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

[Andrew Reid]

Cinema5D have put the Panasonic GH5 on test, with a few charts and have some harsh words for the new 10bit mirrorless camera.

I don't agree with much of their criticism.

Here's an alternative opinion.

Read the full blog post

[Zmaslanik]

Bravo! I like Cinema5D, among other websites like yours, but I appreciate your breakdown of this article. Also, I enjoy reading your articles in general!

I plan to buy a GH5 when it comes out and these little criticisms... Don't really bother me. I have used the GH4 for two years, and I 

I have used previous GH models in the past, so I am looking forward to getting my hands on the GH5!

[etidona]

Very well said Andrew! Maybe I'm biased but to me looks like

1)the fs7 has less artifacts in middle-dark areas but more in the bright areas (look at the thumb)

2)they missed focus on the fs7 or the gh5 has far more details

[Borbarad]

Thanks for the article!

I also read the cinema5D "review" and thought that something doesn't stack up. First I taught that their samples really didn't provided the clear evidence and then the text wasn't lined up properly.  On top it should also be clear that the upcoming 400 codec is made for 10Bit,as this was stated elsewhere ( the PDF or?)

 

B

[Liszon]

I was expecting this and I think it is a major mistake to release the GH5 without the 400Mbps update. First impressions mean a lot and everyone is eager to read the early reviews. Once people see "10bit GH5 worth nothing" a few times it will be hard to make them forget it half a year later even if the new firmware brings a completely new experience. Think of the Ursa Mini 4.6K and how bad the first batch was. Rental houses even state it's not suitable for professional work even today..

[OliKMIA]

Question @Andrew Reid, is there a real benefit for 10bits 4K at 150mbps ? Does this extra 50mpbs over the GH5 really allows to "store" the 10bits info. I mean, the codec is still H.264.
Also, I know the 400mbps update is coming but it will be ALL-I vs more efficient IPB for the 150mbps.

What do you think about that Andrew ? Thanks

[JLemos]

Also, if i'm not wrong, they don't really tell how the lut and grading was applied. For example, if they just plaster it over the image in Premiere with Lumetri, there may be some problems like those if the sequence isn't at "Maximum Bit Depth".
That is even seen if you're using a native ProRes file, with a h264 file should be even worst.
You can see the difference in a fast example i made with a 10bit prores v-log file from a GH4 (recorded with Ninja Assassin), left without "Maximum Bit Depth", right with "Maximum Bit Depth" active.
p.s: I just applied the native "Alexa_Default_LogC2Rec709" from Lumetri

[no_connection]

I may be blind but I don't see any original files to download. No way of telling if the artifacts are there fault or not.

On thing is for certain, the sample pic shows some pretty bad artifacts, does not look like regular compression or bitdepth error as you can see some noise coming through the stairsteps/area, mostly red channel.
Areas with more information/detail seems fine tho. Would not surprise me if it's a bug that has already been corrected.

 

One big benefit of 10bit is you need less noise to avoid banding, so you can have a cleaner image. A cleaner image compresses better so it's win-win.
At higher ISO where it matters less I would rather have those bits put into resolution instead.
 

A bit late now so Ill catch up tomorrow.

[theSUBVERSIVE]

Yeah, these were almost exactly my comments there. C5D does far too often technical mistakes like this, it happens in basically every article. They look more knowledgeable than they actually are and I've seen this in so many articles, enough so I follow the news, but take any technical report with a grain of salt.

They are confusing compression issues with color depth issues, two very distinctive things, propagating terrible misinformation for those that know even less than they do.

To clear this up, they should have done an external recording to rule out compression and codec issues before jumping into 10-bit is no good. Plus, I'm not really into this level of pixel peeping anyway. They even missed some other observations about the other cameras artifacts or even other GH5's advantages too.

In the end, they are spreading a lot of misinformation which will only confuse even more people that already don't eff understand what color depth and chroma subsampling is, this is already a complicated topic to discuss out there because there are already a lot of poorly done tests about this matter and this isn't doing it any favor at all.

 

[jonpais]

3 hours ago, Andrew Reid said:

Cinema5D have put the Panasonic GH5 on test, with a few charts and have some harsh words for the new 10bit mirrorless camera.

I don't agree with much of their criticism.

Here's an alternative opinion.

Read the full blog post

Did you meant to say elevate or alleviate? Thanks for the article!

[Philip Lipetz]

 C5D trashed the Cannon XC10, tanking sales for a year until Andrew published tests that restored its reputation as  brilliant camera under the right conditions. 

[Don Kotlos]

1 hour ago, theSUBVERSIVE said:

To clear this up, they should have done an external recording to rule out compression and codec issues before jumping into 10-bit is no good.

You are right it is not just the bit depth but James Miller did the tests already with an external recorder and showed that these artifacts are still there. You need to have an LUT baked-in in order to get much better results. That is "YUV chroma smearing". So it is not just the codec but the whole pipeline including the Vlog. 

It is visible in other footage from GH5 as well:

and the links that explain it posted from @bunk

That happens also with the s-log from Sony's but I hope the new codec in the summer can fix it. 

[bunk]

I might want to add that there was something 'fishy' going on. Turned out that the 422 10bit version I downloaded was in reality a 420 8 bit version. Emmanuel Pampuri replaced the file some what later and you can now download the correct version. It holds up way better. The "problem" remains, but doesn't jump in your face anymore.

[Michael Ma]

I would like to know how they graded the 10-bit image, and what software and version they used.  I sure hope they didn't use a LUT for their "rough rec.709 grade".  Almost sounds like it.  If that's the case, it's most likely a major source of the problem.  It does bring up the point if IPB could be inadequate for 10-bit 4K LOG.  I would like to see further on this.  I hope Panasonic sends Andrew a GH5 soon!

[theSUBVERSIVE]

5 hours ago, Andrew Reid said:

What Cinema5D are seeing is not a serious “problem with bit-depth” as they put it, it’s just compression and a very poor grade. Perfectly normal and the Sony FS5 initially had macro-blocking in 10bit far worse than what I see with the Panasonic GH5.

Is 10-bit a typo? I remember the FS5 having macroblocking issues when recording 4K internally but the FS5 doesn't record 4K 10-bit 422, it's XAVC-L 8-bit 420.

[mercer]

26 minutes ago, theSUBVERSIVE said:

Is 10-bit a typo? I remember the FS5 having macroblocking issues when recording 4K internally but the FS5 doesn't record 4K 10-bit 422, it's XAVC-L 8-bit 420.

I thought it recorded 10bit 1080p internal?

[IronFilm]

Yes, it would be a reference to the FS5's internal 1080.

That is how I read it.

[Lintelfilm]

The problem is with V-Log. It was an outright disaster on the gh4 - even 10bit to an external recorder it showed huge, horrific banding artefacts like those separated colours in the 5d article. It doesn't surprise me at all that the GH5 is the same. They simply haven't fixed v-log. I'm not very hopeful what difference high bit rate will make.

Fair enough the GH5 10 bit might be better with a non-log profile, but it's hard to get excited about 10bit with 10-11 stops DR. 

Andrew you yourself said you couldn't see any difference between 8bit and 10bit external on the GH4. Why do you think the GH5 will be any different?

xx

[Dimitris Stasinos]

If you set the 10 bit artifacts in Vlog aside and look at their cons list, this review seems quite subjective. "Dynamic range is limited". 10-11 stops is "fine" for such a camera, it's slightly worse than xt-2. "Lowlight is average". No it's not...This is an m43 camera that gives you a usable 3200 iso sensitivity, which is not average, it's "fine". "1080p 180fps gives you less resolution than actual 1080p". Really? Name a camera at this price range (or double the money if you want) that does better than this. Luke's HFR footage was terrific.. "Some canon lenses did not work with speed booster..." Cmon...what has this to do with the camera? In which universe these are cons?

[Jimmy]

13 minutes ago, Dimitris Stasinos said:

"1080p 180fps gives you less resolution than actual 1080p". Really? Name a camera at this price range (or double the money if you want) that does better than this.

So how is their point wrong? It just means all the cameras in this price range also have the same con.

They have had their hands on the camera and have given an opinion, just like every other blogger out there. For those that are convinced the GH5 is great and worth the money (me!), then there is no need to worry. If you are on the fence, then it is another point of view to digest and make conclusions from.

At least we can put to bed the idea that all these other sites just pander to the whims of the manufacturer.