BRAW breaking up in blue skies

The place for questions about shooting with Blackmagic Cameras.
  • Author
  • Message
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

BRAW breaking up in blue skies

PostMon Jun 24, 2019 4:39 pm

Using 3:1 and I'm seeing macro blocking compression artefacts in graduated areas of blue skies. (viewed on 10bit monitor via decklink) I need to do some tests using Q0 and ProRes but it's a bit of a shock to see it.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

Dune00z

  • Posts: 280
  • Joined: Thu Oct 18, 2018 3:40 pm
  • Real Name: Duane Eues

Re: BRAW breaking up in blue skies

PostMon Jun 24, 2019 7:03 pm

Do you have an example we can see?
Offline
User avatar

Jamie LeJeune

  • Posts: 1163
  • Joined: Tue Feb 26, 2013 4:33 am
  • Location: San Francisco

Re: BRAW breaking up in blue skies

PostTue Jun 25, 2019 3:45 am

Never have seen that happen, even in Q5 footage from HD crop

Please share a link to the file so that it can be examined
www.cinedocs.com
http://www.imdb.com/name/nm4601572/
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostTue Jun 25, 2019 8:28 am

The clip is 10gb so not practical to share. I'll do some more tests with Q0 and ProRes when the weather clears and blue skies is back. I've reprocessed it several different ways and the macro blocking is pretty clear when you push the grade. Maybe I'm pushing the grade a bit more than would be normal but I was expecting BRAW to be more robust than this.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostTue Jun 25, 2019 12:23 pm

10GB is a lot of clouds, but you can trim and copy the .braw clip in Media Manager to something manageable, a few seconds. This requires no recompression.
Offline
User avatar

rick.lang

  • Posts: 11875
  • Joined: Wed Aug 22, 2012 5:41 pm
  • Location: Victoria BC Canada

Re: BRAW breaking up in blue skies

PostTue Jun 25, 2019 5:20 pm

At what part of the workflow are you seeing clipping on a 10bit monitor? BRAW is 12bit log.

Have you created optimized media to make it easier to work the Timeline in Resolve Edit and Colour Pages? When I’ve done that it doesn’t look like you can push it much in the grade but that’s misleading as it looks good when you generate your output from the Deliver Page with the correct settings. Make sure you turn on the Advanced options to generate highest quality from your original raw media; never use the optimized or previously rendered media. I’m usually generating h.264 with multiple passes, but h.265 10bit may also be good for you.


Sent from my iPhone using Tapatalk
Rick Lang
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 7:59 am

I'm not using optimised media or Render cache. Still grey skies here so not able to test the different codecs yet. This is an HD section from a UHD frame grab - saved as a Tiff and then converted to JPEG at the highest quality in PS and then examined side by side to make sure there is no jpeg compression adding to the blocking
macro blocking sky_1.1.1.jpg
macro blocking sky_1.1.1.jpg (796.05 KiB) Viewed 10455 times
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline
User avatar

Ulysses Paiva

  • Posts: 813
  • Joined: Sun Sep 01, 2013 8:32 pm
  • Location: Pernambuco, Brasil

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 1:27 pm

For me, it looks like noise, not pixel blocking.
What was your rec settings for this clip? And what was in front of the sensor (lens, filters, etc)?
Ulysses Paiva
UPMD Filmes
http://www.upmdfilmes.com.br
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 2:34 pm

]It's not noise. It's in the sky so it's right at the top of data range. These are the data levels of the ungraded clip
[attachment=0]parade.jpg[/attachment
Attachments
parade.jpg
parade.jpg (67.61 KiB) Viewed 10396 times
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline
User avatar

Robert Niessner

  • Posts: 2637
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 3:12 pm

As we are having 33° Celsius at the moment and a clear blue sky, I did some quick test shots (~12 sec pans) with the UMP G1 in BRAW 3:1, BRAW Q0, ProRes HQ 4.6k, and lossless CDNG.

UMP G1 / FW 6.0 / 24p / 180° / ISO 400
Sigma Cine 18-35mm T2.0 /@ T5.6 / + ND 3

ProRes:
https://we.tl/t-z54FM9WfsX

BRAW 3:1
https://we.tl/t-o3Dimutw39

BRAW Q0
https://we.tl/t-SzUpQJzjtW

CDNG lossless
(is uploading)
Robert Niessner
LAUFBILDkommission
Graz / Austria
---------------------------------------------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 3:23 pm

Many thanks! I'll have a look and report back.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline
User avatar

Robert Niessner

  • Posts: 2637
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 3:38 pm

Here the CDNG lossless link:
https://transferxl.com/08XVrMmd1Ssm6
Robert Niessner
LAUFBILDkommission
Graz / Austria
---------------------------------------------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 4:34 pm

If I push the grade hard on the end part of each sequence where the sky graduation is the greatest I can start to see some roughness. DNG is the worst with BRAW and ProRes looking identical. Interestingly the pattern is quite active and dances around like noise whereas mine is a fixed uneven pattern. Again many thanks for doing this and I'll report back with my own tests when that nice weather makes it up here!
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

CaptainHook

Blackmagic Design

  • Posts: 1586
  • Joined: Wed Aug 22, 2012 4:50 am
  • Location: Melbourne, Australia

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 8:07 pm

John Griffin wrote:The clip is 10gb so not practical to share.

From the RAW panel press "Export Frame" and you can share a single Blackmagic RAW frame easily.
http://www.captainhook.co.nz/blackmagic-cinema-camera-lut/

**Any post by me prior to Aug 2014 was before i started working for Blackmagic**
Offline

pnguyen720

  • Posts: 195
  • Joined: Tue Mar 19, 2019 9:17 pm
  • Real Name: Phong Nguyen

Re: BRAW breaking up in blue skies

PostWed Jun 26, 2019 8:20 pm

DNG is the worst??
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 6:37 am

CaptainHook wrote:
John Griffin wrote:The clip is 10gb so not practical to share.

From the RAW panel press "Export Frame" and you can share a single Blackmagic RAW frame easily.

Thanks
https://we.tl/t-NzRN032lVP
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 9:20 am

pnguyen720 wrote:DNG is the worst??

Open this frame in Photoshop and look at the red channel.
https://we.tl/t-5Qgc7geoVc
I've done the same with BRAW via tiff export from the timeline and the red channel doesn't look nice at all.
Disclaimer - I am applying large tonal adjustments so for most users there shouldn't be a problem but at the same time many users will also choose RAW over ProRes in the expectation that it will take heavy grading in production.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 11:47 am

John Griffin wrote:so for most users there shouldn't be a problem but at the same time many users will also choose RAW over ProRes in the expectation that it will take heavy grading in production.


They may have that expectation, but is it based on anything? Has BMD ever made that claim?

The differences between Prores and raw on this camera are subtle and, in my view, pertain more to workflow. A Prores shot that falls apart when pushed in grading will not survive in raw either.
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 12:55 pm

You can't take any heavier grading than what comes out of camera. Prores is not more resilient than braw, at best, they are equal. Difference between dng and braw on the other hand is in the black-box processing of braw done in camera + braw encoding itself which can introduce macroblocking.

For what there is in the sky and what there isn't, small illustration with some exposure and gamma slamming:
CropperCapture[305].jpg
CropperCapture[305].jpg (166.62 KiB) Viewed 10212 times
I do stuff.
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 2:12 pm

'Slammed' it certainly is!
I'd say BRAW is better than ProRes as it doesn't have an ISO curve baked in. If you expose ProRes at 100ISO and then in your NLE try to make it look like 1000 ISO or vice versa it doesn't work out well but in BRAW it's identical.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 2:43 pm

John Griffin wrote:I'd say BRAW is better than ProRes as it doesn't have an ISO curve baked in. If you expose ProRes at 100ISO and then in your NLE try to make it look like 1000 ISO or vice versa it doesn't work out well but in BRAW it's identical.


You may find that adjustment of raw ISO settings in post is of little practical value in grading shots -- it does nothing which can't be achieved with primary controls, in either raw or Prores, and with greater precision.

And in Prores, the effect of baked-in ISO settings can also be altered in post. For example, shoot a full-range scene at the same f-stop, only changing the ISO, from 100 to 1000. Then grade the shots to look as much alike as possible. You may be surprised at the results.

I think the main advantage of braw over Prores is that it's about the same quality, but with much smaller files. Anyway, it's definitely not the difference between success and failure in the movie business, or between success and failure in the grading suite.
Offline

Dune00z

  • Posts: 280
  • Joined: Thu Oct 18, 2018 3:40 pm
  • Real Name: Duane Eues

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 2:58 pm

The claims here that are saying braw and prores are the same in grading are completely false.

Braw operates like a raw file in that white balance and gain curves are not baked into the file, which makes a big deal depending on how you shoot. Braw allows much, much more versatility when adjusting colors by comparison to braw and a simple white balance stretch test will show this. I also expose to the right for cleaner shadows and braw handles this much better than prores as well in my tests.

Yes, depending on your workflow you may or may not see or use these benefits, but they are there, and braw has much more versatility in post than prores.
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 3:15 pm

Dune00z wrote:The claims here that are saying braw and prores are the same in grading are completely false.


"Completely false"?

I know how braw operates. If you're unable to change color balance with Prores, that's something you might want to work on. And if you're really interested in ISO curves, you might want to undertake the test I proposed above.

Any questioning of the usefulness and/or miracle properties of raw invariably provokes this response, and unfortunately this site is not the venue to settle the dispute with blind tests. I'd just invite anyone interested in the question to look back when BMD first offered a raw camera (2012?) and seek out well-controlled proofs which demonstrate the obvious objective superiority of raw in the BMD cameras, beyond the likes of highlight recovery. If you find any in the past 7 years, let me know.
Offline

Dune00z

  • Posts: 280
  • Joined: Thu Oct 18, 2018 3:40 pm
  • Real Name: Duane Eues

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:04 pm

Its not about changing the color balance, its the degree and control you have to so without causing problems in the grade/correction process. I am fully aware that there are controls outside of a raw tab and nobody is saying that you can't color correct or do adjustments without raw at all.

Saying a 12 bit raw codec has the same latitude as 10 bit prores makes no sense at all and I have done more than enough testing with braw and other raw codecs to prove it does have more latitude to make adjustments without causing major color shifts or other problems. Whether or not you personally see it makes no real difference. Data is data.

This discussion reminds me of arguing with people who claim 8 bit log is just as useful as 10 bit log. The claims here come from people taking 8 bit and 10 bit log, then throwing on a lut and simple color or curve adjustment, then throwing these up as a blind test to prove something. It proves nothing at all.

There is no magic in raw as there is no inherent magic in 10 bit vs 8 bit. More data is more data is more data.
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:13 pm

I tried making ProRes shot at 100ISO look like ProRes shot at 1000ISO at the same exposure and it didn't work very well. When the shadows were pushed in the 100ISO clips it looked very muddy and and conversely when the highlights were expanded in the 1000ISO clips they just didn't separate as nicely all due to the baked in ISO curve. That's not to say there is anything inherently wrong with ProRes but you do get extra room to pull the tones around with BRAW if you need it.
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:18 pm

Dune00z wrote:Saying a 12 bit raw codec has the same latitude as 10 bit prores makes no sense at all


There's reality, and then there's supposition, hope and fantasy. You can argue with proofs, or you can pound the table and say "it stands to reason that...."

What makes sense is demonstrating the truth of your claims, not insisting on the truth of them because on paper 12 is better than 10. We're looking at images, not data. What's measurable, even assuming it's there, may not be visible.

In any event, I don't know what you shoot, but I think it can be assumed that folks investing in a $1300 camera probably don't have a whole lot of production resources. The difference braw and prores, under these circumstances, is beyond trivial.
Offline

John Griffin

  • Posts: 268
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:20 pm

Dune00z wrote:Its not about changing the color balance, its the degree and control you have to so without causing problems in the grade/correction process. I am fully aware that there are controls outside of a raw tab and nobody is saying that you can't color correct or do adjustments without raw at all.

Saying a 12 bit raw codec has the same latitude as 10 bit prores makes no sense at all and I have done more than enough testing with braw and other raw codecs to prove it does have more latitude to make adjustments without causing major color shifts or other problems. Whether or not you personally see it makes no real difference. Data is data.

This discussion reminds me of arguing with people who claim 8 bit log is just as useful as 10 bit log. The claims here come from people taking 8 bit and 10 bit log, then throwing on a lut and simple color or curve adjustment, then throwing these up as a blind test to prove something. It proves nothing at all.

There is no magic in raw as there is no inherent magic in 10 bit vs 8 bit. More data is more data is more data.

BRAW isn't 'RAW'. RAW is the RGB sensor data straight off the AD converter. BRAW goes via another internal processing stage where it gets debayered, converted to YCBCR and compressed. It's better than ProRes but it's a long way off being true RAW. I kind of think BM should have called it B-RES.....
Dell XPS 9560 i7 7700HQ,GTX 1050m, 32gb 1Tb NVME C drive, 1TB SSD 2nd internal as Cache. GTX 1080ti eGPU, 4x 7200rpm 1.5tb HD's in RAID 0 on USB-C as media storage. Win10 x64, Resolve Studio (all latest versions and up to date drivers)
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:21 pm

John Griffin wrote:I tried making ProRes shot at 100ISO look like ProRes shot at 1000ISO at the same exposure and it didn't work very well. When the shadows were pushed in the 100ISO clips it looked very muddy and and conversely when the highlights were expanded in the 1000ISO clips they just didn't separate as nicely all due to the baked in ISO curve. That's not to say there is anything inherently wrong with ProRes but you do get extra room to pull the tones around with BRAW if you need it.


Well, look. I suggested the test because my own results indicate it's next to impossible to tell the difference between the shots after they're graded. And noise balances out -- for example, the noise at 100, thanks to underexposure of areas of the shot, is offset by the greater ISO-related noise at 1000. The former may actually be less.

The variables here are the underlying shot, and the grading. If you want to upload your ungraded shots, I'm willing to give it a go, since I'm the guy making the claim. Beyond that, I doubt we can settle the matter here.
Last edited by John Paines on Thu Jun 27, 2019 4:46 pm, edited 1 time in total.
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 4:45 pm

Dune00z wrote:Saying a 12 bit raw codec has the same latitude as 10 bit prores makes no sense at all and I have done more than enough testing with braw and other raw codecs to prove it does have more latitude to make adjustments without causing major color shifts or other problems. Whether or not you personally see it makes no real difference. Data is data.

You are forgetting the simple fact that it is not 10bit vs 12bit but rather 10bit log vs 12bit lin. And bit depth says nothing about latitude (dynamic range) anyway. And white balance and gain are baked into raw well enough. One comes from sensor response curves, another from analog gain and controlled amount of light reaching sensor.
I do stuff.
Offline
User avatar

Jamie LeJeune

  • Posts: 1163
  • Joined: Tue Feb 26, 2013 4:33 am
  • Location: San Francisco

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 6:01 pm

Hendrik Proosa wrote: You are forgetting the simple fact that it is not 10bit vs 12bit but rather 10bit log vs 12bit lin

BRAW is 12bit log (...and it's worth noting that even mighty ARRIRAW is 12bit log)
www.cinedocs.com
http://www.imdb.com/name/nm4601572/
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 6:13 pm

Jamie LeJeune wrote:BRAW is 12bit log (...and it's worth noting that even mighty ARRIRAW is 12bit log)

What is your source for braw having internal log encoding?
I do stuff.
Offline

youlikeny

  • Posts: 146
  • Joined: Fri May 10, 2019 9:45 pm
  • Real Name: Alessandro Penazzi

Re: BRAW breaking up in blue skies

PostThu Jun 27, 2019 6:30 pm

Hendrik Proosa wrote:
Jamie LeJeune wrote:BRAW is 12bit log (...and it's worth noting that even mighty ARRIRAW is 12bit log)

What is your source for braw having internal log encoding?


Blackmagic says "Images are encoded using a custom non‑linear 12‑bit space designed to provide the maximum amount of color data and dynamic range."

So, if non-linear is the opposite of linear, I guess logarithmic is the only other option...
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostFri Jun 28, 2019 9:04 am

youlikeny wrote:So, if non-linear is the opposite of linear, I guess logarithmic is the only other option...

Gamma curve is also nonlinear. For interest I dug up the opencl kernel code analysis article and there the conclusion was that curve is in part linear, in part quadratic. Where the threshold is set was not said though, probably linear part serves the same function as in rec709, to keep the noisy end at bay.
I do stuff.
Offline

Steven Abrams

  • Posts: 267
  • Joined: Tue Feb 24, 2015 12:43 am
  • Location: LA La Land

Re: BRAW breaking up in blue skies

PostFri Jun 28, 2019 9:09 pm

Hendrik Proosa wrote:
youlikeny wrote:So, if non-linear is the opposite of linear, I guess logarithmic is the only other option...

Gamma curve is also nonlinear. For interest I dug up the opencl kernel code analysis article and there the conclusion was that curve is in part linear, in part quadratic. Where the threshold is set was not said though, probably linear part serves the same function as in rec709, to keep the noisy end at bay.


Yeah, every log function for a camera I've seen has a linear section at the bottom. For noise yes, but also tell me what log(0) is?

Arri logc is
Code: Select all
(x > cut) ? c * log10(a * x + b) + d : e * x + f

So below (and at) "cut" is the linear section of e * x + f.

But for dng and even arri raw files if you look into them, the log function for storing the data is different to the log curve used for grading. Arri has a log encoding for the raw files that is not the above equation. Reverse blackmagics linearization tables from the dngs as you see the log curve is different to the one in resolve for the camera footage.
Offline

Steven Abrams

  • Posts: 267
  • Joined: Tue Feb 24, 2015 12:43 am
  • Location: LA La Land

Re: BRAW breaking up in blue skies

PostFri Jun 28, 2019 9:13 pm

Hendrik Proosa wrote:For what there is in the sky and what there isn't, small illustration with some exposure and gamma slamming:
CropperCapture[305].jpg

What is this image? The forum has made it small so I only see a lot of noise in the sky, not macroblocking. Is it the dng or braw?
Offline

Steven Abrams

  • Posts: 267
  • Joined: Tue Feb 24, 2015 12:43 am
  • Location: LA La Land

Re: BRAW breaking up in blue skies

PostFri Jun 28, 2019 9:16 pm

pnguyen720 wrote:DNG is the worst??

It would appear in this example dng has more problematic artefacts in the sky from my testing here. But it makes sense. There was once posted here a dng of sky at 4:1 and it had VERY bad macroblocking and artefacts. Its a known problem with dng from these cameras.
Offline
User avatar

Robert Niessner

  • Posts: 2637
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: BRAW breaking up in blue skies

PostFri Jun 28, 2019 10:24 pm

Steven Abrams wrote:
pnguyen720 wrote:DNG is the worst??

It would appear in this example dng has more problematic artefacts in the sky from my testing here. But it makes sense. There was once posted here a dng of sky at 4:1 and it had VERY bad macroblocking and artefacts. Its a known problem with dng from these cameras.


The DNG sample I shot is lossless compressed - there can't be any artifacts or macroblocks through compression.
Robert Niessner
LAUFBILDkommission
Graz / Austria
---------------------------------------------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostSat Jun 29, 2019 6:46 am

Steven Abrams wrote:
Hendrik Proosa wrote:For what there is in the sky and what there isn't, small illustration with some exposure and gamma slamming:
CropperCapture[305].jpg

What is this image? The forum has made it small so I only see a lot of noise in the sky, not macroblocking. Is it the dng or braw?

Open it in separate tab or download for original size. It is from braw frame from this same thread.

Macroblocking in braw is usually not visible in color image, it is apparent in chroma components after luma-chroma conversion though.
I do stuff.
Offline

Bunk Timmer

  • Posts: 81
  • Joined: Sat Oct 15, 2016 8:14 pm

Re: BRAW breaking up in blue skies

PostSat Jun 29, 2019 3:08 pm

It looks like the OP shot was made with the ETTR technique. Sky runs between 82 to 96 ire, not a lot of info there. My guess is that causes the grain pattern to pop up under heavy grading as it is absolute abscent in Niessner’s footage which has some 20 ire reserved for the sky. He does have an other problem though…
@Robert Niessner
You might want to check you sensor a lot of weird fixed vertical stripes pop up when going wild with the footage you provided.
Screen Shot 2019-06-29 at 08.17.04.png
sensor problem?
Screen Shot 2019-06-29 at 08.17.04.png (875.51 KiB) Viewed 9750 times
Offline
User avatar

Robert Niessner

  • Posts: 2637
  • Joined: Thu Feb 21, 2013 9:51 am
  • Location: Graz, Austria

Re: BRAW breaking up in blue skies

PostSat Jun 29, 2019 4:41 pm

Well, Bunk - it might be the black shading. On that particular camera it was done several months ago when it was far colder ~ 18° Celsius - when shooting that sample it had 33° Celsius and yesterday I have done another sky test (which was much pur bluer that time) with and without a polarizer and it had 38° Celsius.
Robert Niessner
LAUFBILDkommission
Graz / Austria
---------------------------------------------------------------------
Blackmagic Camera Blog (German):
http://laufbildkommission.wordpress.com

Read the blog in English via Google Translate:
http://tinyurl.com/pjf6a3m
Offline

deezid

  • Posts: 231
  • Joined: Wed Sep 19, 2018 10:38 am
  • Real Name: Dennis Schmitz

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 1:40 am

Downloaded the samples.
This is the thing I keep talking about until John Paines chimes in saying he doesn't see anything wrong and so nobody else ever does either... :mrgreen:

The "artifacts" you are seeing are not caused by the codec but are leftovers caused by fast spatial noise reduction in both Braw and ProRes. I really wonder if BMD ever adds an option to bypass internal noise reduction in Braw/ProRes which may help in some situations.

To reproduce these artifacts just apply fast spatial noise reduction to the cDNG clip at around 0.5, make sure to turn off sharpening in the RAW tab as well.

It will basically look the same then.
Download my 55M Advanced Luts for the Pocket 4K and 6K here:
www.55media.net/55m-p4k-color
Offline

John Paines

  • Posts: 3000
  • Joined: Tue Jul 28, 2015 4:04 pm

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 1:57 am

deezid wrote:This is the thing I keep talking about until John Paines chimes in saying he doesn't see anything so wrong and nobody else ever does either... :mrgreen:


Well, I'm flattered, but I have no following here, and you were invited more than once to provide samples of all the terrible things you were seeing with braw -- the lack of texture, the sharpening and the smoothing, which (allegedly) was making everything look plastic. And do you recall what you produced? Meanwhile, others here have done somewhat more systematic testing, and haven't been able to confirm your observations, at least not under 3x zoom, and even then, you have to strain,.

I'm not a software engineer and don't know what's created the artifacts seen after these shots are subject to stress -- or how the footage from other cameras compares, similarly stressed. But until all this is apparent to the naked eye, it's significance tends to fall in the realm of the theoretical rather than the physical, wouldn't you say?
Offline

deezid

  • Posts: 231
  • Joined: Wed Sep 19, 2018 10:38 am
  • Real Name: Dennis Schmitz

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 2:16 am

John Paines wrote:the lack of texture...and the smoothing

an issue indeed, as well as the "artifacts" in these examples. Can be reproduced as explained earlier.

John Paines wrote:the sharpening

I think that one has been fixed in either later firmwares or Resolve 16 beta.
Cannot reproduce it anymore... which is good and reason why I'm using Braw quite comfortably as my main driver atm.

John Paines wrote:Meanwhile, others here have done somewhat more systematic testing, and haven't been able to confirm your observations, at least not under 3x zoom, and even then, you have to strain,.

I agree. I think things have improved. Using Q0 I don't have many complaints anymore. Can't reproduce the halos anymore which made it look sharpened. Still missing the texture/grittiness of cDNG though - which could be achieved in Q0 quite easily I think by adding at least adding an option like the one mentioned earlier.

John Paines wrote:I'm not a software engineer and don't know what's created the artifacts seen after these shots are subject to stress -- or how the footage from other cameras compares, similarly stressed. But until all this is apparent to the naked eye, it's significance tends to fall in the realm of the theoretical rather than the physical, wouldn't you say?

For the main user I don't see this as a big issue either especially when the delivery takes place on platforms such as YouTube or worse. For a DCP, Blu-Ray etc. the extra level of texture and even grain could be a plus though.
Download my 55M Advanced Luts for the Pocket 4K and 6K here:
www.55media.net/55m-p4k-color
Offline

CaptainHook

Blackmagic Design

  • Posts: 1586
  • Joined: Wed Aug 22, 2012 4:50 am
  • Location: Melbourne, Australia

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 2:32 am

Hi,

When evaluating in Resolve please make sure your timeline is the same resolution as the source files, or that you don't have optimized playback enabled. If you do and the timeline is half the resolution (or smaller) of the source files then Resolve will get the SDK to use a lower quality decode even if the clip settings are set to highest quality (full res). This should not affect renders, but if you zoom into the UI viewer you will only be zooming into a lower quality decode.

Optimized playback settings are in preferences > user settings > playback performance in Resolve 16 (notice Optimized Decode is enabled when set to automatic mode).

optimized playback.png
optimized playback.png (35.41 KiB) Viewed 9535 times


deezid wrote:I think that one has been fixed in either later firmwares or Resolve 16 beta.
Cannot reproduce it anymore... which is good and reason why I'm using Braw quite comfortably as my main driver atm.... Using Q0 I don't have many complaints anymore. Can't reproduce the halos anymore which made it look sharpened


This has not changed FYI.
http://www.captainhook.co.nz/blackmagic-cinema-camera-lut/

**Any post by me prior to Aug 2014 was before i started working for Blackmagic**
Offline

Andrew Kolakowski

  • Posts: 5817
  • Joined: Tue Sep 11, 2012 10:20 am

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 10:34 am

95% people when watching high resolution sources in Resolve 15/16 believe they see full quality, which is of course not true at all.
Making those performance setting "hidden" made it even more confusing.
And amount of youtubers who claim how BRAW is so well performing (and still claiming they watch full quality) is 99%. People believe too much in that they see :D
Last edited by Andrew Kolakowski on Mon Jul 01, 2019 2:36 pm, edited 1 time in total.
Offline

Hendrik Proosa

  • Posts: 740
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 10:42 am

With shameless self-promotion I can suggest using my braw player for viewing the detail in braw files. What it draws is the direct decoded image buffer in full float precision. With custom view shaders one can cook up all kinds of "views" for pixel-peeping, luma-chroma representation and whatnot. Currently it is using linear filtering for scaling (which looks better when downscaled) but I will change that soon so that when zoomed in above 1:1 pixel level it will show actual unfiltered pixels for more elaborate peeping. A sampler tool is also in my mind, for sampling the actual pixel values live.
I do stuff.
Offline

John Brawley

  • Posts: 2149
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: London UK

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 2:33 pm

deezid wrote: Using Q0 I don't have many complaints anymore. Can't reproduce the halos anymore which made it look sharpened.


Does that mean, that it was how you had your settings in Resolve ? A playback issue ?

JB
John Brawley
Cinematographer
London UK
Offline

levisdavis

  • Posts: 170
  • Joined: Mon Apr 22, 2013 11:43 pm
  • Location: Phoenix, AZ

Re: BRAW breaking up in blue skies

PostMon Jul 01, 2019 4:10 pm

Experienced this 3-months ago and wondered why this could ever get passed QC. ( Ended up with no EXT B-roll from the BMPCC4K on this production. ) The spatial noise renders this camera worthless in my professional opinion. For future reference, please don't scrub through any of our INT shots either.

( The client's belief in BM and my belief in the Beta Tester's positive comments on BRAW vs CDNG really did not help anyone on my end. )
Levi Davis
Professional Imagery. Simple.
azcamera.biz
Offline
User avatar

Steve Holmlund

  • Posts: 416
  • Joined: Fri Jun 28, 2013 11:30 pm
  • Location: Montara, California

Re: BRAW breaking up in blue skies

PostTue Jul 02, 2019 2:43 am

Can the OP clarify if this is in fact a playback quality settings issue? Or something else?
Steve
Steve Holmlund
Hobbyist
BMPCC, vintage Rokkor lenses, Olympus 12-40 and 12-100, Panasonic 100-300 II
SmallHD Focus, i7 8700k / GTX 1080
Offline
User avatar

Jamie LeJeune

  • Posts: 1163
  • Joined: Tue Feb 26, 2013 4:33 am
  • Location: San Francisco

Re: BRAW breaking up in blue skies

PostTue Jul 02, 2019 9:19 am

To those in this thread who are cranking the saturation up far higher than anyone ever would on any real shot and then declaring proof of some defect in the BRAW codec, you're not testing what you think you're testing

Raising saturation essentially stretches the contrast on the color channels. Jack up the saturation on any image while watching the parade scope and you'll see it.
On top of that, the blue channel in any digital video signal is always the weakest and noisiest. So, when you go and crank the saturation on some blue sky image, you're just revealing noise and any image artifacts you see in the result aren't any kind of macroblocking codec defect, they are just the same type of image noise you'd see if you pushed the contrast too far or added too much gain or raised the ISO. Try it on a shot from an Alexa and you'll see the same thing, just a different noise pattern/character. It's not some defect of BRAW.
www.cinedocs.com
http://www.imdb.com/name/nm4601572/
Next

Return to Cinematography

Who is online

Users browsing this forum: Bing [Bot] and 61 guests