Jump to content

Discovery: 4K 8bit 4:2:0 on the Panasonic GH4 converts to 1080p 10bit 4:4:4


Andrew Reid

Recommended Posts

  • Replies 158
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

Pros are wondering what the benefit of 4K is to them in terms of overall image quality, when mastered and delivered for 2K / 1080p. A lot of work is still shot in 1080p and cameras like the Cano

I think that Cineform Studio free version (now called GoPro Studio 2.0) can do this.  In step 1, convert your file to CFHD (in the same resolution).  It is inflated to a 10-bit file.  Step 2, select f

Finally, a GH4 vid that doesn't look like 480 mush. Absolutely beautiful.

The BM pocket aint consumer level. It's nearly impossible to use for anything pro, let alone for consumer use... The average consumer doesnt spend £300-£400 on media to get an hours worth of recording time, a decent evf, rigging, and lenses to fulfil consumer purposes. The gh4 will deliver everything a consumer needs while also servicing the enthusiast, professional and semi pro sectors.


This makes no sense. I have a pocket, I don't have any rigging on it 99% of the time, nor do I have (or need) an evf. You'll be able to format your cards in camera, but that's not a big deal to me.

The GH4 will require the same amount of $$ for lens as the pocket. And shooting 4k on it, you're gonna pay a pretty penny for your sd cards, too.

The GH4 sounds great, but don't be so damn negative about other cams!
Link to post
Share on other sites
  • Administrators

Previous discussion thread: '?do=embed' frameborder='0' data-embedContent>>

 

Discussion continues here, thank you to all contributors so far (and please let's stay on topic as I feel there's a lot to be gained from this info...)

 

1D C - yes same should apply to that. As for the actual workflow technique, perhaps others can chip in on how they would go about doing this... I have never tried it myself.

Link to post
Share on other sites

This recent DVX discussion seems to suggest that the 4:4:4 you get are not the same as real acquired 4:4:4.   You get more info per pixel but not any new color information.  The Hues that you get in 10bit from initial will not be there, you will only reduce artifacts and noise which is better but not true 10bit.

 

DVX Discussion on 4k 4:2:0 to 1080p 4:4:4

 

 

Until we get footage from the GH4 and we process it this way to clear banding and do green screen work, we will probably never know for sure because the codec implementation could be the deciding factor here.

Link to post
Share on other sites
  • Administrators

What camera does real acquired 4:4:4 though? Not many, especially not at $1999.

 

Let's see the end results and compare to 10bit 1080p 4:2:2 on other cameras and then judge it.

 

The theory holds tight mathematically.

Link to post
Share on other sites

We can try a test with the 4:2:0 1080p we have now on current cameras.      Shoot a clip that has banding or green screen and reduce it to 1/4 1080p or 480x270 pixels on a 16bit After effects timeline after doing the conversion to cineform 10 bit.

 

If the 8bit codec sample is cured of banding and you get a clean key then the theory holds true.    It will be the same average but down-sampled to 1/4 HD res.

Link to post
Share on other sites

should be easily testable as someone with an external recorder could do both types of recording on the gh4.

What camera does real acquired 4:4:4 though? Not many, especially not at $1999.

 

Let's see the end results and compare to 10bit 1080p 4:2:2 on other cameras and then judge it.

 

The theory holds tight mathematically.

Link to post
Share on other sites

So I downloaded the 4K Panasonic footage from youtube.  My editing system, Edius 5.51 does not do 4k. I dumped the 4k footage into Edius AVCHD Converter and converted it to Edius HQ 1080p .avi file.  Is there any benefit to doing this other than being able to edit it?  Keep in mind that Edius 5 is 8-bit.  

Link to post
Share on other sites
  • Administrators

Don't download anything from YouTube. It is mega heavily compressed. YouTube do not serve the originally uploaded file like Vimeo. You will not be able to download a high bitrate file from there let alone a ProRes clip.

 

To test this we need original 100Mbit MOV files from the GH4 or a 4K ProRes master uploaded somewhere.

Link to post
Share on other sites

If you have 4 pixels with the same value (there is noise so this isn't going to happen), you will get the same value afterwards. And that value was obviously just an rounup into 8bit, so it's not the same as the one you would get with 10bit. Obviously it's not the same as taking 10bit directly.

Link to post
Share on other sites

let's make this even simpler, and use a dynamic range to show why you can't always resurrect higher bit depth (even with error diffusion):

 

Assume there are two types of A/D conversion:

 

1 bit (taking on the value of 0 or 1) 2 choices 

2 bit (taking on the values of 0 1 2 3) 

 

Let's assume that analog values are:

 

(0.1) (1.2) (2.0) (2.1) (3.0) (4.1)

 

and that A/D conversion assigns the closes possible digital value. 

 

1 bit A/D conversion becomes:

 

(0) (1) (1) (1) (1) (1)

 

2 bit A/D conversion becomes

 

(0) (1) (2) (2) (3) (3)

 

at half resolution you get either:

 

(0) (2) (3) or (1) (2) (3)

 

either one represents 3 levels of light , which you cannot represent in just 1 bit. 

 

Is this a contrived example, yes. But the point is to try and show they are not mathematically equivalent. 

Link to post
Share on other sites
  • Administrators

OK let's put this in context HurtinMinorKey please. What affect does your theory have on the end result, are we arguing here over a tiny technicality / mathematical proof, or is it a serious issue which will mean we get nowhere near a higher bit depth as David and others are suggesting?

Link to post
Share on other sites

OK let's put this in context HurtinMinorKey please. What affect does your theory have on the end result, are we arguing here over a tiny technicality / mathematical proof, or is it a serious issue which will mean we get nowhere near a higher bit depth as David and others are suggesting?

 

If a pixel in the camera reads 14 bits of data you CANNOT get all of it back once you truncate to 8 bits of data.  

 

Certainly, there will probably be modest color/luma improvements by downsampling fomr 4K, but only within its 8bit dynamic range.  

 

That is to say, IN PRACTICE, if you shoot a scene that falls within the CODEC's dynamic range output you may get better color nuance through average of neighboring pixels.  But if the neighboring pixels are choppy then you're just going to create artifacts.

 

However, you cannot get values from those RAW pixels that were out of the 8bit range they took from the 14bits.

 

I don't mean to be rude, but you're confusing color bits with compression bits.  People who read this thread who think the GH4 is going to do what the Blackmagic cameras, ML RAW, or high end RAW based cameras do should understand this.

Link to post
Share on other sites

It means that this is no magical cure for banding. Once you compress your 8bit range to get out the dark bits into safe zone and highlights you end up with even less range which gives you even more banding, and banding affects more than 4pixel clusters so you will still have it.

Link to post
Share on other sites

OK let's put this in context HurtinMinorKey please. What affect does your theory have on the end result, are we arguing here over a tiny technicality / mathematical proof, or is it a serious issue which will mean we get nowhere near a higher bit depth as David and others are suggesting?

 

I'm not sure exactly about the practical implications. The math seems to indicate that we cannot get 10-bit DR benefits out of 8-bit by downsampling.  

 

Maxotics, I was wondering when you'd show up! You probably have as good a handle on this as anyone. 

 

In Andrew's defense, I think he has been led astray by a Software guy at GoPro.  

 

But the idea of error diffusion got me thinking. While we cannot resurrect the full DR of 10 bit, perhaps we can do a decent job of approximating this precision within the DR limits. 

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...