no, i see difference, but only on red object's edges.
I render the same footage to 4:2:0 720p and 4:2:2 720p, and importing back to timeline, and mute one track.

So why do you say it's only size reduction?

To what format are you rendering your timeline that you see a difference?

keep in mind that source is 1080 4:2:0 and i render to 720p

Sorry for the cross-edit. It's a better question now.

Format is: High profile, High level Mpeg2 - 35mbit 1280x720 50p
One time 4:2:0, and then 4:2:2

So upload your full resolution screen shots somewhere. it would be interesting to compare.

ok, tomorrow

[I]"But why is red so different than other colors?
On nature scenes (blue sky, green bottom) i don't see any difference at all"[/I]

Off the top of my head because red is the most depreciated colour in the Y',Cb,Cr matrix used. In a natural scene there isn't saturated colours with hard edges although I have noticed say deep red flowers against green foliage can look a bit pixelated at the edges.

My understanding and experience is that the difference between 4:2:0 and 4:2:2 chroma sampling is just noticeable, the difference between 4:2:2 and 4:4:4 not so unless it's being heavily processed in post. But it's hard to make an apples to apples comparison because all the cameras that'll let one choose between 4:2:0 and 4:2:2 record 4:2:2 at a higher bitrate e.g. 35Mbps and 50Mbps.

Bob.

Bob's last sentence says it afaict.

From a little test I just ran, I can confirm that there is a difference: however I attribute it to the higher bitrate of the 422 profile:
.9 bits per pixel as opposed to .6 bits per pixel with the 420.

If it pleases you, there is nothing wrong with using an outsized container. Size remains a consideration in that case. But if you're going to do that, just as well wrap it as 10 bit 422 and see if that gives a slight reduction in banding.

TBH, I use a 10 bit 422 intermediate from Vegas to Handbrake; because Vegas' pattern dither is lousy.

4:2:0
http://kepfeltoltes.hu/160111/420_www.kepfeltoltes.hu_.png

4:2:2
http://kepfeltoltes.hu/160111/422_www.kepfeltoltes.hu_.png

See the red edges on the right!

I also (think) I'm seeing more detail in the red gradient on the petals.

I'm not working with 4K yet, but followed the posts about how downconversion to 1080 can lift 4:2:0 color to 4:2:2. I haven't looked into the math behind it, but could the 1080 to 720 downconvert give SOME sort of a similar increase in color resolution, even if its not properly 4:2:2?

Petals edges at green background definetly has more detail, or no pixelated edges.
Still not 4:4:4, but much better than 4:2:0 :)

4K downconvert to 1080 should give 4:4:4 colors., because of 4x resolution.

Maybe you'll want to read Glenn Chan's (somewhat murky) article.
Resolution definitely does not equate color enhancement.
It's bits per pixel.

Bitrate and thus bits per pixel do rule the roost. Since 4:2:2 has twice the color data of 4:2:0 to encode, so it had better have a higher bitrate than 420 to not lose quality relatively speaking. Bits per pixel will be higher since the pixel count is the same.

I do not understand you musicvid10.
Why should i read a murky article?
You will find lot of info on the web about Panasonic GH4 4k 420 to 1080 444 conversion.

[url=http://www.eoshd.com/2014/02/discovery-4k-8bit-420-panasonic-gh4-converts-1080p-10bit-444/]

Sorry I was referring to an article on chroma subsampling, not your camera.
And it is a murky subject.