1080p instead. You go into the camera menu and change to AVCHD. Now suddenly there is a conflict because V-Log/V-LogL is all recorded at FULL levels but AVCHD is always interpreted at VIDEO levels. With Resolve, you can choose how it is interpreted, VIDEO, FULL or AUTO. AUTO is a bit of a best guess and not 100% for the same reason mentioned, thus you have the option to specify it manually. But let's assume you chose AUTO for an AVCHD file that was recorded V-LogL. It's going to be interpreted wrong. That's just one example, but wrongful interpretations of recording levels result in files that can have raised blacks, crushed blacks, clipped whites, or dull gray whites. But too often, the fix used by the colorist is to twirl the color dials, offset, shadow, highlights, gain, gamma etc, or lay blame somewhere else.

If the file is brought into the application with a lut in front of it, circumstances can be even worse. Luts do this, they convert from a large log color volume (x,y gamut and z luma) to a smaller 709 color volume. In other words, 13 stops gets remapped into 9. If your levels are input wrong in the first place, and you subsequently color correct or adjust tone, you're moving data out of range. It gets clipped. Now you have all kinds of problems, raised blacks, crushed blacks, colors out of bounds. You're trying to fix what was brought in at the wrong levels in the first place to the lut with the color correction tools, and the lut is only making it worse because it clips anything that goes out bounds.

Make sure you know what your recording levels will be for the file type or gamma type, and make sure the coloring app uses the corresponding recording level before it goes to the lut, and before you start CC'ing it.
 
If exposure is botched I would advice to try and grade by hand. If you are not using resolve just try using saturation, curves/contrast get the skin tones as close as you can then desaturate the highlights and shadows.


Tom Roper said:
A picture is emerging, that potential problems exist in how the coloring app is interpreting the footage levels. Not talking about exposure here, but VIDEO levels versus FULL. Lifted blacks is a red flag. You could expose and light everything properly in-camera, and still get caught in a trap. I'm not a Premier user, so I'll describe how it works in Resolve, and how it works inside the Panasonic camera. V-Log, S-Log2/3 are all based on recording FULL levels, 10 bit, 0-1023 code value. V-LogL only records to 80% IRE, or about 815 but works the same as full levels because the mapping to 80% is the same as regular V-Log, just abbreviated.

But...HLG is recorded with VIDEO levels, 64-940 so there is a key difference right there but that's not all. The GH5/GH5s are recording the levels properly. How the coloring app interprets the file can be based on metadata, or in the absence of metadata base its selection on the file container. Assume you are recording 4k-UHD 10 bit 422 with the GH5 into an mp4 container. V-LogL gets recorded FULL, HLG gets recorded VIDEO. Since Panasonic h.264 is proprietary, most coloring apps can recognize from metadata that V-LogL is recorded FULL, and the app can bring it into the project that way, with the correct matching level. But the app often doesn't know what the levels are with HLG, or even that it is HLG.

But now you make a change, your client doesn't want or need 4K-UHD and requests you deliver in 1080p instead. You go into the camera menu and change to AVCHD. Now suddenly there is a conflict because V-Log/V-LogL is all recorded at FULL levels but AVCHD is always interpreted at VIDEO levels. With Resolve, you can choose how it is interpreted, VIDEO, FULL or AUTO. AUTO is a bit of a best guess and not 100% for the same reason mentioned, thus you have the option to specify it manually. But let's assume you chose AUTO for an AVCHD file that was recorded V-LogL. It's going to be interpreted wrong. That's just one example, but wrongful interpretations of recording levels result in files that can have raised blacks, crushed blacks, clipped whites, or dull gray whites. But too often, the fix used by the colorist is to twirl the color dials, offset, shadow, highlights, gain, gamma etc, or lay blame somewhere else.

If the file is brought into the application with a lut in front of it, circumstances can be even worse. Luts do this, they convert from a large log color volume (x,y gamut and z luma) to a smaller 709 color volume. In other words, 13 stops gets remapped into 9. If your levels are input wrong in the first place, and you subsequently color correct or adjust tone, you're moving data out of range. It gets clipped. Now you have all kinds of problems, raised blacks, crushed blacks, colors out of bounds. You're trying to fix what was brought in at the wrong levels in the first place to the lut with the color correction tools, and the lut is only making it worse because it clips anything that goes out bounds.

Make sure you know what your recording levels will be for the file type or gamma type, and make sure the coloring app uses the corresponding recording level before it goes to the lut, and before you start CC'ing it.
Click to expand...
 
Well, I did tests over the weekend and I guess I'm back on the V-Log train. Though I determined that my original method of exposure was correct, after all. However, it is just a fact that occasionally, dark gray such as asphalt or tires can be plagued with magenta/cyan splotches. This was much worse on the GH4, when recording V-Log in 8-bit and was a main reason shooters disliked using V-Log in 8-bit. Something about the way its flavor of h.264 encoded that dark part of the image.

In any case, I learned a lot over the weekend, and created a new LUT that solves basically all the issues I had with V-Log (and HLG). I discovered the IWLTBAP LUT Generator, which allows you to make precise color corrections in Lightroom or a Camera Raw app, and then create a LUT based on those corrections. Since the official Panasonic LUT is expecting V-Gamut as the input, and the GH5 does not use V-Gamut, the color is wonky. Blue is too light, yellow is somewhat pale, etc. SO I found a reference color chart shot by both the GH5 and a Varicam LT online, and used those to adjust colors and make a new LUT. Since the Varicam LT uses V-Gamut, my corrections seem to get colors to look a lot better than before. Blue skies are blue again, not pastel light pale blue. Skin tones have more of a hint of yellow. Reds aren't oversaturated anymore. I'm happy. It's not perfect, but a vast improvement.

I also applied this LUT to my HLG footage (after using the free HLG to V-Log conversion LUT from Emotive Color), and the footage looks pretty great.

Tom Roper said:
A picture is emerging, that potential problems exist in how the coloring app is interpreting the footage levels. Not talking about exposure here, but VIDEO levels versus FULL. Lifted blacks is a red flag. You could expose and light everything properly in-camera, and still get caught in a trap. I'm not a Premier user, so I'll describe how it works in Resolve, and how it works inside the Panasonic camera. V-Log, S-Log2/3 are all based on recording FULL levels, 10 bit, 0-1023 code value. V-LogL only records to 80% IRE, or about 815 but works the same as full levels because the mapping to 80% is the same as regular V-Log, just abbreviated.
Click to expand...

Premiere has always been weird with how it interprets video levels. It incorrectly interprets Pro Res footage shot on an Atomos Shogun, and one has to adjust those levels before adding a LUT (this is a well-known method and very easy). What I DIDN'T know until this wekkend's experiments, is that Premiere incorrectly interprets V-Log footage shot internally. It correctly interprets Rec 709 footage, and HLG, but V-Log is interpreted incorrectly in the exact OPPOSITE way that Pro Res files are. V-Log footage, which should have a range of about 8 to 80 IRE, only goes from 13-75 IRE! I have never read this on any forum!

News to me, and is certainly one reason I was having problems. So now I know, have fixed the problem, and am happy again. Premiere Pro users may want to do some testing and see if this knowledge improves their workflow.
 
That's so funny to hear that you've gone full GH. I did exactly the same thing. I shot exclusively with the FS7 but decided with the GH5 being able to shoot 4k in 4:2:2 10 bit (same as FS7), it was time to switch once the price dropped to $1,500. The biggest issue for me was the form factor of the FS7. It was way too bulky for trying to get to locations I like to shoot and it would draw too much attention when I was just trying to get in and out of a location. It is a great camera, but wow, the GH5 seems to be one hell of a camera. I too need to play around with things, but I would like to try my hand at VLog keeping in mind what others have said here. VlLog sounds closer to one of the Hypergammas in the FS7, where you don't overexpose and maybe even go 1/2 to 1 stop under at times.
 
" don’t miss the size/weight of the FS7 and 18-110 plus vlocks and Gratical EVF etc..."

I also dumped my FS7 for a GH5 and 5S. I liked the picture from the FS7 but in terms of daily use, size, weight, inconspicuousness, etc. the GH cameras win. Also love the presets, moving from 30 to 180 fps can be done without a thought. Not so on the FS7. The FS7 always struck me as being a bit too delicate, flimsy, as well as bulky.
 
I went with Cine D instead of V Log on this big project I’ve just shot. I just couldn’t get consistent results with V Log in a fast paced mixed light environment, when testing, so went with the baked in profile. Jury’s still out for me. Lots of advantages of the GH5/S over the FS7, but various gotchas as well of course.

I’ve kept my Sony 18-110 (actually bought it back) so I can use both systems and see what Sony replace the A7S2 and FS52 with. Overall, my biggest criticism of the GH5 is the over sharpened image. Looks a bit small chip video-y sometimes.

Ben.
 
bengiles said:
I went with Cine D instead of V Log on this big project I’ve just shot. I just couldn’t get consistent results with V Log in a fast paced mixed light environment, when testing, so went with the baked in profile. Jury’s still out for me. Lots of advantages of the GH5/S over the FS7, but various gotchas as well of course.

I’ve kept my Sony 18-110 (actually bought it back) so I can use both systems and see what Sony replace the A7S2 and FS52 with. Overall, my biggest criticism of the GH5 is the over sharpened image. Looks a bit small chip video-y sometimes.

Ben.
Click to expand...

I am guessing you shot everything with NR and sharpening all the way down? One odd thing that was found is that electronic lenses added further sharpening. Someone concluded this by taping off certain contacts on the back of the lens. I think it was just with electronic lenses. People also found 5k open gate mode to have less sharpening.
 
sgttom said:
I am guessing you shot everything with NR and sharpening all the way down? One odd thing that was found is that electronic lenses added further sharpening. Someone concluded this by taping off certain contacts on the back of the lens. I think it was just with electronic lenses. People also found 5k open gate mode to have less sharpening.
Click to expand...

Yes, both dialled right down. It’s a shame there’s nobody at Panasonic Lumix division to have a sensible dialogue with. The people at the trade shows are usually sales teams with a very low level understanding of the products. I’d like to be able to switch sharpening off completely, while still enjoying the stabilising abilities of the native lenses. And if there’s a good reason they’ve set sharpening/NR at a certain minimum level, it would be good to know why.

Ben
 
Cary Knoop said:
You absolutely need at least some noise reduction in order to get a reasonable compression efficiency (you can't compress noise).
Click to expand...

Cheers Cary. It's the sharpening that's the issue for me - I'd like to have the option of seeing how the picture looks with less sharpening.

Ben.
 
Cary Knoop said:
If I put sharpness at -5 I do not see anything overly sharpened.
Click to expand...

I guess that's subjective - I'm comparing it with a 3+ year working knowledge of the FS7, which has a noticeably softer presentation - but which responds well to a gentle bit of sharpening when needed.

My gut feeling is that the GH5/S image could be just a bit more pleasing if Lumix gave us the option of lower sharpness - especially with native electronic lenses.

Ben.
 
Last edited:
Plenty of options in post to soften the image........ I don't find it too sharp but in 4k/UHD it's downsampled so the perception of sharpening may be from the resize algorithm rather than any added sharpening. When I use it in open gate mode i.e with no in-camera downsampling it looks a bit less 'sharp'. Toying with the prospect of an S1 or S1h and the later features an OLPF - does the FS7 have one?
 
You must log in or register to reply here.
Back
Top