Ursa RAW 4:1 - Bright Vertical lines that are not FPN

I shot RAW 4:1 at 4.6K. All seemed good until I uploaded the files in Davinci. The footage had a very odd issue. Vertical lines, that are not FPN.

These are bright vertical lines, that can be seen in the brightly exposed areas of the sky, after applying a contrast curve.

DNG is provided below.

I checked project settings in Davinci, to go as "full resolution decoding" and being sure the timeline and delivery format is also 4608x2592

I will do another test tomorrow, comparing the UM 4.6K with the UM Pro.

Any info on the matter will be appreciated.

Update:

as stated by a Blackmagic rep, this issue happens because of the RAW 4:1 compression, but only on rare occasions under a heavy contrast curve.

Case closed.

Here is a DNG.

The vertical lines can be seen after applying a contrast curve.

http://www.newactorsstudio.com/storage/ ... 000010.dng

Emilian I downloaded your RAW

IMHO there's no issue here except no noise reduction and over sharpening... Set your Spatial Threshold to Luma: 2 Chroma: 10 and don't set your sharpening radius less than .47 (.48 optimal)... Problem solved!

Seems to me that the vertical lines are maybe viewer related ? I'm not seeing them here. Have you tried rendering a frame and seeing if they are still there ?

The "dancing pixels" sounds and looks like aliasing from the BMD camera not having an OLPF. This is the kind of situation that looks like this. Very fine moving detail on a high contrast background.

jb

https://www.flickr.com/gp/johnbrawley/023fhr

Thanks for the advice, the sharpening is at default value. Need to do more testing.

Yes I rendered to 12-bit DNxHR and the lines are there.

John, in order to see the vertical lines, please use a color convrrtion to rec709, or just brighten thr image with a curve.

Will try to replicate it again with and without the internal ND, just to be sure.

Yes I rendered to 12-bit DNxHR and the lines are there.

John, in order to see the vertical lines, please use a color convrrtion to rec709, or just brighten thr image with a curve.

Will try to replicate it again with and without the internal ND, just to be sure.

I'm not seeing it ? Heres a REC 709 and the curve lifted ?

https://www.flickr.com/gp/johnbrawley/318j33

JB

Hi John, sorry to mis-leading you. Its not visible in a bright sky, like in your example.

Please try again, without going to REC709, just use a contrast curve like that:

http://www.newactorsstudio.com/storage/umpro_raw.png

If you still dont see it, then it must be some display setting in Davinci.

Hi John, sorry to mis-leading you. Its not visible in a bright sky, like in your example.

Please try again, without going to REC709, just use a contrast curve like that:

http://www.newactorsstudio.com/storage/umpro_raw.png

If you still dont see it, then it must be some display setting in Davinci.

Yeah I'm not seeing it. Here's your curve both with and without a straight 709 transform.

https://www.flickr.com/gp/johnbrawley/KD0bV4

https://www.flickr.com/gp/johnbrawley/514p98

I'd also say, that curve is a pretty extreme grade...

JB

Alright John, thanks for the feedback! So it is some kind of a display issue.

Will try to find anything related to "quality" inside Davinci.

Emilian

I see it too. when boosting up the contrast and bring the sky down there is some strangeness. its also visible in BM-Film Colour and Gamma.

Wow alright so I am worried again...

Could not find any "quality" settings in Davinci, aside from Project / timeline resolution and "Full" decoding, which is on by default.

I didnt have time to do more tests today, but tomorrow I will surely compare the UM Pro with the UM46 and also RAW 4:1 vs RAW Full.

Here is the whole sequence from the 1st test, if its worth looking at:

http://www.newactorsstudio.com/storage/ ... 2_C005.rar

Alright guys, so I did test both cameras in both settings:

Ursa Pro in RAW and RAW 4:1

Ursa 4.6K in RAW and RAW 4:1

I shot the same bright scene with blue sky. I cannot see the vertical lines in any of the files.

I cannot reproduce the problem.

So I am back at square one. Maybe it was a 1 time event, that the UM Pro sensor was a bit off.

To me this looks like a compression issue, the trees will stress any codec. did you do the new test also with content that goes hard on the compression like trees, grass or floating water with lots of waves. Although its RAW and its only 4:1 it should not produce fragments like this. What was the f-stop on the original shot?

maybe its a one time issue and something in the camera did not boot correctly.... things happen....

Yes - the same scene as before.

The compression artifacts show up only after a very strong contrast curve, which I guess may be expected even in RAW. Even in full RAW as well as 4:1.

I cannot see the vertical lines now.

Here are the files from the 2nd test:

http://www.newactorsstudio.com/storage/ ... D_test.rar

just examined the second test you did. the stripes only show up in the 4:1 compressed image i cant see it in the full raw. so it is a a compression issue.
just out of curiosity i compressed the full-raw with slimraw in 4:1 and 7:1 Compression. the 4:1 looks pretty o.k. the 7:1 is similar compressed as the 4:1 out of the camera. but its clear they could improve the incamera raw compression a lot!
here are the slimraw compressed dng's
https://drive.google.com/file/d/1fFdqNP ... sp=sharing

Hi Thom, thanks for the reply.

I still cannot see the stripes on the second test.

But they are really visible on the 1st test.

As far as I know, the nature of the "compressed raw" is about the same, as in a ZIP archive? That should be "lossless", just taking more time to de-compress. Or am I wrong?

[CaptainHook]

Raw 3:1 and 4:1 are "visually lossless", but can potentially show artefacts when "stressed" in post processing and scene complexity like you've seen/done here. There is the lossless option in camera if you would prefer but comes at the expense of much larger file sizes/data rate.

Thanks for the info, mr. Captain!

Can you advise on these bright stripes, could they be provoked by the 4:1 compression? I could not see them in full RAW for sure.

[CaptainHook]

Yes I've seen these before as a result of 4:1 compression - but only on a fairly complex scene and then changing exposure in post and applying strong processing. In fact I was explicitly trying to find artefacts for development reasons, but otherwise in typical shooting and pushing the image as far as I normally would in post it hasn't been an issue for me at all.

As always I would recommend testing the various options in the conditions you will shoot in and decide the best option for you.

Alright Captain, thanks for clarifying the matter!

For development purposes, here is the DNG for testing.

http://www.newactorsstudio.com/storage/ ... 000010.dng

thanks Captain Hook for clearing up the matter.

Is there a way and a will to imporove the internal compression quality? the uncompressed dng that i compressed to 4:1 with slimraw does not show any of those artifacts.