Denoising diminishes braw and prores detail

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

Wayne Steven

  • Posts: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostFri May 08, 2020 3:04 pm

jack0429 wrote:
John Griffin wrote:AFAIK Red cameras patented the concept of the compressed RAW video codec. Not the compression type, not the codec but the concept of combining these so they form a practical video format. This is why BM had to pull it. BRAW although having ‘RAW’ in it’s name (maybe to stick 2 fingers up to Red?) is not actually RAW as it’s partially processed in camera so doesn’t infringe the patent. As soon as you start pulling a RAW data stream straight off the sensor and into a codec without debayering but with compression to make the data size manageable you are into a patent war with RED. I may be wrong and am happy to be corrected but I believe this is the gist of why there will never be a return to cNDG nor probably any codec that sails close to the wind and could be classed ‘legally’ as a raw video format i.e one that extracts all the sensor data.
Damn... Jim Jannard really loves his patents huh. I mean uncompressed cdng technically doesn't infringe on their patents. But judging by how butt hurt they are over everything I do get bm's decision to stay the heck away anything that could possibly earn them a court notice from red.


As far as I know, they do a few things, some way to do it which is maybe obvious, destinations which are definitely obvious, resolutions and framerates which are definitely obvious, and maybe a potential big class action for the examiner who passed it, to a certain degree. As far as I know what's his name from cineform publically developed it before Red, so not really patentable as the notion of compressing Raw, and I talked about doing it with jpeg before cdng did, do not patentable without any of us being a co-signatory. So, I presume they have some specific thing of value, that despite whatever with other things, that companies need, and if they fight other things, they might not get a license, I imagine. However they don't act like it, with the patent specifying ownership above a certain resolution, framerste, compression ratio, all trivial obvious stuff............. Which isn't relevant if you have a genuinely new non obvious technology. So, what's happening, why did Apple get a Bayer expert and go after them and then cave in, or anybody else on the planet. It just occured to me, they have hidden patent properties they at least license, which they potentially could wave at somebody under NDA negotiation and threaten them on other aspects, causing companies to give up. There could be other compression technologies in question to do with better jpeg and others, that would stop people, but who knows? Would love to know what is the truth here?
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostFri May 08, 2020 3:18 pm

I would like to mention something. The old 4k+ action (which has an expensive manual lens cine case mod) could do over 200mb/s inter I think. Thats Braw territory as far as image compression, and I imagine the ambarella codec chip could be trained to do a Braw like codec.

I think BM needs to do extra quality options and better cameras last month, as going on the raw recording Canon last month, I imagine it might be possible that Panosonic could implement ProRes raw on one of it's cameras end of this year when GH6 will have been announced, if they want that challenge the Canon with a low end Bayer internal recording model.
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

John Griffin

  • Posts: 616
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: Denoising diminishes braw and prores detail

PostFri May 08, 2020 3:24 pm

Once BM decide that BRAW is not really 'RAW' and can therefore do in-camera subsampling they can use higher mp sensors and deal with all the issues at once - resolution, NR and moire.
Which is what I said, and even though this is what I was talking about, there is not too much comparison between your actions at the moment compared to some not around here.
Offline

Wayne Steven

  • Posts: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostFri May 08, 2020 3:37 pm

jack0429 wrote:
Ulysses Paiva wrote:I dont mean to offend the OP. Surely no one here does. But the thing is, he comes with a certain claim, others confront it, a proof that doesnt show what he was talking about is shown, confronted, then an unreliable proof is shown but it cant prove the point. The OP insistis in his claims, others have difficulty acknowledging it and confront it. 6 pages later nor the OP could prove his point nor at least show exactly what he is talking about and the others cant agree or disagree with factual arguments. So far, we have only different people looking at this in different ways. Thats the problem of not having shown what you are claiming. Its an endless debate.
The only thing we have that is close to the truth, if not the best take on this and real truth, is from experienced filmmakers like JB. Still, some (less experienced) likes to disgree, still without actual evidence. Still, an endless debate.

I really would like we could get to the bottom of this. Everyone would benefit. And for my personal experience, I have no problems with it and find it amazing
I will try my best to do some tests this weekend. It is not gonna be the most extensive test, due to the limitation of the lockdown. But it should be providing more samples.


Ulysses, he is not a technical person, he is only a gilmsjers who "uses" things made by technical people, only other relevant technical people really understand. You should only rely on people skilled in examining and the truth.

You are inflaming things by saying no evidence is presented after it has been for years, but some don't want to hear, and even just was again before you posted. What gives! I read something true and real proof is in front of me, I feel obliged to agree with it, not deny it and then say there wasn't evidence here. No body should feel obliged to accept that. There is nothing acceptence in wishing something true, except fur something undetermined. I'm sure other regular persons who say inflammatory thought is doing it behind the blocked messages, superiorily. It's outstanding, Robert just posted pictures, even John says there were differences earlier on after speaking to the technical people. Well that's it, shows why I blocked them. It's just so difficult to see such constant noise pollution especially after the point has been shown. There is no point in having a different discussion at all of that is what it constantly brake down to.
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostFri May 08, 2020 4:19 pm

jack0429 wrote:
John Griffin wrote:One test for everyone to see the ‘bigger picture’ with would be to take your video capable stills cameras ( Sony A73, Panasonic S1 etc) and shoot video at the highest quality settings and then at the same settings take a raw frame in the stills settings. Using the best resampling method downsize the frame to that of the video pixel size and compare. You may be shocked to find that BM are not the only manufacturer who are ‘diminishing’ detail (some may even be destroying it) and shock horror if you then think BRAW is doing a pretty good job......
Unfortunately I only have my g7 with me as of now, which does a weird cropped pixel readout for 4k, so that's unfortunately not possible. I'm really just targeting an option to disable the denoiser, since the codec like prores and braw seems to be fine. So I'm really trying to have more options.


The point is that they are getting the speed and compactness people exclaim is so great, by sacrificing information like a consumer codec.. Iv'e suggested down the line that a consumer codec could do an cinematic image if setup for that look, an image just like Braw does. Braw is a bit more flexible then a brief in image, but still the same idea. I consider everything usually when looking into design, I realise things happen. A
In the early days of consumer HD, I laid out a way to break down an image more compared to how it was being done to preserve visual appearence while sacrificing image quality in parts of the image that mattered less, in a preference order list formula, and sent that to one of the VP's in Ambrella, after having posted it in connection with the early JVC HD pocket camera. H264 appearance improved 10x or more in the consumer market, and I presented an improved version here for Braw, in recent years. But even though it involved identifying faces and hair and choosing who you give more compression space too professionally, these are consumer tricks, as Jack is finding, ultimately you need a higher quality option professionaly. It doesn't matter what subjectively a user uses or prefers, the option there is for cases where that is not enough and for those who use it, and of nobody else's business if it doesn't impinge on their preferred use options. There is a win loose war, in a loose loose fashion, going on to anybody that brings this up, by people who really have no right to attack things. If somebody honestly, realistically and unmistakenly doesn't find the current situation helpful for their use, it doesn't matter what anybody else normally says.

You compare the failings of consumer codecs to Braw, as a reason why Braw is alright and professional, is not that useful. We should be looking towards improvements in Braw instead, my shaping preference lists and higher quality modes (and other suggested improvements) are ways to greatly improve ordinary modes of raw and the highest quality. These are the things should be suggesting and improving on.

I'm starting to wonder wherever I should work with somebody on my alternative way of encoding raw that Red can't touch, and maybe significantly better than the Braw mechanism, but forget where I put my technical note on it. Who does David Newman work for now? Anybody else here real good on codecs and coding with patent business experience?
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

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 10:15 am

So here is my second test, (https://drive.google.com/drive/folders/ ... sp=sharing) I shot this test using my sigma 17-50 with metabones speedbooster ultra at f3.5 (sharpest results without halation), and the scene is lit with a bounced godox 60w. The brightness is adjusted using my shutterspeed. The braw is shot at 3:1 because q0 might have dumb down the bitrate due to it being a static scene. At least for me the results from iso 800 on braw is a lot worse then the cdng due to the denoising. Other low contrast textures such as the towel became very muddy, and the texts on the box all were almost rendered unreadable. And the labels on the microphones are also very blurry. And these kind of blurriness is why the denoising appeared to be so detrimental to our purpose of cropping, even at 1.5x you could easily tell that it's a quite muddy image. Would love to know your thoughts.
Offline

John Griffin

  • Posts: 616
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 10:44 am

I'm not seeing anything here different to what Robert did in his test - one persons unacceptable NR and softening is another's acceptable NR and softening. Disregarding the differences between BRAW and CDNG the main issue I would have with a camera creating these images would be the significant moire and edge aliasing and a lens with obvious lateral Chromatic Aberration.
Which is what I said, and even though this is what I was talking about, there is not too much comparison between your actions at the moment compared to some not around here.
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 6:25 pm

Thanks for re-doing your tests.

Few things,

Would make a lot more sense if you made the camera original files rather than PNG files, especially because it can affect how an image is processed. I'm guessing you went through resolve by the file names, but how would I know :-)

Also, again a minor thing, but noise and sharpness perceptions changes on shots in motion over a frame grab. ( I shoot 5-10 second takes when I know others will be downloading)

I would have gotten rid of the speed booster and used a more "known" prime lens as well. To take your optical path out of the equation as well. Optics can certainly influence sharpness and contrast. I've always been pretty skeptical of optical focal reducers (speedboosters)

Skin tones are the real test and again, your tests haven't replicated the useage case you are arguing for, being able to punch in on a shot on someones face. Your test for the scenario you have brought to us all to consider don't match and we're not able to independently look at your post processing methods to see if anything else is being introduced. We can't replicate your process to get to the end results. We can only see your end results.

But assuming all things are equal and the processing is "correct".....

I've had a look at your shots and thank you for that.

I have to say the differences are certainly there, but they're in-line with what I normally would expect between cDNG and BRAW. I see a loss of very fine detail, but I'd disagree that anything is "muddy" which in colour correction terms maybe means something different that what I THINK you're trying to describe which is all to do with sharpness and detail levels.

Muddiness to me infers a lack of contrast and difficulty in finding colour separation, finding tones in the grade to be able to manipulate the image.

I do observe a heck of a lot of moire as well ! You seem to have a super power for finding that.

So far, your images to my experience of shooting, are in line with what I normally expect between BRAW and cDNG.

I'd wonder about adding some sharpness back into the BRAW and see if that made your matches better. If we had your original files we could try that. Even better on someones face and in motion. Blind test anyone ?

I think you're unhappy with the non user adjustable default settings of BRAW and one of the few that prefer the detail with all the baggage that goes with cDNG.

It's a moot point now as I am think we will never see DNG on a BMD camera again, so now we're left with either trying to improve BRAW (which I've no doubt they will do) or moving to another camera with more resolution if you're going to expect to be able to continue to punch in without IQ cost.

You're sure not the first to complain about the loss of detail in BRAW and I know that BMD would have this on their list of things to look at, but I also know, a lot of the issues of cDNG are around encoding "false" information.

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline

WahWay

  • Posts: 131
  • Joined: Tue Apr 07, 2020 11:54 am
  • Real Name: Simon Chan

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 7:22 pm

Yes but there is still cDNG in BMD cameras albeit in stills. Its not like someone saying you cant have it or we will take you to court.
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 8:12 pm

WahWay wrote:Yes but there is still cDNG in BMD cameras albeit in stills. Its not like someone saying you cant have it or we will take you to court.


DNG and cDNG are different.

Compressed cDNG is what BMD have been using for the last 5 years and that’s a problem.

No media is going to keep up with uncompressed DNG and as resolution increases so does the data rate. 4k is already not practical. Let alone 4k+

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline

WahWay

  • Posts: 131
  • Joined: Tue Apr 07, 2020 11:54 am
  • Real Name: Simon Chan

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 8:25 pm

John Brawley wrote:
WahWay wrote:Yes but there is still cDNG in BMD cameras albeit in stills. Its not like someone saying you cant have it or we will take you to court.


DNG and cDNG are different.

Compressed cDNG is what BMD have been using for the last 5 years and that’s a problem.

No media is going to keep up with uncompressed DNG and as resolution increases so does the data rate. 4k is already not practical. Let alone 4k+

JB


I get what you mean but hang on here.
We had loseless cDNG on UM4k and 4.6k for years and we have to rely on expensive CF cards to record to. Now we have USB-C out that can record to cheaper faster SSD and suddenly you are saying its not practical.
Are you saying lossless cDNG is also different to DNG?
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 8:29 pm

WahWay wrote:
I get what you mean but hang on here.
We had loseless cDNG on UM4k and 4.6k for years


But still technically compressed, even the lossless version. They haven't done TRUE uncompressed since the OG BMCC and the data rate is too high. Can't stress that enough.

WahWay wrote:and we have to rely on expensive CF cards to record to.


Often using dual cards just to keep up.

WahWay wrote: Now we have USB-C out that can record to cheaper faster SSD and suddenly you are saying its not practical. This is like one step forward and two step back.


For one camera maybe, the P4K. Maybe. And yeah it isn't practical.

It's not a strategy going forward. DNG is gone. The workflow sucks. It's not a "good" codec. No one processed it properly and I think BMD shot themselves in the foot because a lot of people do not know how to grade it and then would fire up Resolve, also never having used it and then you got endless discussion about how "poor" the colour was from BMD cameras, that skin tones were always brown.

It's just not going to happen, especially not for one single model.

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline

WahWay

  • Posts: 131
  • Joined: Tue Apr 07, 2020 11:54 am
  • Real Name: Simon Chan

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 8:35 pm

I always thought cDNG is CinemaDNG not CompressDNG.
Have to admit that is a new one :o
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 8:39 pm

WahWay wrote:I always thought cDNG is CinemaDNG not CompressDNG.
Have to admit that is a new one :o


viewtopic.php?f=2&t=33575#p199074

Since 2015 :-)

Uncompressed DNG is a lot of data.

Even the lossless version. (which is 30% less than the uncompressed version!)

And as I understand the complication of cDNG use is compression.

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostSun May 10, 2020 10:15 pm

John Brawley wrote:Thanks for re-doing your tests.

Few things,

Would make a lot more sense if you made the camera original files rather than PNG files, especially because it can affect how an image is processed. I'm guessing you went through resolve by the file names, but how would I know :-)

Also, again a minor thing, but noise and sharpness perceptions changes on shots in motion over a frame grab. ( I shoot 5-10 second takes when I know others will be downloading)

I would have gotten rid of the speed booster and used a more "known" prime lens as well. To take your optical path out of the equation as well. Optics can certainly influence sharpness and contrast. I've always been pretty skeptical of optical focal reducers (speedboosters)

Skin tones are the real test and again, your tests haven't replicated the useage case you are arguing for, being able to punch in on a shot on someones face. Your test for the scenario you have brought to us all to consider don't match and we're not able to independently look at your post processing methods to see if anything else is being introduced. We can't replicate your process to get to the end results. We can only see your end results.

But assuming all things are equal and the processing is "correct".....

I've had a look at your shots and thank you for that.

I have to say the differences are certainly there, but they're in-line with what I normally would expect between cDNG and BRAW. I see a loss of very fine detail, but I'd disagree that anything is "muddy" which in colour correction terms maybe means something different that what I THINK you're trying to describe which is all to do with sharpness and detail levels.

Muddiness to me infers a lack of contrast and difficulty in finding colour separation, finding tones in the grade to be able to manipulate the image.

I do observe a heck of a lot of moire as well ! You seem to have a super power for finding that.

So far, your images to my experience of shooting, are in line with what I normally expect between BRAW and cDNG.

I'd wonder about adding some sharpness back into the BRAW and see if that made your matches better. If we had your original files we could try that. Even better on someones face and in motion. Blind test anyone ?

I think you're unhappy with the non user adjustable default settings of BRAW and one of the few that prefer the detail with all the baggage that goes with cDNG.

It's a moot point now as I am think we will never see DNG on a BMD camera again, so now we're left with either trying to improve BRAW (which I've no doubt they will do) or moving to another camera with more resolution if you're going to expect to be able to continue to punch in without IQ cost.

You're sure not the first to complain about the loss of detail in BRAW and I know that BMD would have this on their list of things to look at, but I also know, a lot of the issues of cDNG are around encoding "false" information.

JB
Oh yep, I would love to shoot some far away faces as well, but in times of quarantine, I really couldn't figure out a way to do that in my tiny apartment, there certainly no models that you can book right now. But I would say the jsfilmz sample demonstrated that pretty well. The reason why I did a stills out of resolve is because since the cdng sample is also a still, I felt like if I included the original braw clip it would be a bit unfair, I colorgraded the footage is resolve using the p4kalex lut, which bring a lot of artifact out of the image, so I used that to demonstrate the muddiness on the towel and the boxes for example. I definitely don't want to bring back cdng, it's a clunky codec, I also just really want to improve braw's resolution. So it's good to hear that Blackmagic is set out to improve it.
Offline

Uli Plank

  • Posts: 7139
  • Joined: Fri Feb 08, 2013 2:48 am
  • Location: Germany and Indonesia

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 2:46 am

From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.
Resolve Studio 16.2.2 and Fusion Studio under MacOS Mojave 10.14.6
iMac 2017 Radeon Pro 580 8 GB VRAM and 32 GB RAM
Mac mini 16 GB RAM plus eGFX Breakway Radeon RX 580
Offline

pnguyen720

  • Posts: 363
  • Joined: Tue Mar 19, 2019 9:17 pm
  • Real Name: Phong Nguyen

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 2:56 am

Uli Plank wrote:From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.


Has this been confirmed via release notes or anything?
Offline

Adam Silver

  • Posts: 297
  • Joined: Wed Dec 14, 2016 4:12 pm

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 3:36 am

jack0429 wrote:Oh yep, I would love to shoot some far away faces as well, but in times of quarantine, I really couldn't figure out a way to do that in my tiny apartment, there certainly no models that you can book right now.


So, you're on a 24 hour lockdown and can't go outside at all? You're also saying that you have no friends or family that can help you? You are completely isolated from the world?

I think you could figure this out. Also, I don't recall if you ever told us what you actually do that you need this particular set of specifications. Just curious.
Offline

Uli Plank

  • Posts: 7139
  • Joined: Fri Feb 08, 2013 2:48 am
  • Location: Germany and Indonesia

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 3:58 am

Sounds a tad aggressive, Adam, doesn't it? Rules of lockdown differ quite a bit over the planet.
Resolve Studio 16.2.2 and Fusion Studio under MacOS Mojave 10.14.6
iMac 2017 Radeon Pro 580 8 GB VRAM and 32 GB RAM
Mac mini 16 GB RAM plus eGFX Breakway Radeon RX 580
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:13 am

Uli Plank wrote:From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.
That's good to hear, sounds quite promising for the future of braw.
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:17 am

Adam Silver wrote:
jack0429 wrote:Oh yep, I would love to shoot some far away faces as well, but in times of quarantine, I really couldn't figure out a way to do that in my tiny apartment, there certainly no models that you can book right now.


So, you're on a 24 hour lockdown and can't go outside at all? You're also saying that you have no friends or family that can help you? You are completely isolated from the world?

I think you could figure this out. Also, I don't recall if you ever told us what you actually do that you need this particular set of specifications. Just curious.
Pretty much yeah? It's not a 24 hour lockdown, but with some recent superspreaders near where I live, I definitely do not want to risk people's health just for some simple camera tests lol. My specification is simple, low contrast textures, lighting and far away faces on a wide shot with plenty of details. Like the jsfilmz sample that I've included. Those shots usually shows the most kind of denoising artifacts.
Offline
User avatar

Fabián Aguirre

  • Posts: 91
  • Joined: Sat Jun 10, 2017 4:58 am
  • Location: Mill Valley, California

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:36 am

jack0429 wrote:My specification is simple, low contrast textures, lighting and far away faces on a wide shot with plenty of details. Like the jsfilmz sample that I've included. Those shots usually shows the most kind of denoising artifacts.
I think what Adam means is that we would love to know the kind of work that requires you to fulfill this set of photographic parameters. What kind of work or projects, exactly, do you do?
Fabián Aguirre
Cinematographer / Steadicam Operator
www.theunderstory.co
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:54 am

Fabián Aguirre wrote:
jack0429 wrote:My specification is simple, low contrast textures, lighting and far away faces on a wide shot with plenty of details. Like the jsfilmz sample that I've included. Those shots usually shows the most kind of denoising artifacts.
I think what Adam means is that we would love to know the kind of work that requires you to fulfill this set of photographic parameters. What kind of work or projects, exactly, do you do?
A lot of different situations, in doors, out doors, at night, and never really properly controlled with lighting. I usually do doc type of work so we need minimal footprint and decent crop-ability. Since you never know when to exactly zoom into the perfect framing.
Offline

John Griffin

  • Posts: 616
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 6:11 am

jack0429 wrote:
Fabián Aguirre wrote:
jack0429 wrote:My specification is simple, low contrast textures, lighting and far away faces on a wide shot with plenty of details. Like the jsfilmz sample that I've included. Those shots usually shows the most kind of denoising artifacts.
I think what Adam means is that we would love to know the kind of work that requires you to fulfill this set of photographic parameters. What kind of work or projects, exactly, do you do?
A lot of different situations, in doors, out doors, at night, and never really properly controlled with lighting. I usually do doc type of work so we need minimal footprint and decent crop-ability. Since you never know when to exactly zoom into the perfect framing.

By this description of your working situations the P4k would seem not to be the ideal choice of camera irrespective of codec ‘issues’.
Which is what I said, and even though this is what I was talking about, there is not too much comparison between your actions at the moment compared to some not around here.
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 6:48 am

By this description of your working situations the P4k would seem not to be the ideal choice of camera irrespective of codec ‘issues’.
Not quite, the flexibility of braw in terms of white balance and dynamic range means that we have a ton of safe space while shooting, and the dual native iso and the deep dof also means that we can practically shoot almost where ever we need to without some bulky on camera lights. Put some security 12v battery banks in our pocket, and we are set for an entire day of shooting. And the 4k's braw file size were basically perfect for our size of production, so that's why we chose the p4k. W used to shoot with dvc180 and digibetas, so this style of operation suits us fairly well.
Offline
User avatar

Fabián Aguirre

  • Posts: 91
  • Joined: Sat Jun 10, 2017 4:58 am
  • Location: Mill Valley, California

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 6:55 am

jack0429 wrote:
By this description of your working situations the P4k would seem not to be the ideal choice of camera irrespective of codec ‘issues’.
Not quite, the flexibility of braw in terms of white balance and dynamic range means that we have a ton of safe space while shooting, and the dual native iso and the deep dof also means that we can practically shoot almost where ever we need to without some bulky on camera lights. Put some security 12v battery banks in our pocket, and we are set for an entire day of shooting. And the 4k's braw file size were basically perfect for our size of production, so that's why we chose the p4k. W used to shoot with dvc180 and digibetas, so this style of operation suits us fairly well.


As a documentary DP myself, I’d love to see some of your work here.
Fabián Aguirre
Cinematographer / Steadicam Operator
www.theunderstory.co
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 7:36 am

Fabián Aguirre wrote:
jack0429 wrote:
By this description of your working situations the P4k would seem not to be the ideal choice of camera irrespective of codec ‘issues’.
Not quite, the flexibility of braw in terms of white balance and dynamic range means that we have a ton of safe space while shooting, and the dual native iso and the deep dof also means that we can practically shoot almost where ever we need to without some bulky on camera lights. Put some security 12v battery banks in our pocket, and we are set for an entire day of shooting. And the 4k's braw file size were basically perfect for our size of production, so that's why we chose the p4k. W used to shoot with dvc180 and digibetas, so this style of operation suits us fairly well.


As a documentary DP myself, I’d love to see some of your work here.
I'm sorry but I usually like to remain private on forums and social media. And plus it's nothing impressive really, It's usually for tv stuff. and since it's all in a foreign language, the clients never really uploads them onto the web.
Offline

CaptainHook

Blackmagic Design

  • Posts: 1713
  • Joined: Wed Aug 22, 2012 4:50 am
  • Location: Melbourne, Australia
  • Real Name: Hook

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 7:43 am

pnguyen720 wrote:
Uli Plank wrote:From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.
Has this been confirmed via release notes or anything?

Uli may be referring to a bug we fixed in version in 1.2 - from the read me notes:
• Fixed issue with lower bounds clamp clipping noise.


For the rest of this thread I can't get into the details and won't comment on personal preferences, but there is no "denoising algorithm" in our cameras/BRAW like what you might find in software such as Resolve that's "looking for noise" to reduce etc.

Previously our cameras used a very similar demosaic to the one in Resolve that was used for cDNG processing so that ProRes in camera and DNGs would match closer, this had its issues that we didn't like (especially once compressed 3:1 and higher) and we developed a new one that is now used in camera and in Blackmagic RAW processing through the SDK. This along with other differences favours "smooth edges" and a "3D-like depth" over sharpness and some problems with false detail, aliasing etc. I'm also in the camp that believes resolution and sharpness are different and that they are better ways to achieve resolution that don't end up feeling 'harsh' or 'hard'.

High frequency noise is also reduced which is generally the enemy of dynamic range and compression as mentioned by others, and the reference to Andy's quote earlier also explains that the new demosaic algorithm results in less (high frequency) noise. But there's no "denoising algorithm" to turn off or turn down, everything is a product of the whole system working together to produce nice smooth images with good dynamic range that can be compressed fairly well with good visual results that are robust in typical grading situations. We of course tested (very) many variations of all these elements for a very long time and settled on what is now available.

We do hear you on the feedback and take it into consideration for future updates/products.

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

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 7:45 am

CaptainHook wrote:
pnguyen720 wrote:
Uli Plank wrote:From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.
Has this been confirmed via release notes or anything?

Uli may be referring to a bug we fixed in version in 1.2 - from the read me notes:
• Fixed issue with lower bounds clamp clipping noise.


For the rest of this thread I can't get into the details and won't comment on personal preferences, but there is no "denoising algorithm" in our cameras/BRAW like what you might find in software such as Resolve that's "looking for noise" to reduce etc.

Previously our cameras used a very similar demosaic to the one in Resolve that was used for cDNG processing so that ProRes in camera and DNGs would match closer, this had its issues that we didn't like (especially once compressed 3:1 and higher) and we developed a new one that is now used in camera and in Blackmagic RAW processing through the SDK. This along with other differences favours "smooth edges" and a "3D-like depth" over sharpness and some problems with false detail, aliasing etc. I'm also in the camp that believes resolution and sharpness are different and that they are better ways to achieve resolution that don't end up feeling 'harsh' or 'hard'.

High frequency noise is also reduced which is generally the enemy of dynamic range and compression as mentioned by others, and the reference to Andy's quote earlier also explains that the new demosaic algorithm results in less (high frequency) noise. But there's no "denoising algorithm" to turn off or turn down, everything is a product of the whole system working together to produce nice smooth images with good dynamic range that can be compressed fairly well with good visual results that are robust in typical grading situations. We of course tested (very) many variations of all these elements for a very long time and settled on what is now available.

We do hear you on the feedback and take it into consideration for future updates/products.

Thanks.
Ah I see, sorry for my uneducated ramble, I only has the best wishes for braw going forward, as it's a great codec with a great workflow. So thank you very much for the explanation.
Last edited by jack0429 on Mon May 11, 2020 7:58 am, edited 1 time in total.
Offline

Bunk Timmer

  • Posts: 100
  • Joined: Sat Oct 15, 2016 8:14 pm

Re: Braw is soft and ridden with denoising artifacts

PostMon May 11, 2020 7:48 am

John Griffin wrote:I'm scrolling around the BRAW vs CDNG in Photoshop at 100% on a UHD monitor (not far off my nose) There is plenty of low contrast detail if you look hard enough esp in shadow areas. What I can see is lots more false detail and aliasing in very fine detail in CDNG. Very little if any difference in low contrast areas. Some difference in medium detail like foliage . My overall take is the general level of aliasing and false detail is more obvious and distracting than any advantage in apparent detail in other areas and I'd be wanting to apply something on post to knock this false detail back.

I took the liberty to post some of those areas. Both files were put on a timeline, scaling set to crop and zoom to 1.330. Can you point out to me where you spot the fals details at 100% view? Thanks.
Braw_vs_cDNG_02.jpg
Braw_vs_cDNG_02.jpg (988.55 KiB) Viewed 1203 times
jack0429 wrote:If anyone can somehow shoot some low contrast situations with some relatively far away faces (like the jsfilmz samples) that would be very nice.
@Jack, is this the clip you talk about...
Braw_vs_cDNG_01.jpg
Braw_vs_cDNG_01.jpg (282.29 KiB) Viewed 1203 times

As that is a horrible example, again scaling set to crop and zoom to 1.330. Despite all odds the BRAW file is sharper. In this case clearly caused by the lens.
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Braw is soft and ridden with denoising artifacts

PostMon May 11, 2020 7:52 am

@Jack, is this the clip you talk about...
Braw_vs_cDNG_01.jpg

As that is a horrible example, again scaling set to crop and zoom to 1.330. Despite all odds the BRAW file is sharper. In this case clearly caused by the lens.
I explained earlier that the cdng sample was shot on a sigma fp, through a soft lens, and was focused on the side of the tree branches. So that's why I only included the braw pngs in my folder.
Offline

John Griffin

  • Posts: 616
  • Joined: Sun Aug 23, 2015 4:18 pm

Re: Braw is soft and ridden with denoising artifacts

PostMon May 11, 2020 8:12 am

Bunk Timmer wrote:
John Griffin wrote:I'm scrolling around the BRAW vs CDNG in Photoshop at 100% on a UHD monitor (not far off my nose) There is plenty of low contrast detail if you look hard enough esp in shadow areas. What I can see is lots more false detail and aliasing in very fine detail in CDNG. Very little if any difference in low contrast areas. Some difference in medium detail like foliage . My overall take is the general level of aliasing and false detail is more obvious and distracting than any advantage in apparent detail in other areas and I'd be wanting to apply something on post to knock this false detail back.

I took the liberty to post some of those areas. Both files were put on a timeline, scaling set to crop and zoom to 1.330. Can you point out to me where you spot the fals details at 100% view? Thanks.

I can't see it in your screen grab (how was this generated). I was also looking at it on the timeline when the clips were later uploaded.
Last edited by John Griffin on Mon May 11, 2020 8:40 am, edited 1 time in total.
Which is what I said, and even though this is what I was talking about, there is not too much comparison between your actions at the moment compared to some not around here.
Offline

Adam Silver

  • Posts: 297
  • Joined: Wed Dec 14, 2016 4:12 pm

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 8:13 am

Uli Plank wrote:Sounds a tad aggressive, Adam, doesn't it? Rules of lockdown differ quite a bit over the planet.


Yeah, I'm an ass. I think I'm just frustrated with this thread. Sorry Jack.
Offline

Adam Silver

  • Posts: 297
  • Joined: Wed Dec 14, 2016 4:12 pm

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 8:18 am

jack0429 wrote:
Fabián Aguirre wrote:As a documentary DP myself, I’d love to see some of your work here.
I'm sorry but I usually like to remain private on forums and social media. And plus it's nothing impressive really, It's usually for tv stuff. and since it's all in a foreign language, the clients never really uploads them onto the web.


Umm okay. You can't easily explain what kind of work you do, and you have no samples of ANY work you've done. :lol: :roll:
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 8:20 am

Adam Silver wrote:
Uli Plank wrote:Sounds a tad aggressive, Adam, doesn't it? Rules of lockdown differ quite a bit over the planet.


Yeah, I'm an ass. I think I'm just frustrated with this thread. Sorry Jack.
Nono it's fine, it's just little internet thing. I was a bit over aggressive in the beginning of this thread as well, especially with my uneducated rambles, so I understand. I'm just happy that my objective is resolved and have nothing else to complain about.
Offline

jack0429

  • Posts: 112
  • Joined: Wed Nov 06, 2019 7:55 am
  • Real Name: Jack Jin

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 8:24 am

Adam Silver wrote:
jack0429 wrote:
Fabián Aguirre wrote:As a documentary DP myself, I’d love to see some of your work here.
I'm sorry but I usually like to remain private on forums and social media. And plus it's nothing impressive really, It's usually for tv stuff. and since it's all in a foreign language, the clients never really uploads them onto the web.


Umm okay. You can't easily explain what kind of work you do, and you have no samples of ANY work you've done. :lol: :roll:
I thought explained it? Doc style shooting for tv? It could be news, programs, shows etc. Whoever is willing pay me lol. Asian countries are still very behind on copyright, so they are experimenting with putting their top dramas and things on the web, but overall the kind of small programs I shoot for don't get much attention on the web.
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 3:04 pm

First of all, thanks to Hook for weighing in !

Seems to me like what he is describing in the process is more like a pre filter of high frequency detail.

So by ignoring or pre-filtering the very very fine detail they can use that as a way to get a nicer image to then compress efficiently.

And as MOST high frequency detail is “false” detail due to a lack of OLPF then for many, it’s not that noticeable, and it’s made up info.

So by ignoring that info and only compressing “real” detail with maybe some overlap between the real and not real we get BRAW.

It also sounds like there’s no option to vary this pre-filter because it’s intrinsic to the process. The level has to be set somewhere, weighing up the compression ratios and data rates.

I would also argue that none of the footage posted has come close to demonstrating a problem for me and the way I use BRAW and I think many would agree with me, that the case hasn’t been made, especially by the language used in the original topic title.

I am still expecting when we get an actual replication of the usage being talked about, people’s faces, were going to be squinting into 1.3 crops and not be able to pick much difference.

Thank you Jack for raising it, it’s always worth examining where things stand and each user will have their own perspective on what’s acceptable and what is not.

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline

Wayne Steven

  • Posts: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:01 pm

WahWay wrote:I always thought cDNG is CinemaDNG not CompressDNG.
Have to admit that is a new one :o


+1

https://en.m.wikipedia.org/wiki/CinemaDNG

I thought it was strange they didn't out it in a MXF wrapper to speed up processing, but it's worse than that the wrapper format is a standard option in cdng but BM chose a series of frames in native folders instead, which isn't streamlined. Is there an option to resave a clip into the standard wrappered option in resolve, to see just how slow it is in a contiguous file (defragmentated) compared to Braw?


People don't seem to understand that scientifically there is no such thing as false detail in a true raw format, it is just how the sensor sees the real world. Braw however is not like his the Bayer sensor sees, then you decide just how well you are going to process it. This is master level stuff, not simplified, but for Jack, it is the only option for his use outside of abandoning BM Pocket 4k, which we would prefer he doesn't do.
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:09 pm

jack0429 wrote:Oh yep, I would love to shoot some far away faces as well, but in times of quarantine, I really couldn't figure out a way to do that in my tiny apartment, there certainly no models that you can book right now. But I would say the jsfilmz sample demonstrated that pretty well. The reason why I did a stills out of resolve is because since the cdng sample is also a still, I felt like if I included the original braw clip it would be a bit unfair, I colorgraded the footage is resolve using the p4kalex lut, which bring a lot of artifact out of the image, so I used that to demonstrate the muddiness on the towel and the boxes for example. I definitely don't want to bring back cdng, it's a clunky codec, I also just really want to improve braw's resolution. So it's good to hear that Blackmagic is set out to improve it.


We heard you many times before, a long time ago, at: I don't want to bring back cdng, but improve Braw, like me. But unnoficial opinion and marketing policing have been an issue around here.
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 4:44 pm

CaptainHook wrote:
pnguyen720 wrote:
Uli Plank wrote:From my observations, BMD already has improved BRAW from its first incarnation.
The very first version somehow squeezed noise down against 'superblack', which is not the case anymore.
Has this been confirmed via release notes or anything?

Uli may be referring to a bug we fixed in version in 1.2 - from the read me notes:
• Fixed issue with lower bounds clamp clipping noise.


For the rest of this thread I can't get into the details and won't comment on personal preferences, but there is no "denoising algorithm" in our cameras/BRAW like what you might find in software such as Resolve that's "looking for noise" to reduce etc.

Previously our cameras used a very similar demosaic to the one in Resolve that was used for cDNG processing so that ProRes in camera and DNGs would match closer, this had its issues that we didn't like (especially once compressed 3:1 and higher) and we developed a new one that is now used in camera and in Blackmagic RAW processing through the SDK. This along with other differences favours "smooth edges" and a "3D-like depth" over sharpness and some problems with false detail, aliasing etc. I'm also in the camp that believes resolution and sharpness are different and that they are better ways to achieve resolution that don't end up feeling 'harsh' or 'hard'.

High frequency noise is also reduced which is generally the enemy of dynamic range and compression as mentioned by others, and the reference to Andy's quote earlier also explains that the new demosaic algorithm results in less (high frequency) noise. But there's no "denoising algorithm" to turn off or turn down, everything is a product of the whole system working together to produce nice smooth images with good dynamic range that can be compressed fairly well with good visual results that are robust in typical grading situations. We of course tested (very) many variations of all these elements for a very long time and settled on what is now available.

We do hear you on the feedback and take it into consideration for future updates/products.

Thanks.


Yes, I have noticed an improvement. At the time it was just a 4.6k implementation which we were examining,, and I had hoped that the pocket 4k version would be better due to its noise profile, allowing better compression, and it seems the results are better. However, the writing on those boxes are real details, as was the missing lines on the bank note and solid shaded fishing net, all gone under Braw. So, special case rendering to preserve edges without sharpening, and other such details, would be a minimal improvement. Identifying and rendering other shape and real detail would answer what Jack is locking for. Jack is merely showing that there is real detail there which gets lost, but it seems to have become an anti dng war.

I think it might have been Fujitsu who made edge and text preserving jpeg codec for their car cams around 17 years ago, to preserve these details. However, even if the demosaic works a certain way, surely there are some parameters and alternative coding which could bring an extra level of performance (with more data usage) that could be enabled to get that bit more. However, I think that it will be at similar datarates that Jack want to avoid, which doesn't worry me.

However, a new feature request to make Braw, and auto ingress into Resolve more 4.6k and Arri Alexa like along with the detail improvement above, sensor accuracy in camera, and style modification with the auto ingress. I think customers would be very delighted with that user selectable option, In would be.

However, when is BM going to move to a new codec under Braw? Many years back a double compression advantage under the windows codec was to be had, which was adopted into a version of jpeg, by overlapping boundaries. At 12:1 that night be helpful. Now we have the new jpeg codecs and infrastructure related video, which will probably be a much more reliable Improvement. Thus would allow a bit more quality with no increase in datarate, and maybe a 20:1 mode, which isn't officially supported in that codec, but still.

Another feature request, 14 bit+ for grading to HDR delivery formats.


Thanks Captain Hook.
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Braw is soft and ridden with denoising artifacts

PostMon May 11, 2020 5:06 pm

Bunk Timmer wrote:I took the liberty to post some of those areas. Both files were put on a timeline, scaling set to crop and zoom to 1.330. Can you point out to me where you spot the fals details at 100% view? Thanks.
Braw_vs_cDNG_02.jpg


Bunk, thanks for this. I couldn't see straight late the other night and have up and have been off to other things. This combo frame is convenient.

So, the Braw is on the left and the nicer looking dng on the right.

Most Work, is not cinema, so it's useful to have a little strong image for video, and HDR orientation. I wonder what those drongos in England who did that clip think about this conversation. I'm glad Braw is getting better (as I drape myself in the American Flag, patrioticlly implying that I have see an improvement in Braw..I'm joking, I've had little to do with suggestions for it).

Re-edit;. Actually, an ability to choose look styles in camera in Btaw for different types of work (apart from user custom looks) for monitoring and direct handoff, would be a good suggestion?
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: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 5:24 pm

Anyway, time for new Braw Micro camera. The 64MP Lenovo Legion phone I was waiting for has some better cooling and processing, but ufs3.1, USB 2.0 I think, no card, so even if you can get it to do higher quality 4k or 8k there is limited scope to off load it live. Back down to maybe one new camera. If the new micro is 6k or 8k, with improved Braw, that would be better.
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

Ulysses Paiva

  • Posts: 915
  • Joined: Sun Sep 01, 2013 8:32 pm
  • Location: Pernambuco, Brasil

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 8:24 pm

So... 7 pages later where did we get? The issue couldnt be replicated?
All this story is so strange...
Ulysses Paiva
Offline
User avatar

Fabián Aguirre

  • Posts: 91
  • Joined: Sat Jun 10, 2017 4:58 am
  • Location: Mill Valley, California

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 10:32 pm

jack0429 wrote:I'm sorry but I usually like to remain private on forums and social media. And plus it's nothing impressive really, It's usually for tv stuff. and since it's all in a foreign language, the clients never really uploads them onto the web.


Hey Jack,

I respect and can appreciate your desire for anonymity-- and yet, I would ask that you consider the time many folks have invested here to try to replicate and discuss the alleged issues you find with Blackmagic Raw. This is a wonderful forum full of talented professionals and an invaluable resource, especially when we get a chance to hear from Hook and others who are intimately involved in the design of these technologies.

I love a good geeky thread-- except when it's derailed by the hyperbolic contributions of armchair cinematographers or, worse yet, armchair scientists-- but at some point you've got to wonder whether you're asking the right question.

If, as you say, your work is "nothing impressive" or just for "TV stuff", wouldn't it make more sense to devote some of the energy spent on this thread figuring out how to take advantage of the strengths of Braw? For example, why is it a requirement that a $1200 camera (with features that rival cameras worth many times more) be able to provide you with 1080p crop without any penalty, as opposed to thinking about how to shoot in a way that makes 1080 crops an exception, rather than the main feature?

I get it-- as a full time documentary cinematographer, it's all about flexibility, but it's also about choosing the right tool for the job, and about using that tool in a way that maximizes the capability of that tool. Requiring that a $1200, 4k cinema camera with 13 stops of dynamic range, high speed, internal raw recording, and the industry's best user interface, provide you with a pristine 1080 image (did I mention it's $1200?) tells me that either this is not the best camera for you or that you should re-evaluate the way you approach production so that you don't have to resort to "zooming or reframing in post." Maybe it's because I put so much care into each frame I capture, when possible, or because I hate it when producers tell me "we'll just crop-in in post" instead of taking the time to frame it, or trusting me enough to make the best decision in a given scenario.

Again, I understand the realities of non-fiction cinematography production, but cropping into an image should be the exception and not the rule if what you want is to use a modern high resolution, high fidelity image capture device and actually maximize its potential. If you're finding yourself needing to crop into a 4K frame constantly, even in documentary production, then there are probably other areas of your skillset to improve, even if it's "just" for TV. Ultimately, especially in non-fiction productions, it's the story and what you capture that most often sticks in the minds of the audience, and not whether you should have had a 100mm lens instead of a 35mm.

Here is a short piece I DP'd a little while ago containing a set of vignettes and stories-- a sort of sample "film" that would accompany the release of a book, not quite a short film but also not a trailer. This was shot mostly with an Ursa Mini all over California with a team of two people (second camera was a micro)-- no crops in post and only a couple lenses. No pressure to watch, but if I could go back and do it all over again, my top priority wouldn't be to be able to have an image that would require me to zoom in and reframe in post but, rather, to spend more time with the subjects, to ask different questions, to light or frame a little differently (and, oh, what I would have done to have a BMPCC4K with dual gain and Braw!).



Ultimately, only you know what you need in a camera, but I would urge you to try to keep things in perspective and see just how lucky we are to have these tools at our disposal. I am sure Hook and team have a lot of plans for Braw going forward, and something tells me we are "this" close to being shocked and surprised by whatever new toys and innovations are coming from BMD imminently. Either way, you can be sure we'll be right back here with people complaining it's not good enough.

Cheers,

Fabián
Last edited by Fabián Aguirre on Mon May 11, 2020 10:51 pm, edited 1 time in total.
Fabián Aguirre
Cinematographer / Steadicam Operator
www.theunderstory.co
Offline

Oyvind Fiksdal

  • Posts: 349
  • Joined: Sun Jul 02, 2017 7:12 pm

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 10:44 pm

Great info from Captain hook and others. I have learned more about what makes BRAW, and the ideas behind it. Good stuff!

I remember the first reaction to BRAW, on its arrival. Many of us saw a cinematic image that looked smoother and deeper, in a good way, and showed incredible results even at 12:1 compression. It was ground breaking and salute all over.

But, as all geeks, we started to look into those tiny pixels to see if there was anything not quite right…because this could not be possible. And we found some…errors. There was detail loss going on when compared to cDNG. Unforgiven? Not for all, but many. That’s the core of this whole endless discussion.

If we never had seen cDNG (without OLPF), than BRAW would most likely stand out as an even bigger momentum in the line of RAW formats. A bit unfair actually. Comparing BRAW and Prores would be much more relevant IMO.


This is why I feel BRAW is a game changer, even though there is detail lost compared to cDNG.

First option: Choosing between Prores and cDNG is hard. I always have to weigh the need to shoot RAW. I prefer RAW, but Prores is often used. Because its good enough(mostly), faster, and economic. (Prores also suffer detail loss compare to cDNG)

Second option: Choosing between Prores or BRAW is easy. Its Prores, ONLY when the customer requires it. Meaning, I shoot RAW(BRAW) most of the time.
Offline

John Brawley

  • Posts: 2557
  • Joined: Tue Aug 21, 2012 7:57 am
  • Location: New Orleans, Louisiana

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 11:07 pm

Fabián Aguirre wrote:
Here is a short piece I DP'd a little while ago containing a set of vignettes and stories


What a great film Fabián, well done. Very affecting.

JB
John Brawley
Cinematographer
New Orleans Louisiana
Offline
User avatar

Fabián Aguirre

  • Posts: 91
  • Joined: Sat Jun 10, 2017 4:58 am
  • Location: Mill Valley, California

Re: Denoising diminishes braw and prores detail

PostMon May 11, 2020 11:33 pm

John Brawley wrote:What a great film Fabián, well done. Very affecting.

JB


Thanks so much for watching, John, and for the kind words. It's such a treat to have someone with your level of experience contributing on the forum. Really appreciate it.
Fabián Aguirre
Cinematographer / Steadicam Operator
www.theunderstory.co
Offline

Robert Castiglione

  • Posts: 57
  • Joined: Sat Apr 26, 2014 2:36 am

Re: Denoising diminishes braw and prores detail

PostTue May 12, 2020 2:20 am

Thanks Captain Hook for your response.

"This along with other differences favours "smooth edges" and a "3D-like depth" over sharpness and some problems with false detail, aliasing etc. I'm also in the camp that believes resolution and sharpness are different and that they are better ways to achieve resolution that don't end up feeling 'harsh' or 'hard'."

I think that these are aesthetically good choices and one of the main reasons I am using the camera.

Cheers,

Rob
Offline

Uli Plank

  • Posts: 7139
  • Joined: Fri Feb 08, 2013 2:48 am
  • Location: Germany and Indonesia

Re: Denoising diminishes braw and prores detail

PostTue May 12, 2020 2:22 am

Oyvind Fiksdal wrote:First option: Choosing between Prores and cDNG is hard. I always have to weigh the need to shoot RAW. I prefer RAW, but Prores is often used. Because its good enough(mostly), faster, and economic. (Prores also suffer detail loss compare to cDNG)

Second option: Choosing between Prores or BRAW is easy. Its Prores, ONLY when the customer requires it. Meaning, I shoot RAW(BRAW) most of the time.


Did you know that 90% of the work on Alexa in Germany is shot in ProRes? Got this first hand from the biggest rental service. Just to put things in perspective.
Resolve Studio 16.2.2 and Fusion Studio under MacOS Mojave 10.14.6
iMac 2017 Radeon Pro 580 8 GB VRAM and 32 GB RAM
Mac mini 16 GB RAM plus eGFX Breakway Radeon RX 580
Offline

Wayne Steven

  • Posts: 3082
  • Joined: Thu Aug 01, 2013 3:58 am
  • Location: Earth
  • Warnings: 1

Re: Denoising diminishes braw and prores detail

PostTue May 12, 2020 4:52 am

Thanks Uli, you are one of the few experts who contribute some reality here.


Jack,

But innovation is another level again, which all the arm chair experts here don't really realise, that the sort of deep reality explorer innovative is generally the sort of person that ultimately sees and identifies the problems and the innovative solutions they will use, all generally hidden is the back ground, mocked and mistreated by those which can't see, but making tomorrow. Generally everybody they praise is actually not the person coming up with and designing things they are praising them for. They mistake what they themselves do, as valid to the reality of this innovation, they use, but understand very little from the arm chairs, inflexible and unknowing, instead of flexible and considering the scope of what can be done in their estimation and what could be in exploring design, which are the must haves to correct innovation. They disrespectfully sulk and plague and bully to have to be expert and right, focusing on themselves rather than the expanse not the truth, which they degrade. Should you listen to them, yes, with a grain of salt, to a block of salt and even more, to sift out what's real. But you should do so based on capacity in handling real truth. It's like
film industry, fine framing distribution is as important as focusing which is as important as story which is as important acting, which is as important as lighting and color, which is as important as set (see framing) which is as important wardrobe and location etc. If any of these things are off you have a B grade or less production, which if people are being truthful they would agree and that there are not many who get there. I generally don't comment on those, things, prefering to be positive and encouraging, except for the negative people trying to unrealistically tear into us, who want to act like something they are not and dominate, and I see a sea of orange on my little screen Jack, a word of advise, only appreciate much those that have truely tried to help you and the truth here, not those. You can show them aome appreciation as well, for little bits and pieces of actual truth they have contributed in all they have. I and others saw what you were talking about, what you meant, what the issue was, what you were trying to achieve, and didn't get bogged down in irrelevancies, mistake you wanted but still try to beat you over the head for it, and make war on you for saying something they should understand to defend something they do didn't, and we tried to help. Robert is put in a Stella effort. We might disagree at times, but at least he and some others have an ability to see and contribute. There are three types of people on thus forum, those who contribute, those who remain hidden a lot, and those who negatively miscontrbute, who have to be seen as credible. Even when trying to be careful and correct, their speech is full of mistakes. Make no mistake, accept for me and a few engineers, generally no really talented person is coming here much to contribute, you are largely on your own with a bunch promoting what they want. The top end people are largely working or on other forums.

But you have exposed a floor, the market is mainly video production and what they need, not cinema cameras, the bigger market is stills, which is being eroded by phones, but those other areas need to be addressed just as much, not degraded (except that stills is a handy work and personal thing, that can sell extra cameras, but not the main focus).

Written earlier: I started out successfully contacting companies around 13 or 15, and probably have contributed more to the world economy then anybody here, because seeing and avoiding technical use problems (something often neglected) is a big cost. When working on the OS design project I aimed to reduce computer usage costs to one seventh, major savings to the world economy, and it turned out my design was very mainframe like, which is well ahead of what was out there on personal computers and maybe still is. I still continue to back channel improvements to the most popular OS, which they implement their way. But, if it saves people billions of hours a year, that's great. So, being as trained, experiemced, and knowing as much as I do, does anybody generally really know anything completely? So, it pays to be careful and circumspect in your speech, and to limit it to what can be possible, and to regard others speech as to what could be possible (beyond ones limited knowledge).

A quick example. I assert that there is likely no other civilisation in the universe unmatched to our own that evolved, due to the processes against such a thing (making it uniikely). Then an dude comes down on a ship claiming to be Lord whatever from whatever. Then I have to look at what proof and likely hood of proof, is there they are what they say they are, and consider and explore (if time). Maybe I would not have the ability to concretely find out if it is, true but ha E to consider the evidence I can price as to how likely they are true. Not, act like we see here, where the thought and evidence Braw produces less detail, let alone that Braw is changing what the sensor is seeing, which is Bayer) is like telling somebody who firmly that there are no aliens, that you are alien Lord whatever and they want to go to war on you. Dudes, get a life, go do some work instead or harrassing innocent people.


I'm having some issues here preventing me from getting to assess your footage, but what I can say through my first blurry look (where it doesn't have to be native format only lossless version of the native frame) is that Braw is only a render of what cdng is supposed to be able to record lossless, not superior, and the results people criticise are really the results of the demosaic and processing stream, which is what Braw does. So, trying a better demosaic will likely make the cdng look a lot better. I know you were concentrating on there being more actual detail there, but if people who wrongly assert people as being pixel "preepers" after they were told that we are seeing the difference full screen at optimal viewing distances, and areas of missing image, or and broad areas of whited out image from a fish net) and can't accept that true reality, they are going to get stuck on the bad demosaicing effects left over from Bayer without a optical low pass filter. While in lock down, you might try getting a cdng frame and try processing it through many different demosaicing routines in different software, even photo shop, and see how close you can get to Braw with detail. You could also enable highlight recovery cleanup noise floor with temporal noise removal and optical flow, and compress/shape more latitude stops into it. Then compare the two half frames in one frame to see if you can get something better that brww could do.

Have a good day Jack.
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
PreviousNext

Return to Cinematography

Who is online

Users browsing this forum: carlomacchiavello, David Hutchinson, Kristian Lam, Uli Plank and 29 guests