Is BRAW 444, 422...?

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

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostSun Jul 12, 2020 6:13 pm

antoine wrote:I haven't made any test but theoretically ProRes 4444 will contain more chroma information so better for green screen keying yeah


No.

This is wrong.

Yes I’ve done Chroma work with BRAW.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

cj-adams

  • Posts: 35
  • Joined: Fri Jun 12, 2020 12:43 pm
  • Real Name: CJ Adams

Re: Is BRAW 444, 422...?

PostSun Jul 12, 2020 9:24 pm

What did you find? Are you pretty much limiting to BRAW 3:1 and Q0?
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostSun Jul 12, 2020 9:31 pm

cj-adams wrote:What did you find? Are you pretty much limiting to BRAW 3:1 and Q0?


For VFX And chroma key work 3:1 would be what I would shoot.

JB
Last edited by John Brawley on Mon Jul 13, 2020 1:38 pm, edited 1 time in total.
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Mike Potton

  • Posts: 193
  • Joined: Fri Nov 25, 2016 11:19 pm

Re: Is BRAW 444, 422...?

PostSun Jul 12, 2020 10:45 pm

Always worth remembering that best green screen practice (even lighting, shutter angle considerations for fast moving scenes and depth to avoid spill) are the critical factors for an easy key.

From my limited testing any braw flavour seems to key well, presumably you get a bit more headroom as you reduce compression.

Iv'e shot a few chroma projects in Q5, then converted to Prores 4444 for workflow reasons. No complaints from the VFX team, perfect keying around some rather wavy hair.
Offline

cj-adams

  • Posts: 35
  • Joined: Fri Jun 12, 2020 12:43 pm
  • Real Name: CJ Adams

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 1:53 am

Has anyone ever tried this with the Chromaflex system or chromatte material setup ?

i was told that you needed to watch bitrate cause the billions of cateyes made the sensor think there was a lot of data going on. Which would leave out using Q0 or Q5 .

im just not getting great results. Its almost like it is seeing the screen too well and every imperfection.
.
Offline
User avatar

Uli Plank

  • Posts: 21278
  • Joined: Fri Feb 08, 2013 2:48 am
  • Location: Germany and Indonesia

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:28 am

It’s good practice to keep the background out of focus. Works well with those reflective ones.
No, an iGPU is not enough, and you can't use HEVC 10 bit 4:2:2 in the free version.

Studio 18.6.5, MacOS 13.6.5
MacBook M1 Pro, 16 GPU cores, 32 GB RAM and iPhone 15 Pro
Speed Editor, UltraStudio Monitor 3G, iMac 2017, eGPU
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:31 am

cj-adams wrote:Has anyone ever tried this with the Chromaflex system or chromatte material setup ?

i was told that you needed to watch bitrate cause the billions of cateyes made the sensor think there was a lot of data going on. Which would leave out using Q0 or Q5 .

im just not getting great results. Its almost like it is seeing the screen too well and every imperfection.
.


Are you taking about the reflecting material that works off a LED ring light ?

I’ve always found they don’t work well. There’s an inherent “shadow” with the way the lighting direction works with that system.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 7:46 am

John Brawley wrote:
antoine wrote:I haven't made any test but theoretically ProRes 4444 will contain more chroma information so better for green screen keying yeah

No. This is wrong.

I'm sure you convinced a lot of readers of this topic here :D
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline

cj-adams

  • Posts: 35
  • Joined: Fri Jun 12, 2020 12:43 pm
  • Real Name: CJ Adams

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 9:44 am

Yes that is the one.
it was fine back in the day mine is over 20 years old now.

i think the visual fidelity of modern sensor tech seems to see too well.

guess i will keep trying and worst case grab a fabric one and lights.
cj


John Brawley wrote:
cj-adams wrote:Has anyone ever tried this with the Chromaflex system or chromatte material setup ?

i was told that you needed to watch bitrate cause the billions of cateyes made the sensor think there was a lot of data going on. Which would leave out using Q0 or Q5 .

im just not getting great results. Its almost like it is seeing the screen too well and every imperfection.
.


Are you taking about the reflecting material that works off a LED ring light ?

I’ve always found they don’t work well. There’s an inherent “shadow” with the way the lighting direction works with that system.

JB
Offline
User avatar

Uli Plank

  • Posts: 21278
  • Joined: Fri Feb 08, 2013 2:48 am
  • Location: Germany and Indonesia

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 10:23 am

My answer was only referring to the granularity problem.
Other than that I agree with JB. Even in the days of HD those ring-light solutions were producing a small, but very visible shadow around the foreground. They are a quick solution, but not a pretty one.
No, an iGPU is not enough, and you can't use HEVC 10 bit 4:2:2 in the free version.

Studio 18.6.5, MacOS 13.6.5
MacBook M1 Pro, 16 GPU cores, 32 GB RAM and iPhone 15 Pro
Speed Editor, UltraStudio Monitor 3G, iMac 2017, eGPU
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 1:07 pm

John Brawley wrote:
antoine wrote:I haven't made any test but theoretically ProRes 4444 will contain more chroma information so better for green screen keying yeah


No.

This is wrong.

Yes I’ve done Chroma work with BRAW.

JB


BRAW should be the same when it comes to "chroma info" as ProRes444. If it isn't then it's crap RAW format.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 1:39 pm

antoine wrote:
John Brawley wrote:
antoine wrote:I haven't made any test but theoretically ProRes 4444 will contain more chroma information so better for green screen keying yeah

No. This is wrong.

I'm sure you convinced a lot of readers of this topic here :D


Why don’t you tell us how ProRes 444 has more chroma information than BRAW instead ?

Go ahead and make the case if you can...

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 1:45 pm

cj-adams wrote:Yes that is the one.
it was fine back in the day mine is over 20 years old now.

i think the visual fidelity of modern sensor tech seems to see too well.

guess i will keep trying and worst case grab a fabric one and lights.
cj



I tried them a couple of times back in the day but they were always a pretty poor result.

I like using regular screens and then using a saturated light for them. It used to be a coloured kino tube, but sky panels can do that very easily now. Or an Astera or Titan / coloured LED of choice.

I also never used to like the fact you’re lighting your foreground with coloured LED light either with the older systems. Not very nice for skin tones.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 4:05 pm

John Brawley wrote:
Why don’t you tell us how ProRes 444 has more chroma information than BRAW instead ?

Go ahead and make the case if you can...


Inside each .BRAW frame there is 4x less information for the UV plane than for the luminance plane. Now, that doesn't mean it is pure 4:2:2 I cannot prove that (yet). But it does tell you there is less chroma information than in pure 4:4:4 or RGB codecs
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 4:13 pm

OK then, let's be really precise about this...

antoine wrote:
Inside each .BRAW frame there is 4x less information for the UV plane than for the luminance plane.


Do you maybe mean "YUV" not UV ?

How do you know it's 4x less information. Do you mean file size ? Be specific.

antoine wrote: Now, that doesn't mean it is pure 4:2:2 I cannot prove that (yet). But it does tell you there is less chroma information than in pure 4:4:4 or RGB codecs


Again, you're going to have to spell this out because I don't understand the case you're trying to make.

"pure" 444 is a meaningless phrase. For starters 444 only applies to ENCODED video.

You're arguing that an ENCODED video signal, that is a signal that's literally generated from the sensor data by an algorithm has more information than the BRAW file.

How do you see that as possible ? That ProRes 444 would have more chroma information than the BRAW ?

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Howard Roll

  • Posts: 2506
  • Joined: Fri Jun 03, 2016 7:50 am

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 4:43 pm

"As in the previous example, the resolution of the Cb and Cr images are a quarter of that of the luminance image."

"In the RGB Bayer filter example, the chrominance images have a quarter of the number of pixels as the luminance image."

"Conveniently, the relative resolutions of the luminance image 110 and chrominance images 114R and 114B and spatial alignment of their pixels in the present embodiment means that they can be treated as a YCbCr 420 image data in further processing."

"encoding the luminance values and the difference values to generate image data compatible with a YCbCr 420 format."

"In a preferred embodiment, throughout this process it is preferable that the image data preserves the range of colors captured by the sensor, such that the color range is not limited or clipped, thus enabling the data to be effectively treated in downstream processing as if it were raw image data."

Good Luck
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 4:52 pm

How BRAW is "stored" is meaningless as long as you can recover 100% of original RAW data. If "transformation" is 100% reversible then BRAW is fine, if not then it's not really a proper RAW format. Also if sensor resolution is very close to debayered final file resolution then RAW is not going to "provide real" 4:4:4 data anyway.
If BRAW does an additional data reduction on actual chroma info due to way how it's internally formed then this is another story, but nothing desired for sure. All comes down to the fact tat BM's SDK doesn't provide access to decoded RAW data, but only to already debayered one, so 100% valid test can't be really performed. Fast CinemaDNG company can recover RAW data out of BRAW but not sure if they can guarantee it's 100% "original RAW" as they do use debayered RAW with processing.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:04 pm

If you have a ProRes444 file and BRAW file of the same pixel size, then how on earth can the ProRes 444 somehow have more information.

This malarkey about less chroma resolution ALWAYS gets conflated with a totally seperate encoding process of making 444 or 422 files.

These ratios have NOTHING AT ALL to do with the ratio of Green and Red and BLUE photosites, but people often make a link between BAYER photosite ratios of 2 green and 1 Blue and 1 red with that somehow meaning there’s a similar ratio of chroma reduction to what is inferred by 4:2:2 and 4:4:4

It’s just not what happens. Most debater codecs are at least 80% efficient, AND you’re working from RBG not YUV.

Even better, let’s look at some real world examples. Everyone always makes this case, but no one is ever able to show it visually.

If anything ProRes used to be better than the cDNG raw days because it was in effect “pre-filtered” and has less aliasing. Nothing to do with a lack of chroma information.

Yes, there’s a tiny kernel of truth to the idea that chroma resolution varies within the Bayer array, but it’s not even close to being the often made incorrect conclusions about 444 vs 422. Those ratios have nothing to do with how a sensor is de-mosaiced.

If this was true then a 1920 444 super ampler image derived from a RAW sensor 4K image would key a lot better. It just doesn’t work that way.

JB
Last edited by John Brawley on Mon Jul 13, 2020 5:06 pm, edited 1 time in total.
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:05 pm

antoine wrote:
John Brawley wrote:
Why don’t you tell us how ProRes 444 has more chroma information than BRAW instead ?

Go ahead and make the case if you can...


Inside each .BRAW frame there is 4x less information for the UV plane than for the luminance plane. Now, that doesn't mean it is pure 4:2:2 I cannot prove that (yet). But it does tell you there is less chroma information than in pure 4:4:4 or RGB codecs


Is this not because of Bayern pattern mapping?
Bayer has no 4:4:4 info but 50%,25%,25%. Does this fit at 1:1 to 4:2:2 model? Is not direct Green=Y, Red/Blue=UV ?
Last edited by Andrew Kolakowski on Mon Jul 13, 2020 6:49 pm, edited 1 time in total.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:08 pm

Andrew Kolakowski wrote:
antoine wrote:
John Brawley wrote:
Why don’t you tell us how ProRes 444 has more chroma information than BRAW instead ?

Go ahead and make the case if you can...


Inside each .BRAW frame there is 4x less information for the UV plane than for the luminance plane. Now, that doesn't mean it is pure 4:2:2 I cannot prove that (yet). But it does tell you there is less chroma information than in pure 4:4:4 or RGB codecs


Is this not because of Bayern pattern mapping?
Bayer has no 4:4:4 info but 50%,25%,25%. Does this fit at 1:1 to 4:2:2 model? Is not direct Green=Y, Red/Blue=UV ?

This looks for me like exact Bayer to YUV 4:2:2 mapping:

https://www.baslerweb.com/en/sales-supp ... ork/15182/


But that’s confusing two totally different encoding setups.

One is YUV and one is sensor RGB.

There’s an algorithm that mathematically re-creates the colours. At worst it’s 80% efficient, most of the really good ones (resolve) are into the 90’s....

So it’s a misleading and commonly made mistake to conflate these two ratios that happen to seem like they are talking about the same thing....

This theory often comes out as being the reason you should have more resolution and supersample.

The only way to get 444 worth of REAL and PURE colour in a 1920 is supersample it from a incorrectly called 4:2:2 RAW 4K image.

You’re arguing that the key would be better from a 1920 YUV encoded image over the 4K raw.


JB
Last edited by John Brawley on Mon Jul 13, 2020 5:12 pm, edited 3 times in total.
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:09 pm

John Brawley wrote:If you have a ProRes444 file and BRAW file of the same pixel size, then how on earth can the ProRes 444 somehow have more information.
JB


It easily could be less chroma info if BM wanted it for example to save on bandwidth, but I'm not convinced if BRAW does it.
You can filter tons of informtion - final pixel sizes tells you nothing about it.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:11 pm

Andrew Kolakowski wrote:
John Brawley wrote:If you have a ProRes444 file and BRAW file of the same pixel size, then how on earth can the ProRes 444 somehow have more information.
JB


It easily could be less chroma info if BM wanted it for example to save on bandwidth, but I'm not convinced if BRAW does it.
You can filter tons of informtion - final pixel sizes tells you nothing about it.



I think as other threads here have discussed, there IS pre-filtering of noise with BRAW. A lot of that false noise you get as a by-product of aliasing for example....Noise is the enemy of efficient compression....

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:12 pm

John Brawley wrote:
But that’s confusing two totally different encoding setups.

One is YUV and one is sensor RGB.

There’s an algorithm that mathematically re-creates the colours. At worst it’s 80% efficient, most of the really good ones (revolve) are into the 90’s....

So it’s a misleading and commonly made mistake to conflate these two ratios that happen to seem like they are talking about the same thing....

JB


You don't understand. You basically take Bayer pattern (B&W) data and map it to YUV and encode with some existing/new schema to create new RAW format. This mapping (or possibly more additional processing) is need to avoid Red patent. It's quite simple idea and can be executed in many ways.

Look at Bayer pattern image and YUV 422 schema image.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:13 pm

Andrew Kolakowski wrote:
You don't understand. You basically take Bayer pattern (B&W) data and map it to YUV and encode with some existing/new schema to create new RAW format. This mapping or (maybe more additional processing) is need to avoid Red patent. It's quite simple idea and can be executed in many ways.

Look at Bayer pattern image and YUV 422 schema image.


I do understand what you’re saying.

But saying the ratio of RGB in a Bayer sensor is the same as 4:2:2 is just simplistic. It’s wrong.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Howard Roll

  • Posts: 2506
  • Joined: Fri Jun 03, 2016 7:50 am

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:19 pm

That's not how Braw works. The luminance value is created from all pixels not just the green. Red and Blue pixels are sampled twice, once for luminance then again for the chrominance. That's why it's 4:2:0 and not 2:1:0.

Good Luck
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:23 pm

Howard Roll wrote:That's not how Braw works. The luminance value is created from all pixels not just the green. Red and Blue pixels are sampled twice, once for luminance then again for the chrominance. That's why it's 4:2:0 and not 2:1:0.

Good Luck



Again, these encoded terms of 4:2:0 and 4:2:2 have no relevance when talking about sensor data.

Yes, I know BRAW is doing a partial de-mosaic in camera, however, when you have an encoded video...the 444, 422 or 420, it’s always been de-mosiaced already. It happens in the camera or later in Resolve NO MATTER WHAT CODEC. You can’t use the video until it’s encoded by a de-mosaic process.

They’re not the right terminology to use when talking about sensor / RGB data.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

Howard Roll

  • Posts: 2506
  • Joined: Fri Jun 03, 2016 7:50 am

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:52 pm

[quote="John Brawley"Again, these encoded terms of 4:2:0 and 4:2:2 have no relevance when talking about sensor data.
JB[/quote]

...sensor data from a Bayer sensor you mean? In a 3 chip system the ratio is highly relevant and likely the source of most of this confusion.

Uncompressed bayer is 4:0:0 at the sensor level, Braw is more complicated.

Good Luck
Offline

cj-adams

  • Posts: 35
  • Joined: Fri Jun 12, 2020 12:43 pm
  • Real Name: CJ Adams

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 5:54 pm

i'm starting to understand what you mean.. I guess my question still is.. has BMD sdk talked about what they are getting rid of other than some noise and such.. are they binning color data?
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:12 pm

Howard Roll wrote:
...sensor data from a Bayer sensor you mean? In a 3 chip system the ratio is highly relevant and likely the source of most of this confusion.

Uncompressed bayer is 4:0:0 at the sensor level, Braw is more complicated.


A Bayer sensor as a ratio of red, green and blue pixels ?

There’s no red, green and blue in 4:2:2 or 4:4:4.

The first number is both the luminance, and the colour information of Green. Then the second and third numbers are the colour difference signals of R-Y and B-y, or the “UV” part of YUV.

They aren’t RED and BLUE, they are RED and BLUE when multiplied and subtracted with the Y signal. They have no brightness information.

In one system the luminance is encoded on the Y part of the YUV.

In RGB the luminance is encoded into all the channels. R, G and B all have luminance values.

This makes them fundamentally different and the use of 4:2:2 type terminology to talk about sensor data isn’t correct and creates false impressions.

Here’s a phrase from RED’s education site about Bayer sensors.

“Note: Those with a video encoding background may want to try and apply the 4:2:2, 4:1:1, etc. categorizations to a Bayer sensor, but this terminology is intended for compression methodologies and final images, not the sensors themselves. A 4K Bayer sensor is capable of producing full 4K 4:4:4 RGB files, for example; 4:2:2 is what could be applied to this file afterwards.”

https://www.red.com/red-101/bayer-sensor-strategy

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:15 pm

cj-adams wrote:i'm starting to understand what you mean.. I guess my question still is.. has BMD sdk talked about what they are getting rid of other than some noise and such.. are they binning color data?



They are processing colour information in the same was as any other Bayer based sensor would be de-mosiaced.

What’s different is that it gets partially converted to YUV sooner. But that doesn’t mean the choma resolution is inherently different to if you de-mosaiced it in RESOLVE once you take the card out of the camera, like you would if it’s DNG. You always have to do that step with any raw file. What’s changing is when it happens.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline
User avatar

joe12south

  • Posts: 782
  • Joined: Tue Jan 16, 2018 4:14 pm
  • Location: Nashville, TN
  • Real Name: Joseph Moore

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:19 pm

Somewhere in Australia there is someone laughing their ass off after reading this thread...or maybe they are crying.

If this topic is important to you, go make a camera. If making great images is important to you, stop reading this topic and expend an equal amount of time and energy studying lighting. You'll reap 10x the benefit.
Dedicated curmudgeon. Part-time artiste.
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:42 pm

John Brawley wrote:
Andrew Kolakowski wrote:
You don't understand. You basically take Bayer pattern (B&W) data and map it to YUV and encode with some existing/new schema to create new RAW format. This mapping or (maybe more additional processing) is need to avoid Red patent. It's quite simple idea and can be executed in many ways.

Look at Bayer pattern image and YUV 422 schema image.


I do understand what you’re saying.

But saying the ratio of RGB in a Bayer sensor is the same as 4:2:2 is just simplistic. It’s wrong.

JB


I'm saying that if this is the case then nothing is lost. You have all original RAW data preserved :D
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:45 pm

John Brawley wrote:
Howard Roll wrote:That's not how Braw works. The luminance value is created from all pixels not just the green. Red and Blue pixels are sampled twice, once for luminance then again for the chrominance. That's why it's 4:2:0 and not 2:1:0.

Good Luck



Again, these encoded terms of 4:2:0 and 4:2:2 have no relevance when talking about sensor data.

Yes, I know BRAW is doing a partial de-mosaic in camera, however, when you have an encoded video...the 444, 422 or 420, it’s always been de-mosiaced already. It happens in the camera or later in Resolve NO MATTER WHAT CODEC. You can’t use the video until it’s encoded by a de-mosaic process.

They’re not the right terminology to use when talking about sensor / RGB data.

JB


Exactly, but because BRAW is not "clear" RAW data, but "some" YUV mapped data (de-noised, pre-debayered) then in theory it could be 4:2:2 as well, but I think it's not the case. I'm still more convinced it's full RAW data mapped to YUV signal in some way. 4:4:4 or 4:2:2 etc. terminology as you said is somehow irrelevant here.
Last edited by Andrew Kolakowski on Mon Jul 13, 2020 6:52 pm, edited 1 time in total.
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:49 pm

joe12south wrote:Somewhere in Australia there is someone laughing their ass off after reading this thread...or maybe they are crying.

If this topic is important to you, go make a camera. If making great images is important to you, stop reading this topic and expend an equal amount of time and energy studying lighting. You'll reap 10x the benefit.

Your post is not very informative. Please educate us.
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 6:52 pm

Andrew Kolakowski wrote:
They’re not the right terminology to use when talking about sensor / RGB data.

JB


Exactly, but because BRAW is not "clear" RAW data, but "some" YUV mapped data (de-noised, pre-debayered) then in theory it could be 4:2:2 as well, but I think it's not the case.[/quote]


OK, so then we can agree that this is the wrong terminology...

Anytime you use some “clear” RAW data in any given post platform, what do you think happens to be able to see an image ?

It’s the same process as what’s partially happening in camera.

A de-mosaic.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 7:00 pm

John Brawley wrote:
Here’s a phrase from RED’s education site about Bayer sensors.

“Note: Those with a video encoding background may want to try and apply the 4:2:2, 4:1:1, etc. categorizations to a Bayer sensor, but this terminology is intended for compression methodologies and final images, not the sensors themselves. A 4K Bayer sensor is capable of producing full 4K 4:4:4 RGB files

https://www.red.com/red-101/bayer-sensor-strategy

A GRBG bayer pattern of size 3840x2160 can be used to create an image of 3840x2160 with three R,G,B channels but it does require some kind of extrapolation step, and if the algorithm isn't super smart we could end up with a picture a bit blurry. Probably enough to say "this looks a sharp 1920x1080 picture".
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 7:06 pm

John Brawley wrote:[

“.... A 4K Bayer sensor is capable of producing full 4K 4:4:4 RGB files, for example; 4:2:2 is what could be applied to this file afterwards.”

https://www.red.com/red-101/bayer-sensor-strategy

JB


Of course, but if sensor is the same resolution as output then it's not 100% "original" data, but partially interpolated. This is why to get even better results you need oversampled (for given final resolution) sensors. Then you can have some filtering, etc, as well and still "sharp" final output.
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 7:38 pm

antoine wrote:
A GRBG bayer pattern of size 3840x2160 can be used to create an image of 3840x2160 with three R,G,B channels but it does require some kind of extrapolation step, and if the algorithm isn't super smart we could end up with a picture a bit blurry. Probably enough to say "this looks a sharp 1920x1080 picture".


All BAYER CMOS cameras must do this though. That’s the case with every single camera out there.

We’re using the WRONG terminology to describe this issue.

There’s a MASSIVE difference between 422 and 444 if you’re colour grading and especially if you’re doing keys.

That’s not accurately describing what the difference is here though and on top of that you’re using the wrong terminology.

Let’s talk about the differences but at least use the right words.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline

John Brawley

  • Posts: 4267
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: Los Angeles California

Re: Is BRAW 444, 422...?

PostMon Jul 13, 2020 7:40 pm

Andrew Kolakowski wrote:

Of course, but if sensor is the same resolution as output then it's not 100% "original" data, but partially interpolated. This is why to get even better results you need oversampled (for given final resolution) sensors. Then you can have some filtering, etc, as well and still "sharp" final output.


I concur totally.

In an ideal CMOS sensor, the more you oversample the better. +20% is what I’ve heard from the people that design sensors for a living is a good minimum, but the more the better because you also develop other advantages.

JB
John Brawley ACS
Cinematographer
Currently - Los Angeles
Offline
User avatar

Ulysses Paiva

  • Posts: 994
  • Joined: Sun Sep 01, 2013 8:32 pm
  • Location: Pernambuco, Brasil

Re: Is BRAW 444, 422...?

PostTue Jul 14, 2020 8:08 pm

I couldnt care less about 444/422 chroma subsampling, RGB or YUV, debayering algorithm or any of that stuff as long as it can produce an image that is pleasing and satisfying to help me make my things.

You guys are so naive with this discussion...

I still love Revenge of the Sith. And The Phantom Menace. And it was all much lesser capable cameras than what we got on our hand "for cheap" now. Heavy VFX. Never bothered with the image. Not even todays.
And you guys worried if Braw is good enough. Young padawans...
Ulysses Paiva
Offline

Wayne Steven

  • Posts: 3362
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth

Re: Is BRAW 444, 422...?

PostTue Jul 14, 2020 11:20 pm

Wouldn't that have been shot on the HDC-F950/900 three chip 4:4:4 uncompressed which shoots more data than we are orobably talking here.

It was one of the main cameras that motivated us to change the industry with cameras a fraction of the cost. But, it was three chip 4:4:4 uncompressed which is better than cheap and nasty bayer that got used, for vfx Except three chip prisms had a few limitations, and limited the lens options, so Bayer went forwards.
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them
Offline

Howard Roll

  • Posts: 2506
  • Joined: Fri Jun 03, 2016 7:50 am

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 2:14 am

It doesn't really matter if you're going to roto every frame, the next Star Wars could easily be shot on an iPhone.

Good Luck
Offline

Wayne Steven

  • Posts: 3362
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 4:45 am

Yeah, there are ways around things, and they managed to get something that will auto calculate 4:2:0 into it years ago, and my proposal would auto 3D everything live on something which could fit in a pocket camera. However, there is a spiritual successor to what I wanted to do, by a guy called Eric, for the movie industry. But, it's unlikely to be as small and low energy, but you get what you invest for (same as with cameras).

However, 3chip uncompressed is a quality thing, and suitable for upscaling. You probably could calculate the missing pixels a bit by the low pass filters affect on surrounding pixels (probably something no other human on Earth at the moment, has ever thought about before) and do a bit of HDR highlight recovery. All the tools are there, it is just up to us to realise and figure out what to do with it.
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them
Offline

Wayne Steven

  • Posts: 3362
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 4:59 am

Howard Roll wrote:. if you're going to roto every frame, the next Star Wars could easily be shot on an iPhone.

Good Luck


Now, who's for voting for the next Starwars to be shot on an iPhone in h266? :D
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them
Offline

Wayne Steven

  • Posts: 3362
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 5:23 am

By Justin Bieber, on his personal phone without rig.

As long as the tech makes it look as good as the script of epidsode 7, I supose it might be ok then.


Just watching our illustrious leader of the Australian Broadcast Commision on Meet the Press on ABC News on pause at the moment, and I can see the new 4:2:0 interlace issue (which they probably let be forced on us with the 4:2:2 conversion excuse, because the other version of 4:2:0 was going be too good) on the blocky ABC logo graphic which looks like a reject Commodore 64 graphic, with an overall picture quality of maybe a 30% or less mpeg (what a backwards embarrassment, as much as the couch chair political commentary and structuring we were getting, and this guy wants more funding). Sure, iPhone for Starwars, why not!
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 7:14 am

Ulysses Paiva wrote:I couldnt care less about 444/422 chroma subsampling, RGB or YUV, debayering algorithm or any of that stuff as long as it can produce an image that is pleasing and satisfying to help me make my things.

That's the whole point of discussing this matter.
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline
User avatar

antoine

  • Posts: 497
  • Joined: Sat Nov 24, 2018 2:20 am
  • Real Name: Antoine Dornstetter

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 7:23 am

John Brawley wrote:
antoine wrote:
A GRBG bayer pattern of size 3840x2160 can be used to create an image of 3840x2160 with three R,G,B channels but it does require some kind of extrapolation step, and if the algorithm isn't super smart we could end up with a picture a bit blurry. Probably enough to say "this looks a sharp 1920x1080 picture".


All BAYER CMOS cameras must do this though. That’s the case with every single camera out there.

We’re using the WRONG terminology to describe this issue.

There’s a MASSIVE difference between 422 and 444 if you’re colour grading and especially if you’re doing keys.

That’s not accurately describing what the difference is here though and on top of that you’re using the wrong terminology.

Let’s talk about the differences but at least use the right words.

JB


If you're angry about terminology please create a new word or tell me which one to use.

In all case, it remains true that "4:2:2" could be interpreted as "chroma channels have twice less vertical and horizontal resolution compared to the luminance channel" and this likely the case for .BRAW. Now, maybe it wouldn't be possible (or not practical) to get something better out of the camera/sensor anyway so that's not a big issue
BRAW Studio FREE and Premium for Adobe Premiere Pro and After Effects, with a brand new Desktop .BRAW Player and Color Grader, as well as an automatic White Balance Color Picker tool
Offline

Andrew Kolakowski

  • Posts: 9207
  • Joined: Tue Sep 11, 2012 10:20 am
  • Location: Poland

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 12:54 pm

BRAW is not "final YUV data" so you can't ever tell if it's 4:2:2 or 4:4:4 etc. It's just a data stored as YUV. Without further debayering etc. this data is unusable, so any terminology which applies to typical YUV data is meaningless. What counts is how final data is formed and this is rather 4:4:4.
You can think about YUV as a "carrier" here, not a final signal. It's like sending RAW data over eg. SDI. As long as "carrier" can pass through all RAW data its own parameters are not very important (again- final output counts).
Offline
User avatar

Ulysses Paiva

  • Posts: 994
  • Joined: Sun Sep 01, 2013 8:32 pm
  • Location: Pernambuco, Brasil

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 2:42 pm

Howard Roll wrote:It doesn't really matter if you're going to roto every frame, the next Star Wars could easily be shot on an iPhone.

Good Luck

Now lets agree to have some limits, ok?
:lol:

Lets focus on pleasing and satisfying images. That should rule out some gear. Of course.
Ulysses Paiva
Offline

cj-adams

  • Posts: 35
  • Joined: Fri Jun 12, 2020 12:43 pm
  • Real Name: CJ Adams

Re: Is BRAW 444, 422...?

PostWed Jul 15, 2020 3:14 pm

In addition to pleasing and satisfying is all well and good but... I also would like to include VFX/Post in this discussion.
PreviousNext

Return to Cinematography

Who is online

Users browsing this forum: No registered users and 102 guests