Panasonic GH5 10 bit internal recording not good enough

[interceptor121]

I have spent some time decomposing the codecs and I have had a real bad surprise with the 150 Mbps 422 10 bit codec.

I was hoping this to be an equivalent to Prores 422 but it is really a poor codec implementation and I am not sure the 400 mbps all intra is good enough 

full read of the findings here

https://interceptor121.com/2018/08/13/panasonic-gh5-demystifying-movie-recording-settings/

looks like no 10 bit recording really possible without an external recorder

[sgreszcz]

Do you suggest the 400 Mbps 422 10 Bit All intra 25 fps (1.9 MB per frame) or better to just use the 10 bit FHD All Intra 200 Mbps codec if you don't need 4k?

[interceptor121]

I am not suggesting anything however the FHD formats are all almost equivalent in terms of image quality and 200 mbps is higher than Prores HQ so if you wanted to shoot LOG and do serious grading this codec will hold a lot of beating before breaking down

I am just disappointed the FHD 422 10 bit 100 mbps is genuine IBP while the 4K has a poor codec probably the onboard processor can’t manage all that info but then I would have not even created that option. Having said that Sony codecs are the same with only I and P frames I was expecting more from Panasonic

 

[jonpais]

Most of the articles and videos I’ve seen have tended to confirm your findings that 10-bit 150Mbps is to be avoided. 

According to an article written by wolfcrow before the GH5 firmware update, 10-bit 4:2:2 150Mbps only contains 10% more data than 8-bit 4:2:0 100Mbps, not nearly enough to realize a difference in quality.

He further calculated that 10-bit 4:2:2 would require 560Mbps just to be better than 8-bit 4:2:0 100Mbps. 

Yet when he made a video about the GH5 after firmware update v2.1, he recommended shooting 400Mbps internally and not using the Inferno (I think he claimed to see artifacts when recording externally). 

At the same time (I’m relying on my memory here) I seem to recall some people saying that 10-bit 150Mbps did offer a real advantage over 8-bit in that V-log footage was no longer plagued with banding artifacts.

I never shoot 8-bit or V-log, so I can’t say - and that technical stuff is way over my head - but in my limited experience, HLG 10-bit 400Mbps can be manipulated in post quite nicely.

 

[Shirozina]

Despite the poor specs of the 150mbps LongGOP codec it does hold up very well in grading. Mainly I think due to it's 10bits which make the luma channel near immune to banding unless you apply insane curves or tonal shifts and bends. The main reason to avoid it is the stress it places on most editing rigs trying to decode it on teh fly for smooth playback. The 400mbps codec is better in this respect but still harder on the CPU and GPU than ProRes or DNxHD. Given that most people will expose and color balance near optimally and not treat it as RAW the inadequacies you describe are purely academic and do not translate into real world problems. Until other mainstream manufacturers adopt 10bit 4.2.2 internal codecs the GH5 is well ahead of the game at the moment and a joy to work with compared to ubiquitous 8 bit 4.2.0 codecs with often even more compression.

[interceptor121]

39 minutes ago, jonpais said:

Most of the articles and videos I’ve seen have tended to confirm your findings that 10-bit 150Mbps is to be avoided. 

According to an article written by wolfcrow before the GH5 firmware update, 10-bit 4:2:2 150Mbps only contains 10% more data than 8-bit 4:2:0 100Mbps, not nearly enough to realize a difference in quality.

He further calculated that 10-bit 4:2:2 would require 560Mbps just to be better than 8-bit 4:2:0 100Mbps. 

Yet when he made a video about the GH5 after firmware update v2.1, he recommended shooting 400Mbps internally and not using the Inferno (I think he claimed to see artifacts when recording externally). 

At the same time (I’m relying on my memory here) I seem to recall some people saying that 10-bit 150Mbps did offer a real advantage over 8-bit in that V-log footage was no longer plagued with banding artifacts.

I never shoot 8-bit or V-log, so I can’t say - and that technical stuff is way over my head - but in my limited experience, HLG 10-bit 400Mbps can be manipulated in post quite nicely.

 

I have read that blog and the assumptions are totally incorrect the guys does not understand the difference between all intra and motion interpolation it is just a coincidence he is correct.

if Panasonic had implemented IPB on the 422 10 bit this would have been roughly equivalent to Prores 422 net of motion artefacts but it has not probably because the processing is limited 

 

12 minutes ago, Shirozina said:

Despite the poor specs of the 150mbps LongGOP codec it does hold up very well in grading. Mainly I think due to it's 10bits which make the luma channel near immune to banding unless you apply insane curves or tonal shifts and bends. The main reason to avoid it is the stress it places on most editing rigs trying to decode it on teh fly for smooth playback. The 400mbps codec is better in this respect but still harder on the CPU and GPU than ProRes or DNxHD. Given that most people will expose and color balance near optimally and not treat it as RAW the inadequacies you describe are purely academic and do not translate into real world problems. Until other mainstream manufacturers adopt 10bit 4.2.2 internal codecs the GH5 is well ahead of the game at the moment and a joy to work with compared to ubiquitous 8 bit 4.2.0 codecs with often even more compression.

I don’t see any more strain with this codec than prores the only difference is that you need a program that supports it and not all of them do ir do well

In terms of real life if you don’t beat it this codec performs ok but a 150 MBPS IPB would have been better

[jonpais]

3 minutes ago, interceptor121 said:

I have read that blog and the assumptions are totally incorrect the guys does not understand the difference between all intra and motion interpolation it is just a coincidence he is correct.

if Panasonic had implemented IPB on the 422 10 bit this would have been roughly equivalent to Prores 422 net of motion artefacts but it has not probably because the processing is limited 

 

I've heard many dispute his claims as well. Like I said, that technical stuff's way over my head!

[interceptor121]

To say that the inferno has to be avoided compared to all intra has no foundation

the only reason for prores 422 HQ to look similar to all intra was if the camera is not actually able to produce 10 bits

both prores and avc-intra have no interpolation so there are no possible artifafts as frames are captured individually 

maybe he has a cheap SSD skipping frames on the inferno lol!

[jonpais]

1 minute ago, interceptor121 said:

maybe he has a cheap SSD skipping frames on the inferno lol!

I highly doubt that.

[Shirozina]

I think the title of your post is a bit 'click-bait'  in the context of a consumer cameras internal media they are good enough (more than good enough infact) for most users esp if you expose and color balance fairly well.  

[Deadcode]

3 hours ago, interceptor121 said:

I have spent some time decomposing the codecs and I have had a real bad surprise with the 150 Mbps 422 10 bit codec.

I was hoping this to be an equivalent to Prores 422 but it is really a poor codec implementation and I am not sure the 400 mbps all intra is good enough 

full read of the findings here

https://interceptor121.com/2018/08/13/panasonic-gh5-demystifying-movie-recording-settings/

looks like no 10 bit recording really possible without an external recorder

You mathematically proved that internal codec is sh1t.

Rent an external recorder and show us the difference in real world situation please. 

After seeing this awesome video from Andyax, i think the difference is negligible for most of the users.

 

[KnightsFan]

Nice analysis. I'm glad you actually delve into the numbers and explain the different types of encoding!

However, it would be better if you included files straight out of camera and ran analysis on them, both qualitative and quantitative. Since Prores and AVC intra use different algorithms, simply comparing data rates is not an accurate representation of their fidelity. I  suggest encoding an uncompressed video into Prores, and also into AVC intra, and then running a script to compare the compressed images to the original and get an exact number on how much scene data is lost.

Also, your title is meaningless unless you specify what the 10 bit internal is not good enough for. Is the 10 bit theoretically as good as Prores? Maybe--let's run some tests and see. Is the GH5's specific implementation inferior to external recording to Prores HQ? We need real world tests again. Are the GH5's 10 bit codecs better than any other photo/video hybrid's codecs? Almost certainly. Is it good enough for anything that we used to use GH3 and GH4's for? Of course!

[jonpais]

+1. 

[interceptor121]

3 minutes ago, KnightsFan said:

Nice analysis. I'm glad you actually delve into the numbers and explain the different types of encoding!

However, it would be better if you included files straight out of camera and ran analysis on them, both qualitative and quantitative. Since Prores and AVC intra use different algorithms, simply comparing data rates is not an accurate representation of their fidelity. I  suggest encoding an uncompressed video into Prores, and also into AVC intra, and then running a script to compare the compressed images to the original and get an exact number on how much scene data is lost.

Also, your title is meaningless unless you specify what the 10 bit internal is not good enough for. Is the 10 bit theoretically as good as Prores? Maybe--let's run some tests and see. Is the GH5's specific implementation inferior to external recording to Prores HQ? We need real world tests again. Are the GH5's 10 bit codecs better than any other photo/video hybrid's codecs? Almost certainly. Is it good enough for anything that we used to use GH3 and GH4's for? Of course!

There isn’t a lot of difference between any of the intra codecs as in terms of image compression there hasn’t been great progress over the last 10 years. Prores avid avc intra xavc intra are pretty much the same. The difference comes from spatial compressions in the specific the 10 bit codecs of the GH5 at 4K are weak

if you look at FHD the implementation is better

with regards to the files they are the same scene and nothing exciting there won’t be much difference in the averages regardless of what you shoot quality will drop ad the bitrate is capped

 

[jonpais]

That’s why I switched to Sony’s 8-bit 100Mbps: the GH5’s 10-bit was shiite. ? Do you have actual footage to share shot with the GH5, @interceptor121 ? 

[KnightsFan]

7 minutes ago, interceptor121 said:

There isn’t a lot of difference between any of the intra codecs as in terms of image compression there hasn’t been great progress over the last 10 years. Prores avid avc intra xavc intra are pretty much the same. The difference comes from spatial compressions in the specific the 10 bit codecs of the GH5 at 4K are weak

You could be right. Do you have a source on that, though? Or a test showing that it is the case that hasn't gone through YouTube compression? In your post, you mainly seem to just compare the bitrates.

1 minute ago, interceptor121 said:

with regards to the files they are the same scene and nothing exciting there won’t be much difference in the averages regardless of what you shoot quality will drop ad the bitrate is capped

I don't expect them to be exciting! However, it would give you more credibility if you could show that the numbers do translate to real world differences.

[interceptor121]

1 hour ago, Deadcode said:

You mathematically proved that internal codec is sh1t.

Rent an external recorder and show us the difference in real world situation please. 

After seeing this awesome video from Andyax, i think the difference is negligible for most of the users.

 

The internal codecs are pretty good at FHD they are excellent and at 4K only the 422 is poor 

This is mathematical demonstration am not suggesting what you should use or not or if you want to buy a recorder or not that is your choice but I would not trust youtube as a tool for comparison amongs recorder and internal

there many other benefits in using a recorder other than bitrates a larger screen helps to expose and focus and HDR

10 minutes ago, KnightsFan said:

You could be right. Do you have a source on that, though? Or a test showing that it is the case that hasn't gone through YouTube compression? In your post, you mainly seem to just compare the bitrates.

I don't expect them to be exciting! However, it would give you more credibility if you could show that the numbers do translate to real world differences.

Read the post carefully the comparison is between the codecs of the GH5 themselves and therefore relative and factual it has nothing to do with perceived image quality

take a jpeg of an 8 mp shot compress it to thise sizes and see by yourself if you notice a difference 

Assuming no artefacts from motion interpolation the size of the I frames determines the quality my calculations just extract the information 

it makes me smile that some can post comparison between codecs not even understanding how they work and get away with it so I wanted to provide some back up info

download ffprobe shou few clips of your choice at different settings and the averages won’t change

relative image quality will change but if you can see it or not depends on many factors that are also subjective 

It does not matter what you shoot the codec processes everything exactly the same the GOP structure is fixed and the ratio between the various frame pretty much too with minimal variations so you dont actually need to see what you shoot to test if it is correct or not it is a fact you see from the data

[Deadcode]

4 minutes ago, interceptor121 said:

The internal codecs are pretty good at FHD they are excellent and at 4K only the 422 is poor 

This is mathematical demonstration am not suggesting what you should use or not or if you want to buy a recorder or not that is your choice but I would not trust youtube as a tool for comparison amongs recorder and internal

there many other benefits in using a recorder other than bitrates a larger screen helps to expose and focus and HDR

i have tried to grade 400Mbps 10 bit GH5 VLOG footage and i was impressed how well the codec holds even with extreme grading. I could not see any difference even in 400% magnification. 

[interceptor121]

15 minutes ago, jonpais said:

That’s why I switched to Sony’s 8-bit 100Mbps: the GH5’s 10-bit was shiite. ? Do you have actual footage to share shot with the GH5, @interceptor121 ? 

That is not what I said the footage is not the point this is an analysis of the in camera codecs relative to each other 

if you can’t understand it is fine too

1 minute ago, Deadcode said:

i have tried to grade 400Mbps 10 bit GH5 VLOG footage and i was impressed how well the codec holds even with extreme grading. I could not see any difference even in 400% magnification. 

That is great news as I am wondering if I should buy a v90 memory card as mine fails after 30 seconds. What equipment do you use to grade? Do you have a 10 bit screen? 

when i say the codec is weak I refer to the 150 mbps. If this codec was IBP it would have been roughly equivalent to 500 mbps

 

[KnightsFan]

7 minutes ago, interceptor121 said:

Read the post carefully the comparison is between the codecs of the GH5 themselves and therefore relative and factual it has nothing to do with perceived image quality

You conclude with "If you want to produce genuine 10 bit colour high dynamic range footage you need to buy an external recorded capable of supporting ProRes 422 HQ or there is not game."

That's why I'm suggesting an actual comparison with Prores that uses real world examples. If you just want to compare between internal codecs, that's a misleading conclusion.

10 minutes ago, interceptor121 said:

take a jpeg of an 8 mp shot compress it to thise sizes and see by yourself if you notice a difference

Not all Jpegs are the same, you can actually specify the quality amount and chroma subsampling of a Jpeg when you encode it. Could you specify which type of jpeg you are referring to? Also--what does jpeg have to do with any of this? I'll admit I'm having a hard time understanding why you use it as an example.

16 minutes ago, interceptor121 said:

Assuming no artefacts from motion interpolation the size of the I frames determines the quality my calculations just extract the information

Is that a valid assumption to make, though? My understanding is that motion artifacts is one of the main reasons people use All Intra codecs.

 

To summarize my skepticism: It seems to me that you are judging codec quality purely in terms of adjusted bitrate. If I am misunderstanding, my apologies.