hvx 200 at ibc

[Jan Crittenden livingston]

[toke lahti] “Thanks for confirmation, Jan!

Can we get another one about horizontal resolution of recorded 1080i/p-picture? “

Not yet.
[toke lahti] ”
I would see that the difference is quite small and need for 4:2:2 with progressive material is greatly reduced. So for the next model, please…”

I think the 4:2:0 only comes to haunt you in color correction. Altough one implementation that we used in the DVCPRO Progressive format, now defunct, was to do a quincunx shift between frames, that is the pixels shifted for the red and blue samples from frame to frame. But straight 4:2:0 can be problematic.

Best,

Jan

Jan Crittenden Livingston
Product Manager, DVCPRO, DVCPRO50, AG-DVX100
Panasonic Broadcast & TV Systems

[Graeme Nattress]

4:2:0 in interlaced video can be quite ghastly, but it’s not too bad in progressive, but I still prefer 4:1:1

Graeme

– http://www.nattress.com – Film Effects and Standards Conversion for FCP

[Peter Corbett]

If NTSC is getting 24P 1080, then us PAL users WILL get 25P. But we never get 24P except in high-end F900’s and the like.

Peter Corbett
Powerhouse Productions
Australia
http://www.php.com.au

[Jean-yves Le moine]

just shoot 25p and we will have 1440X1080 pixel on P2
that is very good for going back to film

jean-yves le moine

temps r

[Toke]

And even better would have been 1440x1080p24 with lesser compression…

[Toke]

I haven’t noticed been haunted when cc’ing progressive 4:2:0.
I’d say that 4:2:2 is only needed with interlaced material of FX pictures.

Problems with cc’ing interlaced 4:2:0 are because there’s one color at lines 1 & 3 and other color at lines 2 & 4.
So there’s no really continuation in colors and this problem is tried to solve by very agressive low pass filtering, which reduces the real chroma resolution greatly.

There’s no such problems with progressive scanning. 4:2:0 just has a little bit less resolution in chroma channels which isn’t usually even noticable in natural pictures. 2 x 2 chroma blocks begin to show up only when you don’t have enough tones and that’s why we need more than 8bit colors. And usually the DCT blocks begin to show way more before than chroma sub sampling blocks. So there’s a need for more advanced compression that uses adaptive dct block sizes than over a decade old dv compression that even dvcprohd uses.

So, in order to make (progressive) picture quality better, right temporal order would be:
1) more tones
2) better compression
3) 4:2:0 -> 4:2:2 -> 4:4:4
Bitrates can be increased all the way as the storages gets faster and bigger. Jump from (h)dv’s 25Mbps to dvcprohd’s 100Mbps is a great advancement in that field. I hope there’s already a new compression in development that might have mAx 200Mbps, 1920×1080 sq px, adaptive dct blocks and user selectable color depth (8-12b), chroma resolution and compression ratio. So the user could choose eg. 200Mbps 12bits 4:2:2 for rich colored natural picture, or 200Mps 10bits 4:4:4 for fx backslates that has little movement and need less cc, or 100Mbps 8bits 4:2:0 for cheap long event documenting.

But why prefer 4:1:1 with progressive picture?
Even that it seems that it will take another decade before we van get cameras that uses only square pixels, human eye’s perception isn’t so different horizontally than vertically, so square pixels should give better real resolution.
And then again, there will be no 4:1:1 delivery formats. After the death of analog and interlaced video, 4:1:1 is as good as 4:1:0…

[Toke]

“…before we will get cameras that _record_ sq px”, I was trying to say.

[Graeme Nattress]

I find 4:1:1 more ameanable to algorithmic improvement than progressive 4:2:0, but really, there’s nothing in it. Interlaced 4:2:0 should be avoided at all costs.

Graeme

– http://www.nattress.com – Film Effects and Standards Conversion for FCP