How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide?

Jon Fairhurst · January 8th, 2013, 01:16 PM

I think the gains of 4:2:2 are marginal, unless you are keying. The bigger advantage (or disadvantage) is the external codec. If it's much better, the images will be much better. And, yes, the extra chroma samples of 4:2:2 might help reduce chroma noise and block artifacts.

Markus Nord · January 9th, 2013, 08:38 AM

I hope that the 4:2:2 will help with less banding, but maybe it needs to get up to 10 bit for that. Hopefully Canon is able to send an clean uncompressed signal AND also keep icons on the 5DIII LCD and record internally. I hope that's the reason why it is taking so long time.

Jon Fairhurst · January 11th, 2013, 08:34 PM

Yeah, banding is related to bits, rather than colorspace bandwidth. You can minimize banding by adding noise, but it has to be added before the 8-bit truncation/rounding.

When grading in post, make sure to work in a 16-bit of floating point mode. Then you can add dither to reduce banding due to the grading process. In fact, I think a lot of people have bashed 8-bit cameras not because of the 8-bit source image but because of the additional banding added when doing 8-bit post work.

Arthur Abramov · February 28th, 2013, 09:21 PM

What's banding?

Jon Fairhurst · March 1st, 2013, 12:23 PM

I think banding is the wrong term for what's being described. I believe that the lack of bit depth on smooth gradients causes an artifact that is called "contouring", which shows hard steps, rather than smooth changes over a nearly flat image, like the sky, a plain wall, or a balloon. It looks similar to the contours that one sees on a topographical map.

I believe that banding is the artifact that we get from rolling shutter - typically light or dark horizontal bands across the image - that happen when the shutter speed doesn't match the frequency of fluorescent lights or when there are strobes/flashes in the scene.

Alex DeJesus · March 8th, 2013, 02:40 PM

Quote:

Originally Posted by Nigel Barker

The basic flaw is that it mirrors the LCD & you cannot remove the overlays.

What do you mean by that? That the HDMI signal was not meant for recording?

Nigel Barker · March 8th, 2013, 04:26 PM

Quote:

Originally Posted by Alex DeJesus

What do you mean by that? That the HDMI signal was not meant for recording?

Correct! At least that is the current situation that the promised new firmware is supposed to correct.

January 8th, 2013, 01:16 PM	#16
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Re: How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide? I think the gains of 4:2:2 are marginal, unless you are keying. The bigger advantage (or disadvantage) is the external codec. If it's much better, the images will be much better. And, yes, the extra chroma samples of 4:2:2 might help reduce chroma noise and block artifacts. __________________ Jon Fairhurst

January 9th, 2013, 08:38 AM	#17
Markus Nord Major Player Join Date: May 2007 Location: Vastervik ,Sweden Posts: 639	Re: How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide? I hope that the 4:2:2 will help with less banding, but maybe it needs to get up to 10 bit for that. Hopefully Canon is able to send an clean uncompressed signal AND also keep icons on the 5DIII LCD and record internally. I hope that's the reason why it is taking so long time. __________________ www.nordproduktion.se https://vimeo.com/nordproduktion

January 11th, 2013, 08:34 PM	#18
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Re: How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide? Yeah, banding is related to bits, rather than colorspace bandwidth. You can minimize banding by adding noise, but it has to be added before the 8-bit truncation/rounding. When grading in post, make sure to work in a 16-bit of floating point mode. Then you can add dither to reduce banding due to the grading process. In fact, I think a lot of people have bashed 8-bit cameras not because of the 8-bit source image but because of the additional banding added when doing 8-bit post work. __________________ Jon Fairhurst

February 28th, 2013, 09:21 PM	#19
Arthur Abramov Regular Crew Join Date: Dec 2008 Location: new york, Israel Posts: 136	Re: How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide? What's banding?

March 1st, 2013, 12:23 PM	#20
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Re: How Much More Image Quality Will 4:2:2 Uncompressed 5d Mk III Really Provide? I think banding is the wrong term for what's being described. I believe that the lack of bit depth on smooth gradients causes an artifact that is called "contouring", which shows hard steps, rather than smooth changes over a nearly flat image, like the sky, a plain wall, or a balloon. It looks similar to the contours that one sees on a topographical map. I believe that banding is the artifact that we get from rolling shutter - typically light or dark horizontal bands across the image - that happen when the shutter speed doesn't match the frequency of fluorescent lights or when there are strobes/flashes in the scene. __________________ Jon Fairhurst