Cinema5D slates the Panasonic GH5, calls V-LOG and 10bit "unusable" - They're wrong

[jonpais]

1 minute ago, Fritz Pierre said:

In those words lie maybe one of the biggest issues regarding the GH5 & 4, or really any other Hybrid/Dslr style camera for that matter....

"I've uploaded way more Panasonic videos than Fuji - not because the image quality of the G85 is better, but because it is so much easier to just pick up and start shooting video."

Jonpais 

When you're shooting a feature or a commercial or any other narrative form, how the camera handles....the interfaces for selecting  images...accessing menus....manually controlling the camera  etc. is everything...and the GH5 comes in squarely, after purely dedicated cinema cameras...and when you're chasing beauty light with 20 min to go, that counts for more anything else does....and of course reliability...we are already into a ridiculous realm in terms of image even in sub $1000 cameras...the reality is it's no longer about the camera, but a lot more about the discussion of the camera!....of course I'm equally guilty of that.... 

 

Image quality doesn't exist in a vacuum: we just want the camera to get out of our way so we can take pictures. With the Lumix cameras, I can shoot and the camera is like an extension of myself. The Fuji's kind of a hassle: I've got to drag along a tripod, or only use small lenses on a gimbal, and at the moment, I've only got one small lens, the 35mm f/2. We should all be focussing more on our craft, and less on gear. Or at least not attacking somebody because they use Canon or Sony or Nikon or whatever. I stopped going to Cinema 5D for information a while back, and if it weren't for Andrew's post, I would never have read them again. I often asked myself why I'm listening to certain online reviewers talking about lighting or sound or photography, when meanwhile, the lighting and sound of some of those videos is atrocious. I'm a video shooter, I don't know of any video shooters I admire posting free online tutorials, so I've been watching a lot of Jason Lanier's videos, pretty inspirational. 

[Philip Lipetz]

Off topic, but Jason Lanier's work in theory of computer interactions is amazing. I assume we are talking about the same guy.

[johnnymossville]

I noticed something else as well.  What's with this DR graphic.

In this picture they try to paint the Panasonic as having lower DR than the Fuji,  but when you put them side by side that are almost exactly the same.  I find this lame at best. 

 

[hyalinejim]

Well, they do say:

Quote

We can see that the sensor’s total dynamic range is higher than those 10 usable stops

So their software is flagging those lower stops as below their chosen noise threshold and this happens earlier than in the other cameras in the comparison. However, as has been pointed out, we don't know what the implications of various types of NR would be across these cameras. Does anyone know if the Fuji has heavy NR going on in F-Log?

It's also worth noting that some tests have shown 10bit to be noisier than 8bit on the GH5. But aside from that, give me noise over noise reduction any day of the week, please, and let me deal with it in post. Furthermore (!) the noise on the GH5 is a really nice grainy luma noise. I have no problem with it in the samples I've seen. Although there is temporal NR with ghosting artifacts at higher ISOs (Emmanuel Pampuri's night shots, the girl eating outside the bistro):

With regard to the chroma smearing, it hasn't been an issue for me in the 24p VLog samples. Yes, I can see it even in Pampuri's beautiful flower girl's lovely nose. But no-one in a million years is ever going to spot that, even in a cinema, I think. However, there is an egregious example of it in Neumann's wonderful 180fps footage:

Uncropped:

 

Cropped 100%:

Now, this you would probably spot on your phone while relaxing on the beach at noon as it's there frame after frame after frame. But this is 180fps.

I think V-Log, even in its current incarnation on the GH5, is going to be totally workable at 24p. For higher frame rates, however, you might want to switch to another profile - I'm  thinking of Leeming Lut One for Cine D, or EOSHD Pro Colour for whatever settings you're supposed to use for that. Also, let's not forget the possibilities of the Like 709 profile which we've heard very little about thus far. Will it be contrasty, or will it be like Wide DR on the Canon C series - a sensible compromise between dynamic range and full range levels.

And finally, even if 400Mbps All-I is not enough to kill the nasties in V-Log, the 200Mbps All-I for HD surely will, at effectively double the bitrate per pixel. I know I'm likely to be shooting in HD for a lot of jobs anyway.

 

[cantsin]

1 hour ago, Philip Lipetz said:

Off topic, but Jason Lanier's work in theory of computer interactions is amazing. I assume we are talking about the same guy.

No. You probably mean Jaron Lanier. (It always seemed to me that Jason was cashing in on this easy-to-be-made confusion.)

[no_connection]

That does not look like a bitrate or bitdepth problem. But straight out encoder problem.
How they did not spot or fix that is in development is beyond me. But so is the C100 temporal noise reduction.

[Clayton Moore]

6 hours ago, Philip Lipetz said:

 

BTW am I worried about this "problem?" Not at all.  And I did not mean to imply that I was. I was shocked to be asked if I was cancelling our nonprofit's orders. Absolutely not. the majority of our work goes to the web, and no one will ever notice. Not even me. If I ever need to shoot for the big screen I will use our Atomos, or 400mbs. 

Indeed, for web delivery all this testing is pretty much nonsensical.  In fact I find the whole idea of these tests zoomed in by 400% or whatever a bit absurd as well.

[TheRenaissanceMan]

They also don't know how to use the Xyla chart correctly. To properly utilize the chart, you adjust the exposure until the first chip is just barely clipping; then, you count each chip down to find your DR. But they consistently count from the third chip, leaving the exact starting point of each camera's DR up in the air.

Panasonic does usability and reliability better then any other hybrid. Most real shooters are willing to trade 2/3 of a stop for a tool that gets out of the way.

[Eno]

@hyalinejim, from what I saw, Fuji by default applies some stronger NR cause the detail is lower than the GH5 and both of them are suppose to be resizing from a 5,5K image to 4K.

@TheRenaissanceM, you are very correct about their lack knowledge how to properly measure DR. It is as Driftwood said:

"Cineon adaptation \ log curve interpretation generally looks different from toe to shoulder between different cameras. Its like Slog 2 is good for highlight detail, slog3 for bringing out shadow detail and vlog sits somewhere in between if not too unlike slog3! C log curve differs too. The point is you have to expose\grade correctly with Log & for the camera and DR youve got."

[Andrew Reid]

10 hours ago, no_connection said:

That does not look like a bitrate or bitdepth problem. But straight out encoder problem.
How they did not spot or fix that is in development is beyond me. But so is the C100 temporal noise reduction.

This is my whole point about the Cinema5D post

They are complaining it's a bit-depth issue with 10bit when it's actually an encoder issue / compression related.

And 10bit looks pretty damned good in the standard colour modes to my eye, yet they claim overall it gives zero benefit. Go figure.

[jonpais]

V-log aside, several forum members here are questioning the superior dynamic range of the Fuji X-T2 over the Panasonic GH5, which is preposterous, for lack of a better word. I am brand agnostic - I've got nothing to gain from acknowledging that camera A has better picture quality than camera B. I've shot with Panasonics since before Andrew convinced me to get the GH3: I've shot with the GM1, the GH4 and the G85, and I've preordered the GH5, so nobody could accuse me of fanboyism. I've got nothing against Panasonic. I shoot with Panasonic nearly every day. Have you ever even shot with an X-T2? Have you had a chance to test a Panasonic camera side-by-side with a Fuji? I have. Have you ever even handled a Fuji camera? I have, and not only does it better the dynamic range of what I've seen so far of the GH5, it also has crisper 4K and much more pleasing color rendition, and I'm not just talking about skin tones. It also has less noise and is cleaner at higher ISOs. If you've shot with both Fuji and Panasonic, I'd love to hear you out. Go ahead, if you've shot with both and can honestly say that the image quality of the Lumix trounces the Fuji, I can respect that. If not, you're just going by lab tests, camera reviewers who know absolutely nothing about photography, hearsay, or lousy footage you've seen online. 

[hyalinejim]

Well, by the same token unless you've shot with the GH5

8 minutes ago, jonpais said:

you're just going by lab tests, camera reviewers who know absolutely nothing about photography, hearsay, or lousy footage you've seen online. 

 

 

[jonpais]

Just now, hyalinejim said:

Well, by the same token unless you've shot with the GH5

 

 

I already qualified my statement by saying, 'from what I've seen so far of the GH5'. Please read my post.

[hyalinejim]

I did. From what you've seen of:

18 minutes ago, jonpais said:

lab tests, camera reviewers who know absolutely nothing about photography, hearsay, or lousy footage you've seen online. 

Please use logic.

[deezid]

Well, there's actually nothing coming close to color rendition compared to Fuji in the consumer area. Even Canon is way worse in terms of color reproduction "from what I've seen". Let's not talk about the sharpening algorithm being using on the X-T2, which seems to be out of this world, much better than the GH5 which looks electronical.

The GH5 seems somewhere between Canon and Sony while being closer to Canon.

Dynamic range on the Fuji X-T2 should be a tad better using F-Log vs. V-log but not by 2 stops, especially not by regarding the comparison above from C5D.

Coming back to the topic: The GH5 seems to show similar encoding artifacts as the GH4, even recording in 10 Bit internally. Only way to fix these is to record externally atm.

[jonpais]

1 minute ago, hyalinejim said:

I did. From what you've seen of:

Please use logic.

No need to get nasty.

[hyalinejim]

Well then, don't be condescending.

[jonpais]

1 minute ago, hyalinejim said:

Well then, don't be condescending.

If you haven't shot with the X-T2, we've got nothing further to say to one another.

[hyalinejim]

You're not making sense, Jon. From an epistemological point of view if it's unacceptable, as you claim, for people other than you to make inferences regarding a camera they haven't used it must logically be unacceptable for you to make statements about another camera that you haven't used. I'm not being nasty here. Putting feelings and emotion aside, this is rationality.

Edit: And you're right, I do actually have nothing useful to offer here in terms of a Fuji - Panasonic comparison. This is just a point of order.

[jonpais]

31 minutes ago, hyalinejim said:

You're not making sense, Jon. From an epistemological point of view if it's unacceptable, as you claim, for people other than you to make inferences regarding a camera they haven't used it must logically be unacceptable for you to make statements about another camera that you haven't used. I'm not being nasty here. Putting feelings and emotion aside, this is rationality.

Edit: And you're right, I do actually have nothing useful to offer here in terms of a Fuji - Panasonic comparison. This is just a point of order.

How about if I rephrase that then? Is there anybody here in the forums who owns an X-T2 that truly believes that the GH5 footage they've seen online, be it good, bad or otherwise, trounces or is in any way superior to the Fuji? I'm not into gear wars, don't care to engage in any, but several forum members seem to be highly skeptical that the Fuji has more dynamic range and detail. Is that acceptable to you?