Denoising diminishes braw and prores detail

Is uncompress or lossless cDNG infringe on anyone's IP?

Don't think so, since uncompressed cdng stills is still available on the bmpcc.

Then I would like lossless cDNG for BMPCC4K. I know some people would go one about file size but we have BMPCC6K with about 2.5x the files size compare to the 4k camera.

I love it when people post their opinions about how something can work.

....

Some of us try to post this information in a "read between the lines" way.

If you push too much data through a system that's designed to work at predetermined ratios you don't "just get a bigger file". You max out the media...

maybe....

or you max out the overhead of the little CPU doing all the work....or it can't be done because the whole system wasn't designed that way or a...

choice...

made for other...

unknown...

engineering reasons.

...


Know that for sure this feature request is heard, and...

I'm sure if it's possible...


JB

...

I haven’t looked at the OPs original test...
but having gone though this with the camera team themselves I’m very well aware of the differences.

...

for daring to test the thinking through debate.

(Or merely disagreeing because they cocked it up Ed)

...


...If you want to argue a point like that, you better come to this public forum with a bit more visual evidence to move the consensus opinion of this forum.

I haven’t looked at the OPs original test...

... Debates like this do help the engineering team to look at it and understand it from a user point of view.

I also know it...

might...

not even be possible to adjust this setting because of the inherent process that happens in camera with this hybrid-RAW step that was created.

JB

The question now is whether Blackmagic Design wants to do this.

...Or if it CAN be done ?

It might...

intrinsic to how they do the first de-mosiac step in camera.

JB

Also remember we're talking about compression here. If we suddenly turn up the sharpness a small change...

could...

make the compression ratios un-sustainable. The camera...

might...

Not be able to encode at that data rate.

Yeah the answer is to make it variable...

Maybe that will come and maybe it won't...

But I think they've settled on a...

choice...

that gives the most number of users the best most RAW like experience without getting into IP trouble.

JB

More noise, more artefacts = more data, much more data. Careful what you wish for.

Exactly. BRAW...

might...

not be able to work if there's not some de-noising.

JB

Please note the datarates and compression ratios used in ProRes and CDNG in the past already on the pocket 4k in question here, which people seemed to have found the same codec technology behind as Braw. Hence it seems likely it...should...do the same datarates. The hardware is supposed to be configurable FPGA, which ... Should ... Configure the codec. I have not heard BM has followed my suggestion to use an off the shelf processing ASIC component instead, which may or may not be more limited.

----

You see just some of all the contradictions I usually see..

But you have improved a lot over the years John, more circumspect, more qualifiers, some more rightness, which is what I was trying to do with everybody.

Then I would like lossless cDNG for BMPCC4K. I know some people would go one about file size but we have BMPCC6K with about 2.5x the files size compare to the 4k camera.

Pretty sure uncompressed DNG isn’t possible on the 6k just on media data rates alone. It’s more than what any media we can use can record.

Another one of the reasons that BMD had to move to BRAW.

JB

Then I would like lossless cDNG for BMPCC4K. I know some people would go one about file size but we have BMPCC6K with about 2.5x the files size compare to the 4k camera.

Pretty sure uncompressed DNG isn’t possible on the 6k just on media data rates alone. It’s more than what any media we can use can record.

Another one of the reasons that BMD had to move to BRAW.

JB

I bet if it was the other way around, probably the same people would be wanting Braw back over CDNG.

These days I still see people doing amazing materials with worse equipment than those discussed here.
I remember back in the days After Effects wasnt a well regarded program in VFX area and we already could do so much amazing things coming out from footage recorded on tapes... right before the DSLR revolution. Then we got here where what matters is the gear in hand to shadow the lack of knowledge or talent. Only a few fellas on the top like JB gets whats matters most. But Kids these days...

I bet if it was the other way around, probably the same people would be wanting Braw back over CDNG.

These days I still see people doing amazing materials with worse equipment than those discussed here.

What's needed is a compulsory double-blind testing, and anyone who fails goes to jail for making claims his own eyes can't support.

In the case above, the complaint as best I can reduce it to an actual case, is that a 1.3 crop of 4K braw footage out of the BMPCC 4K is unacceptably soft on an HD timeline, whereas cDNG, as formerly implemented on that camera, satisfies all desires. (At the same time, the OP insists that he can't accept much increased storage requirement. How that's reconciled, no idea.)

What the OP hasn't bothered to specify is the viewing medium (IMAX? computer monitor? iphone?), why the level of detail he demands, at the expense of artifacts, is essential, and whether anyone can see the difference UNDER NORMAL VIEWING conditions. This means moving pictures, not 3x zooms of stills.

In a word, are we talking about real world requirements and differences objectively and universally observed with normal viewing, or desires at the level of pixel fetish? The argument, "well, I can see it" is not a sound basis for mass-market camera makers, or a reasonable expectation for camera development.

Meanwhile, most shooters want to take the edge off digital, not the other way around.....

Then I would like lossless cDNG for BMPCC4K. I know some people would go one about file size but we have BMPCC6K with about 2.5x the files size compare to the 4k camera.

Pretty sure uncompressed DNG isn’t possible on the 6k just on media data rates alone. It’s more than what any media we can use can record.

Another one of the reasons that BMD had to move to BRAW.

JB

I bet if it was the other way around, probably the same people would be wanting Braw back over CDNG.

These days I still see people doing amazing materials with worse equipment than those discussed here.
I remember back in the days After Effects wasnt a well regarded program in VFX area and we already could do so much amazing things coming out from footage recorded on tapes... right before the DSLR revolution. Then we got here where what matters is the gear in hand to shadow the lack of knowledge or talent. Only a few fellas on the top like JB gets whats matters most. But Kids these days...

My first decent camera was a dvc180 shooting dv. And even then I remember people arguing over the native 16:9 ccd of the xl2 giving "higher resolutions". As consumers we want the best out of the things that we bought, and for my type of work we'd want as much resolution as possible for practical purposes. If the bmpcc4k was suddenly forced to only shoot 3k instead of 4k I'd think a lot people would complain as well. Even though it's "not that big of a difference". We'd still like to get as much out of our gear as possible.

I bet if it was the other way around, probably the same people would be wanting Braw back over CDNG.

These days I still see people doing amazing materials with worse equipment than those discussed here.

What's needed is a compulsory double-blind testing, and anyone who fails goes to jail for making claims his own eyes can't support.

In the case above, the complaint as best I can reduce it to an actual case, is that a 1.3 crop of 4K braw footage out of the BMPCC 4K is unacceptably soft on an HD timeline, whereas cDNG, as formerly implemented on that camera, satisfies all desires. (At the same time, the OP insists that he can't accept much increased storage requirement. How that's reconciled, no idea.)

What the OP hasn't bothered to specify is the viewing medium (IMAX? computer monitor? iphone?), why the level of detail he demands, at the expense of artifacts, is essential, and whether anyone can see the difference UNDER NORMAL VIEWING conditions. This means moving pictures, not 3x zooms of stills.

In a word, are we talking about real world requirements and differences objectively and universally observed with normal viewing, or desires at the level of pixel fetish? The argument, "well, I can see it" is not a sound basis for mass-market camera makers, or a reasonable expectation for camera development.

Meanwhile, most shooters want to take the edge off digital, not the other way around.....

I do understand that many people have different requirements than I do, but I also know a lot of people who'd want extra resolution. That's why if my request was possible, I advocated for options to respect people's choices. And sure, if people were viewing on phones they wouldn't tell the difference between an aging 60d and imax scan at 10k. But when the difference is apparent on my humble 21 inch hd monitor and even my old macbook air , I'm going to be a bit disappointed. From what people had said around here I guess most people shoot narrative works with enough time and space for second angles. So they are pretty much satisfied with the result that braw gives. But there are also going to be some people that are going to need the supposed flexibility of it's “4k” raw. So is it really that hard to get my perspective?

I bet if it was the other way around, probably the same people would be wanting Braw back over CDNG.

These days I still see people doing amazing materials with worse equipment than those discussed here.

What's needed is a compulsory double-blind testing, and anyone who fails goes to jail for making claims his own eyes can't support.

In the case above, the complaint as best I can reduce it to an actual case, is that a 1.3 crop of 4K braw footage out of the BMPCC 4K is unacceptably soft on an HD timeline, whereas cDNG, as formerly implemented on that camera, satisfies all desires. (At the same time, the OP insists that he can't accept much increased storage requirement. How that's reconciled, no idea.)

What the OP hasn't bothered to specify is the viewing medium (IMAX? computer monitor? iphone?), why the level of detail he demands, at the expense of artifacts, is essential, and whether anyone can see the difference UNDER NORMAL VIEWING conditions. This means moving pictures, not 3x zooms of stills.

In a word, are we talking about real world requirements and differences objectively and universally observed with normal viewing, or desires at the level of pixel fetish? The argument, "well, I can see it" is not a sound basis for mass-market camera makers, or a reasonable expectation for camera development.

Meanwhile, most shooters want to take the edge off digital, not the other way around.....

I do understand that many people have different requirements than I do, but I also know a lot of people who'd want extra resolution. That's why if my request was possible, I advocated for options to respect people's choices. And sure, if people were viewing on phones they wouldn't tell the difference between an aging 60d and imax scan at 10k. But when the difference is apparent on my humble 21 inch hd monitor and even my old macbook air , I'm going to be a bit disappointed. From what people had said around here I guess most people shoot narrative works with enough time and space for second angles. So they are pretty much satisfied with the result that braw gives. But there are also going to be some people that are attracted to the apparent flexibility of it's 4k raw. So is it really that hard to get my perspective?

Fair enough but at the moment have you confidently established that CDNG or a new flavour of BRAW is going to get you the ability to substantially crop into the P4k frame and not lose detail? The samples you have provided do not lend itself to this conclusion at all. All they show is a bit more detail in certain areas of the image with certain textures but nothing that looks like the big jump in resolution needed for such a process.

Fair enough but at the moment have you confidently established that CDNG or a new flavour of BRAW is going to get you the ability to substantially crop into the P4k frame and not lose detail? The samples you have provided do not lend itself to this conclusion at all. All they show is a bit more detail in certain areas of the image with certain textures but nothing that looks like the big jump in resolution needed for such a process.

Well I’m really just thinking about disabling the denoiser and the sharpening in the camera. And I know that due to compression braw will still be a bit softer than cdng. But when I could get a decent looking 1080p image out of a 1.5x crop in a low contrast wide shot(quality of something like a gh2). I would be very happy.

Pretty sure Resolve is capable of noise reduction as well, so I'm not sure I see your point.

One is in camera and baked in. One is a different process (sharpening) to the other (noise reduction)
If you sharpen in camera post encoding, then it's the same as doing it in post. Ergo, there's a really simple solution right there. Sharpen in post and it would be the same as BMD adding as sharpness slider in camera. Plus you can dial in as much post sharpening as you want.

He's talking about "denoising" problem, not post sharpening. They are different processes.

The de-noising can't be undone later. Post sharpening in camera could be recorded as metadata and so maybe not destructively added into the BRAW.

The de-noise step as I understand is happening as part of the de-bayer process. Or maybe the emphasis is wrong. The partial de-bayer process intrinsically makes the image a bit softer. The de-noise is part of the recipe that makes BRAW work in these parameters. It's not a user controlled variable because it's a product of the process that was conceptualised with the codec, not an added feature you can dial in or out.

JB

But when I could get a decent looking 1080p image out of a 1.5x crop in a low contrast wide shot(quality of something like a gh2). I would be very happy.

Is that your benchmark ? A GH2 cropped to 1.5X and you're saying you can't get that from the P4K in BRAW ?

JB

But when I could get a decent looking 1080p image out of a 1.5x crop in a low contrast wide shot(quality of something like a gh2). I would be very happy.

Is that your benchmark ? A GH2 cropped to 1.5X and you're saying you can't get that from the P4K in BRAW ?

JB

No I meant the cropped result matching gh2. Right now it’s more of an 5d mark 2

But when I could get a decent looking 1080p image out of a 1.5x crop in a low contrast wide shot(quality of something like a gh2). I would be very happy.

Is that your benchmark ? A GH2 cropped to 1.5X and you're saying you can't get that from the P4K in BRAW ?

JB

No I meant the cropped result matching gh2. Right now it’s more of an 5d mark 2

I think something else is going on here. Something this fundamental is not a codec issue.

Even so, I'm thinking there's something we should look at again from your methodologies. It shouldn't be as bad as what you're seeing if that's the comparison.

JB

I'm just downloading your shots now, but my first thought from your set up....

That sigma zoom lens and then a speed booster on top set's off a few alarms for me.

Have you got a "nicer" lens for the sake of comparing ? A native prime that isn't going through a speedbooster ?

JB

I've looked at the images in your first post.

I couldn't make sense of BRAW crop without the uncropped or a DNG

The 1080 P and 4K dci did show a difference, which I would say is to be expected. For sure, once I zoomed in a bunch to the text on the bottle or the newspaper I could see a difference, but much much less so in wider crops.

I don't see muddiness either. What I did see is a lot less noise in the 3200 shot in BRAW.

I think you need to better illustrate your issue with actual footage in the actual use scenario you're asking about. You can't complain about muddiness in skin tones on a face and then post pictures of a building and a shampoo bottle.

Ideally it's clips and not stills as well. It's hard to extrapolate on other details, especially when we're talking about degrees of sharpness...

You're asking for more, but you're also inferring that it's "soft AND ridden with denoising artifacts"

That's pretty strong language for what I see in your own examples as minor softening that's in-line with BRAW over cDNG. I don't see denoising artifacts either, but I'm assuming you're still talking about softness.

More to the point, I think you'd have a different conversation to the one that's happened here if instead you use a clip from the scenario you're saying that it's "RIDDEN WITH DENOISING ARTIFACTS" is to show us in the work you're seeing it in, where it's on a face and in motion on a clip.

JB

I've looked at the images in your first post.

I couldn't make sense of BRAW crop without the uncropped or a DNG

The 1080 P and 4K dci did show a difference, which I would say is to be expected. For sure, once I zoomed in a bunch to the text on the bottle or the newspaper I could see a difference, but much much less so in wider crops.

I don't see muddiness either. What I did see is a lot less noise in the 3200 shot in BRAW.

I think you need to better illustrate your issue with actual footage in the actual use scenario you're asking about. You can't complain about muddiness in skin tones on a face and then post pictures of a building and a shampoo bottle.

Ideally it's clips and not stills as well. It's hard to extrapolate on other details, especially when we're talking about degrees of sharpness...

You're asking for more, but you're also inferring that it's "soft AND ridden with denoising artifacts"

That's pretty strong language for what I see in your own examples as minor softening that's in-line with BRAW over cDNG. I don't see denoising artifacts either, but I'm assuming you're still talking about softness.

More to the point, I think you'd have a different conversation to the one that's happened here if instead you use a clip from the scenario you're saying that it's "RIDDEN WITH DENOISING ARTIFACTS" is to show us in the work you're seeing it in, where it's on a face and in motion on a clip.

JB

That was my conclusion after viewing the samples yesterday. I think Jack has unrealistic expectations of what CDNG is capable of and has never actually shot with it (unlike many here who are trying too help) and just assumed a '4k' camera would be OK cropped. I still say the fundamental issue is that a 1:1 sampling sensor is not going to look that good once you start to get close to 1:1 pixel sizes. BTW even the GH2 wasn't a 1:1 sampling sensor and downsampled from 16mp which is 2x that of the P4k. While the mp count of the P4k is not big the actual pixel pitch is quite small and is roughly the same as a >40mp 36x24 sensor which does require very careful focus technique to optimise the lens performance esp with manual focus lenses. One of the early issues with the P4k I found was that the focus assist punch in was not that high so critical MF was not that easy.

When I compare the sample images, I agree with user jack0429. The way he described it, that's how I see it when comparing BRAW and cDNG.

@jack0429
Maybe you shouldn't compare cDNG and BRAW, and wishes don't bring (unfortunately!) anything, even if you expressed the wish sensibly, you see how some people here react (yes, not great...). Choose a camera that meets your needs. So buy a cDNG camera if you don't like "only BRAW" camera.

He's talking about "denoising" problem, not post sharpening. They are different processes.

Well I’m really just thinking about disabling the denoiser and the sharpening in the camera.

Hey Jack, the sharpening for Braw is already off, currently it can be baked into Prores (in at least 1 BM camera) not Braw.

Good Luck

When I compare the sample images, I agree with user jack0429. The way he described it, that's how I see it when comparing BRAW and cDNG.

@jack0429
Maybe you shouldn't compare cDNG and BRAW, and wishes don't bring (unfortunately!) anything, even if you expressed the wish sensibly, you see how some people here react (yes, not great...). Choose a camera that meets your needs. So buy a cDNG camera if you don't like "only BRAW" camera.

I agree Lee, except he has the camera already and others bought it having cdng and BM could do better. The cinema camera market is very very slack, it's only because the true competitors don't want to play with low energy, low heat, low cost compact products using ASIC chis. They could do similar better cameras to the pocket for under $500, but hobble cameras instead and go after the above $10k market. They are fortunate to get money enough to keep them afloat. So, there is still space for BM to do better, but of Red lowered their Komodo camera price to $1000-$2000 it would almost be over. But, they like the profit.

Also there has been a lot of excuses in debates like this, on how you are wrong. Like 4:1 CDNG having compression artifacts. I for one am aiming for quality at lossless or near lossless, even if it's 2:1 compression, and I don't imagine Jawere thinking of a low quality solution, so 4:1 does not matter. This is only a quality option, a piece of the puzzle.

When I compare the sample images, I agree with user jack0429. The way he described it, that's how I see it when comparing BRAW and cDNG.

@jack0429
Maybe you shouldn't compare cDNG and BRAW, and wishes don't bring (unfortunately!) anything, even if you expressed the wish sensibly, you see how some people here react (yes, not great...). Choose a camera that meets your needs. So buy a cDNG camera if you don't like "only BRAW" camera.

I agree Lee, except he has the camera already and others bought it having cdng and BM could do better.

Please help us - how can we 'do better'

I agree Lee, except he has the camera already and others bought it having cdng and BM could do better.

You shouldn't bait John. It's not worth it.

JB

Then I would like lossless cDNG for BMPCC4K. I know some people would go one about file size but we have BMPCC6K with about 2.5x the files size compare to the 4k camera.

Pretty sure uncompressed DNG isn’t possible on the 6k just on media data rates alone. It’s more than what any media we can use can record.

Another one of the reasons that BMD had to move to BRAW.

JB

But uncompress cDNG have been around for awhile in 4k. That why I advocate bringing uncompressed lossless cDNG to BMPCC4k. I admit 6k cDNG uncompress may not be possible for current media.

Back in the days when we could produce all the time (2019 seems so long ago) we were shooting an average of one film/tv spot in Braw per week - for months.

It's a proven, brilliant codec on every level that counts - IQ, production, post-production and cost.

Please excuse my lack of pages of unbroken paragraphs and hypothetical thinking.

I've looked at the images in your first post.

I couldn't make sense of BRAW crop without the uncropped or a DNG

The 1080 P and 4K dci did show a difference, which I would say is to be expected. For sure, once I zoomed in a bunch to the text on the bottle or the newspaper I could see a difference, but much much less so in wider crops.

I don't see muddiness either. What I did see is a lot less noise in the 3200 shot in BRAW.

I think you need to better illustrate your issue with actual footage in the actual use scenario you're asking about. You can't complain about muddiness in skin tones on a face and then post pictures of a building and a shampoo bottle.

Ideally it's clips and not stills as well. It's hard to extrapolate on other details, especially when we're talking about degrees of sharpness...

You're asking for more, but you're also inferring that it's "soft AND ridden with denoising artifacts"

That's pretty strong language for what I see in your own examples as minor softening that's in-line with BRAW over cDNG. I don't see denoising artifacts either, but I'm assuming you're still talking about softness.

More to the point, I think you'd have a different conversation to the one that's happened here if instead you use a clip from the scenario you're saying that it's "RIDDEN WITH DENOISING ARTIFACTS" is to show us in the work you're seeing it in, where it's on a face and in motion on a clip.

JB

I’d love to my borrow my work’s 24-70mk 2 for my testing, but unfortunately the lockdown is really significantly restricting what I can shoot in my tiny apartment. The cropped braw sample is from jsfilmz. (https://drive.google.com/file/d/1HjxhVP ... p=drivesdk) that I’ve linked to in my previous posts. The cdng flies though were shot on an sigma fp and out of focus. That’s why I didn’t include it in the cropped sample. That demonstrates what I mean by muddiness on far away textures and faces. You can download his braw sample to see for your self, those were shot at q0,

He's talking about "denoising" problem, not post sharpening. They are different processes.

Well I’m really just thinking about disabling the denoiser and the sharpening in the camera.

Hey Jack, the sharpening for Braw is already off, currently it can be baked into Prores (in at least 1 BM camera) not Braw.

Good Luck

Huh, didn’t really notice that. Then I guess if denoising level could become user selectable that would be nice. Thank you for the info.

Back in the days when we could produce all the time (2019 seems so long ago) we were shooting an average of one film/tv spot in Braw per week - for months.

It's a proven, brilliant codec on every level that counts - IQ, production, post-production and cost.

Please excuse my lack of pages of unbroken paragraphs and hypothetical thinking.

In terms of dynamic range, colorscience, and artifacting braw is a great codec. But for our particular need of cropping it lacks a bit in the resolution department, that’s why I’m advocating for user selectable levels of denoising.

I've looked at the images in your first post.

I couldn't make sense of BRAW crop without the uncropped or a DNG

The 1080 P and 4K dci did show a difference, which I would say is to be expected. For sure, once I zoomed in a bunch to the text on the bottle or the newspaper I could see a difference, but much much less so in wider crops.

I don't see muddiness either. What I did see is a lot less noise in the 3200 shot in BRAW.

I think you need to better illustrate your issue with actual footage in the actual use scenario you're asking about. You can't complain about muddiness in skin tones on a face and then post pictures of a building and a shampoo bottle.

Ideally it's clips and not stills as well. It's hard to extrapolate on other details, especially when we're talking about degrees of sharpness...

You're asking for more, but you're also inferring that it's "soft AND ridden with denoising artifacts"

That's pretty strong language for what I see in your own examples as minor softening that's in-line with BRAW over cDNG. I don't see denoising artifacts either, but I'm assuming you're still talking about softness.

More to the point, I think you'd have a different conversation to the one that's happened here if instead you use a clip from the scenario you're saying that it's "RIDDEN WITH DENOISING ARTIFACTS" is to show us in the work you're seeing it in, where it's on a face and in motion on a clip.

JB

That was my conclusion after viewing the samples yesterday. I think Jack has unrealistic expectations of what CDNG is capable of and has never actually shot with it (unlike many here who are trying too help) and just assumed a '4k' camera would be OK cropped. I still say the fundamental issue is that a 1:1 sampling sensor is not going to look that good once you start to get close to 1:1 pixel sizes. BTW even the GH2 wasn't a 1:1 sampling sensor and downsampled from 16mp which is 2x that of the P4k. While the mp count of the P4k is not big the actual pixel pitch is quite small and is roughly the same as a >40mp 36x24 sensor which does require very careful focus technique to optimise the lens performance esp with manual focus lenses. One of the early issues with the P4k I found was that the focus assist punch in was not that high so critical MF was not that easy.

The gh2 did some pixel binning and line skipping (as expected for a camera in that era), but a lot less then a 5d, that’s why I set it as a little crop benchmark. I would say that I really got the samples perfectly in focus By connecting it to a monitor and looking at the focus precisely, there are some halationing in the trees on the cdng so maybe that’s from diffraction caused at f5.6? I mean the actual lens is stepped down to f8 at this point. So maybe that was the problem?

Even so, I'm thinking there's something we should look at again from your methodologies. It shouldn't be as bad as what you're seeing if that's the comparison.

JB

Braw doesn’t look half bad when there are bright contrast and well defined edges. That’s why the samples are shot in overcast or at night, trying to show the denoising at work. Of course from what I’ve read it might not be possible to adjust the denoising for now, I just hope that bm is aware of this issue and can improve braw with this in mind.

In terms of dynamic range, colorscience, and artifacting braw is a great codec. But for our particular need of cropping it lacks a bit in the resolution department, that’s why I’m advocating for user selectable levels of denoising.

Based on what you've said here, you haven't correctly assessed the resolution of the BMPCC 4K. The GH2 has nowhere near the detail of the BMPCC 4K Q5 UHD braw on an HD timeline. And Q5 is noticeable softer than Q0, if you're looking for that sort of thing.

And yet you insist the GH2 has more detail and that the BMPCC 4K can't survive a 1.5 zoom. Whether it can or can't is up to you, but it survives a lot better than the GH2.

All this suggests that either your BMPCC 4K is out of focus or is otherwise compromised by lens attachments or post-production errors.

But uncompress cDNG have been around for awhile in 4k. That why I advocate bringing uncompressed lossless cDNG to BMPCC4k. I admit 6k cDNG uncompress may not be possible for current media.

They're not going to introduce a feature like that on one camera and not others.

JB

In terms of dynamic range, colorscience, and artifacting braw is a great codec. But for our particular need of cropping it lacks a bit in the resolution department, that’s why I’m advocating for user selectable levels of denoising.

Based on what you've said here, you haven't correctly assessed the resolution of the BMPCC 4K. The GH2 has nowhere near the detail of the BMPCC 4K Q5 UHD braw on an HD timeline. And Q5 is noticeable softer than Q0, if you're looking for that sort of thing.

And yet you insist the GH2 has more detail and that the BMPCC 4K can't survive a 1.5 zoom. Whether it can or can't is up to you, but it survives a lot better than the GH2.

All this suggests that either your BMPCC 4K is out of focus or is otherwise compromised by lens attachments or post-production errors.

If you cropped a bmpcc4k into 1.5x, on a wide shot with faces that are far away, with low contrast textures, the gh2 uncropped in 1080p is gonna have more clarity. Braw does decent with high contrast edges, but the noise reduction hinders it when it comes to low contrast textures. If it was just simply a resolution chart I would be happy with braw's performance. But I often shoot other kinds of low contrast textures as well, that sometimes seem smeared with the noise reduction on braw. This I would say is demonstrated pretty well if you download the jsfilmz sample and crop around with it, the faces just seems a bit too muddy even on a hd screen.

Braw doesn’t look half bad when there are bright contrast and well defined edges. That’s why the samples are shot in overcast or at night, trying to show the denoising at work. .

I think this is the reason you're getting a lot of people so vehement about BRAW.

"Braw doesn’t look half bad"

I think a lot of people think BRAW looks AMAZING. You're saying it doesn't look half bad. Like maybe its as good as ProRes ? A lot of people would disagree with this language.

and as I mentioned

"ridden with denoising artifacts"

Ridden is also a loaded word.

"a combining form meaning “obsessed with,” “overwhelmed by” (torment-ridden) or “full of,” “burdened with” (debt-ridden)"

That's just very strong language to use and it doesn't help the credibility of such a strong position when you don't provide rock solid above reproach examples of how you came to that statement.

By all means make the claim, but you're not going to get a lot of sympathy from those that don't agree with the text version of that statement. You have to show it visually and in a hell of a lot more compelling way than you have so far.

I don't think a single person in this thread is actually arguing against what you're asking for. I think everyone would LOVE to have an option to vary the amount of de-noise in BRAW.

That's a feature everyone in this thread would want and why wouldn't you.

However, your language around your problems with BRAW make the issue sound a hell of a lot worse than most others have found in their own experience. Not opinion about theory, but in actual usage.

You're pitching a feature everyone wants, but doing it in a way that most everyone disagrees to the severity of.

JB

Braw doesn’t look half bad when there are bright contrast and well defined edges. That’s why the samples are shot in overcast or at night, trying to show the denoising at work. .

I think this is the reason you're getting a lot of people so vehement about BRAW.

"Braw doesn’t look half bad"

I think a lot of people think BRAW looks AMAZING. You're saying it doesn't look half bad. Like maybe its as good as ProRes ? A lot of people would disagree with this language.

and as I mentioned

"ridden with denoising artifacts"

Ridden is also a loaded word.

"a combining form meaning “obsessed with,” “overwhelmed by” (torment-ridden) or “full of,” “burdened with” (debt-ridden)"

That's just very strong language to use and it doesn't help the credibility of such a strong position when you don't provide rock solid above reproach examples of how you came to that statement.

By all means make the claim, but you're not going to get a lot of sympathy from those that don't agree with the text version of that statement. You have to show it visually and in a hell of a lot more compelling way than you have so far.

I don't think a single person in this thread is actually arguing against what you're asking for. I think everyone would LOVE to have an option to vary the amount of de-noise in BRAW.

That's a feature everyone in this thread would want and why wouldn't you.

However, your language around your problems with BRAW make the issue sound a hell of a lot worse than most others have found in their own experience. Not opinion about theory, but in actual usage.

You're pitching a feature everyone wants, but doing it in a way that most everyone disagrees to the severity of.

JB

Ahhhh I see what you mean now, but what do you think might be more appropriate for the title of this post? Braw's denoising is excessively smoothing out low contrast textures or something like that? English is my second language so I couldn't really think of more compact title that conveyed what I thought. Sorry for the misleading title.

If you cropped a bmpcc4k into 1.5x, on a wide shot with faces that are far away, with low contrast textures, the gh2 uncropped in 1080p is gonna have more clarity. Braw does decent with high contrast edges, but the noise reduction hinders it when it comes to low contrast textures. If it was just simply a resolution chart I would be happy with braw's performance. But I often shoot other kinds of low contrast textures as well, that sometimes seem smeared with the noise reduction on braw. This I would say is demonstrated pretty well if you download the jsfilmz sample and crop around with it, the faces just seems a bit too muddy even on a hd screen.

Where are your GH2 v. BMPCC 4K tests? These generalities of yours are meaningless; the problem isn't English as a second language, it's that there's no basis for these suppositions of yours. Unless you're privy to braw engineering, you don't know what's actually going on with its design.

jsfilmz is a beginning shooter, with zero credentials. If he's your source, this is an even greater waste of time than first thought.

If you cropped a bmpcc4k into 1.5x, on a wide shot with faces that are far away, with low contrast textures, the gh2 uncropped in 1080p is gonna have more clarity. Braw does decent with high contrast edges, but the noise reduction hinders it when it comes to low contrast textures. If it was just simply a resolution chart I would be happy with braw's performance. But I often shoot other kinds of low contrast textures as well, that sometimes seem smeared with the noise reduction on braw. This I would say is demonstrated pretty well if you download the jsfilmz sample and crop around with it, the faces just seems a bit too muddy even on a hd screen.

Where are your GH2 v. BMPCC 4K tests? These generalities of yours are meaningless; the problem isn't English as a second language, it's that there's no basis for these suppositions of yours. Unless you're privy to braw engineering, you don't know what's actually going on with its design.

jsfilmz is a beginning shooter, with zero credentials. If he's your source, this is an even greater waste of time than first thought.

I'm in a lockdown so I don't have an access to a gh2, what I'm saying is the general clarity should be around that level. And it doesn't matter if he is a "beginning" level shooter, this particular braw footage is tack sharp in focus using a sigma 18-35 and it shows what I'm talking about. So that should be what matters.

The GH2 cannot equal in resolution, under any conditions, UHD braw on the BMPCC 4K. I know, because I own both cameras. Your imaginary low contrast scene, which you won't produce, proves otherwise only in your imagination.

But I'm done. Best of luck.

The GH2 cannot equal in resolution, under any conditions, UHD braw on the BMPCC 4K. I know, because I own both cameras. Your imaginary low contrast scene, which you won't produce, proves otherwise only in your imagination.

But I'm done. Best of luck.

I said it was in a 1.5x crop? Jesus Christ, and I'd love to show other examples that I've shot previously, but those were a from production and it's not a good idea to post people's faces onto the internet. So I'm sorry for the lack of examples, but the samples from jsfilmz is demonstrating what I'm talking about... Man do I really need to repeat myself so many times?

The GH2 cannot equal in resolution, under any conditions, UHD braw on the BMPCC 4K. I know, because I own both cameras. Your imaginary low contrast scene, which you won't produce, proves otherwise only in your imagination.

But I'm done. Best of luck.

Jack. From what I can see, the GH2 is not an ideal comparison. Of course we an image shoot at lower resolution then the sensor or camera supports, might benefit from downscaling and more real image information per pixel because of higher data per pixel in those modes sometimes on codecs that are more efficient in compressing image information. So, a mod on a GH2 can present a more quality image than standard mode as well (so what mode was used here?).

Viewed at a proper testing distance with eyesight that 80%+ of the audience has or has less of (a quality metric to deliver a quality experience to the vast amount of the audience) the problems should be more apparent. If people have less than ideal vision, they should move an appropriate distance closer to the display to see what others are seeing. This is mature production. Of course, if one is just going go to low quality broadcast or streaming, then Braw is not such an issue, but as I've pointed out over the years, a source still effects the quality of image going into compression, potentially affecting outcomes. What you describe is very consumer codec like, but also, because of strong olpf or other anti-alaising technique, cropping in is going bring their softness out a lot more, where as downsampling to 1080p on another camera, is going to help hide some of that softness.

But what you are also experiencing is people trying to bog you down in irrelevancies, because they can't get their way over you. I've watched this behaviour from several members over the years. You answer them, and they suddenly keep shifting the goal posts, trawlling you out. There isn't really an end to the conservation until they win on some perceived baseless ground. Very simple. I used to just let them wear themselves out over a few weeks trying to play me and the system, which by then they are speaking gibberish in terms of logic and likely realising I have started and am still talking about something different then what they bombed in on. But at the handicapped rate I type, spending 6 hours a day to answer one or two but cases, was really a bit much (think and take time to write guys and you might write a lot better. I could be imagining it here, but a certain person is writing a lot better because of lock down) costing me tens of thousands of dollars in damages every few weeks. No apology when eventually they realise and start admiting to themselves how wrong they are. So, I give up wasting hundreds of hours a year reasoning with the nasty cats and just shew them off if they don't want to red, listen and understand, and are playing games with themselves in dishonest self denial, and point things out to them. But you can't grind it out of them.

What you are facing is blatent immaturity, but hey, nearly every body has some. But some people just make a bad show of it. You shew them off eventually, except they try to dominate many threads, so you are stuck with them being intellectually impolite and rude with their antics. Very costly people, and must be a nightmare to get them to do premium quality (I suspect, but have seen examples). Of all the smartest people I know, they are not so prima donna except about the truth, and rarely being an arse. But these people are not wrong according to themselves.

Amusing photos I found while trying to remember how to spell 'prima donna':

Before:
https://www.shutterstock.com/image-phot ... -123123598

https://www.shutterstock.com/image-phot ... -123123589

Perhaps it’s best to pause this thread until Jake (or someone else) can provide better samples. I don’t think this needs to reinvent the wheel by proving BRAW is slightly softer and has more NR than CDNG as no one is arguing against this but he needs to prove that this difference becomes a significant image quality issue when the image is cropped.

Do a simple test is not hard, do a good shooting and press photo button, and do simple “check the difference”.

http://www.macchiavello.com/wp/en/shoot ... a-cameras/

Here you can find an article that I wrote last years like how to shoot photo with cinema camera, where I shoot uncompressed cdng against extracted frame from Q0. Check the picture and tell me if you find difference between extracted frame and uncompressed cdng. I usually use a photo camera, but my photo camera had an accident night before then I use that I have.

If you want to do a test you must do at best of light and picture when you can see the sharpness, if you talk about sharpness.

And remember that you never shoot at cdng at that quality, be cause this is uncompressed and shooting was (my pocket are from original batch) 3:1 or more compressed.
Cdng was good in 2012 with some sensors but today I find that codec like braw is the best combination of quality - go - cpu/gpu stress.

Me I had gh2, I was in the group of hacking mod of gh2, never had the sharpness of pocket, also if shoot at 12:1.. last week when I tried to cut some old shooting with newer seems that I leave my eyes glasses on table when I saw gh2 shoot.


Inviato dal mio iPhone utilizzando Tapatalk

Do a simple test is not hard, do a good shooting and press photo button, and do simple “check the difference”.

http://www.macchiavello.com/wp/en/shoot ... a-cameras/

Here you can find an article that I wrote last years like how to shoot photo with cinema camera, where I shoot uncompressed cdng against extracted frame from Q0. Check the picture and tell me if you find difference between extracted frame and uncompressed cdng. I usually use a photo camera, but my photo camera had an accident night before then I use that I have.

If you want to do a test you must do at best of light and picture when you can see the sharpness, if you talk about sharpness.

And remember that you never shoot at cdng at that quality, be cause this is uncompressed and shooting was (my pocket are from original batch) 3:1 or more compressed.
Cdng was good in 2012 with some sensors but today I find that codec like braw is the best combination of quality - go - cpu/gpu stress.

Me I had gh2, I was in the group of hacking mod of gh2, never had the sharpness of pocket, also if shoot at 12:1.. last week when I tried to cut some old shooting with newer seems that I leave my eyes glasses on table when I saw gh2 shoot.


Inviato dal mio iPhone utilizzando Tapatalk

I have left some samples in my original post that demonstrated what I was talking about. And when I mentioned the gh2, I was talking about hoping that braw would match the gh2 in 1080p while being cropped to 1.5x. Braw is fine if you don't try crop. But for people like us who needs to crop a lot the softness of braw gets greatly exaggerated. I don't need cdng, I want braw with user selectable level of denoising. Also the download on your website seems to be failing after a few megabytes.

Do a simple test is not hard, do a good shooting and press photo button, and do simple “check the difference”.

http://www.macchiavello.com/wp/en/shoot ... a-cameras/

Here you can find an article that I wrote last years like how to shoot photo with cinema camera, where I shoot uncompressed cdng against extracted frame from Q0. Check the picture and tell me if you find difference between extracted frame and uncompressed cdng. I usually use a photo camera, but my photo camera had an accident night before then I use that I have.

If you want to do a test you must do at best of light and picture when you can see the sharpness, if you talk about sharpness.

And remember that you never shoot at cdng at that quality, be cause this is uncompressed and shooting was (my pocket are from original batch) 3:1 or more compressed.
Cdng was good in 2012 with some sensors but today I find that codec like braw is the best combination of quality - go - cpu/gpu stress.

Me I had gh2, I was in the group of hacking mod of gh2, never had the sharpness of pocket, also if shoot at 12:1.. last week when I tried to cut some old shooting with newer seems that I leave my eyes glasses on table when I saw gh2 shoot.


Inviato dal mio iPhone utilizzando Tapatalk

I have left some samples in my original post that demonstrated what I was talking about. And when I mentioned the gh2, I was talking about hoping that braw would match the gh2 in 1080p while being cropped to 1.5x. Braw is fine if you don't try crop. But for people like us who needs to crop a lot the softness of braw gets greatly exaggerated. I don't need cdng, I want braw with user selectable level of denoising. Also the download on your website seems to be failing after a few megabytes.

1080p on the GH2 is subsampled down from 16mp. The P4k is less than 6mp in a 1.5 crop so such a comparison cannot be valid no matter what your hoped for and changing NR is not going to significantly make 6mp look like 16mp.

Do a simple test is not hard, do a good shooting and press photo button, and do simple “check the difference”.

http://www.macchiavello.com/wp/en/shoot ... a-cameras/

Here you can find an article that I wrote last years like how to shoot photo with cinema camera, where I shoot uncompressed cdng against extracted frame from Q0. Check the picture and tell me if you find difference between extracted frame and uncompressed cdng. I usually use a photo camera, but my photo camera had an accident night before then I use that I have.

If you want to do a test you must do at best of light and picture when you can see the sharpness, if you talk about sharpness.

And remember that you never shoot at cdng at that quality, be cause this is uncompressed and shooting was (my pocket are from original batch) 3:1 or more compressed.
Cdng was good in 2012 with some sensors but today I find that codec like braw is the best combination of quality - go - cpu/gpu stress.

Me I had gh2, I was in the group of hacking mod of gh2, never had the sharpness of pocket, also if shoot at 12:1.. last week when I tried to cut some old shooting with newer seems that I leave my eyes glasses on table when I saw gh2 shoot.


Inviato dal mio iPhone utilizzando Tapatalk

I have left some samples in my original post that demonstrated what I was talking about. And when I mentioned the gh2, I was talking about hoping that braw would match the gh2 in 1080p while being cropped to 1.5x. Braw is fine if you don't try crop. But for people like us who needs to crop a lot the softness of braw gets greatly exaggerated. I don't need cdng, I want braw with user selectable level of denoising. Also the download on your website seems to be failing after a few megabytes.

1080p on the GH2 is subsampled down from 16mp. The P4k is less than 6mp in a 1.5 crop so such a comparison cannot be valid no matter what your hoped for and changing NR is not going to significantly make 6mp look like 16mp.

The gh2 is pixel binned in 1080p mode, in fact even the gh4 pixel binned in 1080p. Thus it did not utilize the 16mp of the entire sensor.

Pixel binning from 16mp vs subsampling from 6mp - not exactly a level playing field by which to compare 2 different cameras though is it?

So I did some test shots with my PCC4k.

Setup:

• Camera: PCC4k, received in Oct. 2018, Firmware 6.8
  @ 180° shutter / 24 fps / 4K DCI (4096x2160) / WB ~4.500 K / ISO 200
• Metabones Speedbooster Cine Ultra 0.71x
• Lens: Sigma 18-35 mm f/1.8
  @ f/4.0 (which equals to f/5.6 without speedbooster)
• VND Filter: B+W XS-Pro MRC nano 72mm
  @ 5 stops of ND
• Preview monitor: Atomos Shinobi SDI/HDMI
• Tripod: Sirui BCT-3203 / Head: Sirui BCH-30 / with VCT-14 plate and shoulder mount.

2020-05-06_12-55-53_IMG_1317-50%.JPG (719.67 KiB) Viewed 17481 times

2020-05-06_12-56-07_IMG_1318-50%.JPG (725.52 KiB) Viewed 17481 times

2020-05-06_12-56-18_IMG_1319-50%.JPG (724.11 KiB) Viewed 17481 times

I captured the footage in uncompressed CDNG via stills button, BRAW Q0, BRAW 3:1, and ProRes HQ 422.

Unfortunately pressing the stills button always put enough pressure on the setup so the framing moved a bit vertically compared to the rest. Very annoying - I should have mounted a lens support to prevent this.

Edited in Reslove 16.2.1. Only applied my LBK Neutral HiDesat v10 LUT, available here

NOTE: The 4k DCI JPGs are a bit softer than the PNGs I converted from, because of the forum upload limit of 1 MB I had to compress the JPGs stronger.
You get a better picture with the cropped versions because the JPG compression is far less.

CDNG was developed in Resolve with sharpness set to 0.

CDNG uncompressed

4k DCI_1.1.1.jpg (872.15 KiB) Viewed 17468 times