Page 2 of 2

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 5:13 pm
by Claire Watson
Yes, that was 60p. This one is while working in HDR Rec.2020 at 50p, sorry but I don't have any other 60p to show you right now, I shoot 50p here in UK.

Andrew, both black and red squares (grid) look equally sharp on my monitor. What does that tell you?

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 5:54 pm
by Andrew Kolakowski
With 4:2:2 sampling vertical edges should be sharp (pixel perfect) and horizontal not (so you must be not telling truth :D )
Take image 1, insert at 1:1 pixel and preview. You have to look close. You have from left 4:4:4, 4:2:2 and 4:2:0. 4:2:0 messes both dimension (4:2:2 only x), color sampling is not pixel precise in any axis (you have some mixed colors). Interpolation method varies (missing colors have to be interpolated from surrounding pixels), so end results will as well.
Very important- monitor has to display 1:1 pixels as well, so best to do it from project matching monitor native resolution (or you have to tell Eizo not to scale).
My simulation is based on exports to ProRes 444/422 and standard h264 (which is 4:2:0).
In case of typical video we hardly ever see 4:2:2 (or even 4:2:0) imperfections, but they are definitely there.

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 7:35 pm
by roger.magnusson
2D3D4K wrote:Regarding HDMI 4:2:0 at 4K 60fps, here is a link to B&H webpage for the 4K 12g Extreme. Read the specs:

https://www.bhphotovideo.com/c/product/ ... gLBUfD_BwE
I think the B&H specs are old. That card was upgraded to HDMI 2.0. Users who had the old version could buy an upgraded HDMI mezzanine card.

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 7:57 pm
by SkierEvans
2D3D4K wrote:Regarding HDMI 4:2:0 at 4K 60fps, here is a link to B&H webpage for the 4K 12g Extreme. Read the specs:

https://www.bhphotovideo.com/c/product/ ... gLBUfD_BwE



Maybe look here Barry https://www.blackmagicdesign.com/ca/pro ... s/W-DLK-25

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 9:02 pm
by Mario Kalogjera
I imagine one could use Clean feed, at least on MacOS, if the display could be manually forced into HDR EOTF and BT.2020 colorspace mode. I remember seeing some Panasonic TVs from 2016 or 2017 that you could set up, from regular user menus, to PQ EOTF and BT.2020 colorspace manually and separately...of course you'd need to measure the peak nits to know how to set up your Resolve project...I haven't seen Panasonic's menus since, whether they kept their manual modes, but I imagine their OLEDs are beyond the scope of what's being discussed here anyway, due to price...

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 9:26 pm
by Claire Watson
Andrew Kolakowski wrote:With 4:2:2 sampling vertical edges should be sharp (pixel perfect) and horizontal not (so you must be not telling truth :D )
Take image 1, insert at 1:1 pixel and preview. You have to look close. You have from left 4:4:4, 4:2:2 and 4:2:0. 4:2:0 messes both dimension (4:2:2 only x), color sampling is not pixel precise in any axis (you have some mixed colors). Interpolation method varies (missing colors have to be interpolated from surrounding pixels), so end results will as well.
Very important- monitor has to display 1:1 pixels as well, so best to do it from project matching monitor native resolution (or you have to tell Eizo not to scale).
My simulation is based on exports to ProRes 444/422 and standard h264 (which is 4:2:0).
In case of typical video we hardly ever see 4:2:2 (or even 4:2:0) imperfections, but they are definitely there.


I was telling the truth, but mistakenly added a black and then a red grid over background clips.

So I put your screenshots in a UHD project, turned off input scaling - no resizing and set my Eizo screen settings > Picture Expansion to 'Dot by Dot' (1:1).

Not sure what I was to look for? Red and black lines?

What I see, the 4:4:4 at left looks to be four white vertical lines faintly edged with a dark red while the 4:2:2 in center four vertical lines are made up of white and yellow/green. The 4:2:0 at right has no white lines, instead there are pale blue, orange, pale blue and orange lines.

The horizontal lines at left/center are different to any of the vertical lines above. The horizontal lines at right do look the same as the verticals directly above them).

Hope this makes sense, I had to use a magnifier!

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 9:37 pm
by Tom Roper
Andrew Kolakowski wrote:With 4:2:2 sampling vertical edges should be sharp (pixel perfect) and horizontal not (so you must be not telling truth :D )
Take image 1, insert at 1:1 pixel and preview. You have to look close. You have from left 4:4:4, 4:2:2 and 4:2:0. 4:2:0 messes both dimension (4:2:2 only x), color sampling is not pixel precise in any axis (you have some mixed colors). Interpolation method varies (missing colors have to be interpolated from surrounding pixels), so end results will as well.
Very important- monitor has to display 1:1 pixels as well, so best to do it from project matching monitor native resolution (or you have to tell Eizo not to scale).
My simulation is based on exports to ProRes 444/422 and standard h264 (which is 4:2:0).
In case of typical video we hardly ever see 4:2:2 (or even 4:2:0) imperfections, but they are definitely there.


I made sure all scaling and resizing were off, both in Resolve and the UHDTV to make sure Andrew's image was pixel mapped 1 to 1. The vertical stripes were messed. I next rotated the image 90 degrees, the horizontal stripes were clean and delineated 1 pixel wide stripes of pure R,G and B. The conclusion is this was 4:2:2. My monitor is 55 inches but needed a magnifier too.

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 10:08 pm
by Andrew Kolakowski
Everything is fine for both of you- preview is 4:2:2.
Claire- you ment to use picture named 1, but what you described means everything is as it should be for you as well.

Re: 10-bit HDR Workflow Without BM Card?

PostPosted: Sun Jan 03, 2021 10:14 pm
by Andrew Kolakowski
Mario Kalogjera wrote:I imagine one could use Clean feed, at least on MacOS, if the display could be manually forced into HDR EOTF and BT.2020 colorspace mode. I remember seeing some Panasonic TVs from 2016 or 2017 that you could set up, from regular user menus, to PQ EOTF and BT.2020 colorspace manually and separately...of course you'd need to measure the peak nits to know how to set up your Resolve project...I haven't seen Panasonic's menus since, whether they kept their manual modes, but I imagine their OLEDs are beyond the scope of what's being discussed here anyway, due to price...



Some TVs can be put manually into HDR mode (mentioned Panasonic, some Sony I think), but some (eg. LG?) rely just on HDMI flagging. There are other ways to get HDR preview including boxes from HDFury (or AJA) which allow you to inject any required HDMI flagging.
HDR out of Resolve can be done without BM card, but it requires bit of work, knowledge, measuring and validation that your signal is correct (and you have to keep changing GPU refresh to match your project). With some apps it's ways easier- eg. Scratch (fairly current Flame if I'm correct) as they're designed for this.