CaptainHook

I've just read it up there now (I had seen it projected without audio track), it was shot ProRes after all so raw doesn't obviously apply as the OP had inferred, so ends inaccurate and led others like me to misinformation on that one.
Which means no less curious findings anyway. Just not raw related.
 
In any case, there are many tests out there (to prove the shit BMD staff is used to smoke nowadays is updated and HQ indeed):
https://forum.blackmagicdesign.com/viewtopic.php?f=2&t=91466&start=100#p510511
 
I'm sorry @CaptainHook but denial never helped people to grow up... : ) Life and reality above all are beyond our most secret wishes, it doesn't depend on our mantra. Coexists outside :X
 
Our team's flag can end faded, it happens.
 

Ignoring the made up numbers here, in terms of debayer quality the method for cDNG is much older as not anywhere near as good as Blackmagic RAW. I see a lot of people mistake the artefacts of the DNG debayer as sharpness. Its not that the cDNG debayer is necessarily keeping more details, but it IS creating false detail. An interesting comparison was just posted here that's worth looking at :
https://forum.blackmagicdesign.com/viewtopic.php?f=2&p=510320#p510320
Especially this image - what you're seeing on the DNG is not real information the DNG retained, it's CREATING it (falsely) and it can look like sharpness/detail when it's over areas of real detail, but it's a very "hard" and digital feeling look IMHO :

But also pay really close attention to the resolution image at the above link. You can see on the right hand side the artefacts between the red and yellow resolution lines and around the edges of the circle. These artefacts are not real detail or sharpness that you lose with Blackmagic RAW, they are created in error.

Something to consider as well is we get a lot of feedback from customers that 4.6K Blackmagic RAW still has more resolution than some "other" cameras 8K RAW images - likely because of strong optical low pass filtering that without could produce similar artefacts as shown in DNG.
 
The highlights and shadows sliders in cDNG are NOT debayering RAW controls and work on debayered data only. They are the exact same (mathematically) as the ones in the primaries tab and work exactly the same on Blackmagic RAW as the DNG RAW tab ones. I demonstrated this for someone on Facebook last year..

cDNG is better. Period. And such comparison above-posted proves it once again.
P4K brings higher resolution possible to find on doll's hair as for instance, but loses compared with BMCC as far as color concerns; .braw shows less detail.
Just checked on a screen 80" wide along a calibrated 4K projector from a 1080p copy. Fairly visible.
 
Last but not least and overall, BMPCC outcome is the worst sample shown there IMO.

This is not what noise looks like from the ADC. This is not natural and its not the finest details captured, its MORE. The information is made up in the debayer. I know it is.
People also seem to think cDNG has no processing before the file is stored in camera but cDNG is subject to calibration (which attempts to reduce noise and other issues) and other processing before those "RAW" pixels are stored as DNG also. Its okay to like it more too.
Doesn't mean its the same process or the same thing. An OLPF can look similar to Gaussian Blur too.
I get some people prefer cDNG, that's totally fine. But some people are acting like the image was faultless, which it certainly was not. And when there are incorrect statements made as fact, I will try to correct when I'm able to (IP and other concerns restricts my public involvement). The decisions made over a 2 year process developing Blackmagic RAW were not arbitrary or taken lightly either.

Ignoring the made up numbers here, in terms of debayer quality the method for cDNG is much older as not anywhere near as good as Blackmagic RAW. I see a lot of people mistake the artefacts of the DNG debayer as sharpness. Its not that the cDNG debayer is necessarily keeping more details, but it IS creating false detail. An interesting comparison was just posted here that's worth looking at :
https://forum.blackmagicdesign.com/viewtopic.php?f=2&p=510320#p510320
Especially this image - what you're seeing on the DNG is not real information the DNG retained, it's CREATING it (falsely) and it can look like sharpness/detail when it's over areas of real detail, but it's a very "hard" and digital feeling look IMHO :

But also pay really close attention to the resolution image at the above link. You can see on the right hand side the artefacts between the red and yellow resolution lines and around the edges of the circle. These artefacts are not real detail or sharpness that you lose with Blackmagic RAW, they are created in error.

Something to consider as well is we get a lot of feedback from customers that 4.6K Blackmagic RAW still has more resolution than some "other" cameras 8K RAW images - likely because of strong optical low pass filtering that without could produce similar artefacts as shown in DNG.
 
The highlights and shadows sliders in cDNG are NOT debayering RAW controls and work on debayered data only. They are the exact same (mathematically) as the ones in the primaries tab and work exactly the same on Blackmagic RAW as the DNG RAW tab ones. I demonstrated this for someone on Facebook last year..

CinemaDNG offers significantly more details. The comparison pictures BRAW and CinemaDNG also show that BRAW has no finest details. The image information is lost at both 4K and 1080P and cannot be restored. CinemaDNG is different, because it contains the finest details and you can see how good this is when you scale up 1080P to 4K. 
 
BRAW is also not an OLPF, the moiré improvement can be seen minimally in the horizontal area.  You can do similar things with CinemaDNG by applying "Gaussian Blur" filter (H/V Strength: 0.333) to CinemaDNG. 

Ignoring the made up numbers here, in terms of debayer quality the method for cDNG is much older as not anywhere near as good as Blackmagic RAW. I see a lot of people mistake the artefacts of the DNG debayer as sharpness. Its not that the cDNG debayer is necessarily keeping more details, but it IS creating false detail. An interesting comparison was just posted here that's worth looking at :
https://forum.blackmagicdesign.com/viewtopic.php?f=2&p=510320#p510320
Especially this image - what you're seeing on the DNG is not real information the DNG retained, it's CREATING it (falsely) and it can look like sharpness/detail when it's over areas of real detail, but it's a very "hard" and digital feeling look IMHO :

But also pay really close attention to the resolution image at the above link. You can see on the right hand side the artefacts between the red and yellow resolution lines and around the edges of the circle. These artefacts are not real detail or sharpness that you lose with Blackmagic RAW, they are created in error.

Something to consider as well is we get a lot of feedback from customers that 4.6K Blackmagic RAW still has more resolution than some "other" cameras 8K RAW images - likely because of strong optical low pass filtering that without could produce similar artefacts as shown in DNG.
 
The highlights and shadows sliders in cDNG are NOT debayering RAW controls and work on debayered data only. They are the exact same (mathematically) as the ones in the primaries tab and work exactly the same on Blackmagic RAW as the DNG RAW tab ones. I demonstrated this for someone on Facebook last year..

Seems like BMD isn't open for critic's since Braw/ProRes on the Pocket 4K is perfect.
Since you're here @CaptainHook I may ask for fixing the gamut related issues of the perfect V4 BMD color science as well...
 
Right now I have to convert BMD V4 CS to ACES or IPP2 with some smooth gamut mapping applied.
 

Looks like he used the standard sharpening setting though, yikes. cDNG looks way smoother when it is set to 0. Almost looks like GH5s footage lol
But there is also more detail (and noise) since there hasn't been any fast spatial noise reduction applied.

The ProRes examples show some halos around his face and nose which are caused by the internal unsharp masking algorithm.

Yh whatever, done explaining to you as well.
In case people complain about the camera looking videoish or like a DSLM (which do look videoish as well) we just recommend buying Arri or Red cameras instead, even though it's possible to achieve a similar looking image by just bypassing BMD image processing and color science.
 
Cool

Shoot Braw or ProRes and compare it to cdng, Redcode, Arriraw, Canon Raw etc. In case you don't notice it, it isn't my fault and I don't care nor have the time to convince you.

The old ProRes output didn't have any of these new issues. 
To bypass these newly introduced issues by the heavy internal processing you need to shoot cDNG and work in a better colorspace such as ACES or IPP2.

It isn't.
With Braw and ProRes you get strong sharpening and NR (known as the DSLR look), with cDNG there isn't any - when sharpening set to 0 in Resolve.
With Colorspace V4 you get weird lifeless greens and clipped reds which cannot be recovered. With Red IPP2 or ACES (after pulling back) you get perfect reds and greens.

Small test I made using new Rokinon lenses and cDNG inside Red IPP2 last week - download in 4K
https://drive.google.com/open?id=1RRTci4cyRqRsN7vNH1EMEKV1N89AXY9p

I agree. This video looks really DSLR-like. Probably shot using strong internal processing (Braw or ProRes) and bad color grading in the limited V4 colorspace as well.
But using cDNG and using the right colorspace this camera is actually really cinematic.

Keep up the good work Andrew. 
I hadn't realised that about those groups of youtubers that review cameras being selected by the companies but your right. They all tend to be glowing reviews now that I think about it whatever the camera was. I've seen more than a few by iPhondo for example and always very positive reviews with some token comment here and there to appear unbiased. 
As for Trolls and Brawley dont loose sleep over that lot. The launch sample footage was total crap (not the cameras fault). I have my Pocket 4K and absolutely love it. A very capable camera that fits the spot for those of us that want DSLR size but want a pro camera. 
Cheers,
Simon Bailey
Colourberry.com

Resolve 15.3 has a build of SDK 1.3 which does support LUTs in the sidecar, they just need to be manually placed in the sidecar at this time (we would like to add the ability to add LUTs to the sidecar through the Resolve UI in a future update).

If you're brave I've attached a sample sidecar with a 3DLUT (B&W to make it easy to tell its working) which shows the syntax needed. Open the sidecar in a text editor and you can copy in any Resolve formatted 3DLUT (can also open LUTs in a text editor) into the sidecar file data section (inside the quotes) and make sure the "post_3dlut_sidecar_size" number matches the "LUT_3D_SIZE" from the 3DLUT. Just rename the sidecar file to the same as your clip. If you are going to add it to an existing sidecar, put it at the top of the sidecar data inside the curly brace (the current build in Resolve requires it but a future build will not be as strict) and put a comma after the LUT data quote (this is JSON syntax). If you already have the clip in a Resolve timeline, you'll need to remove the clip from the timeline and from the media pool and relink it as Resolve will cache and not pick up the sidecar change. If you just want to test, making a new project and bringing in the clip might be faster.

You'll need to use Blackmagic RAW player 1.3 which is inside the Resolve app bundle so make sure that version is the one being used when viewing clips not using Resolve.
In general though so you are aware the sidecar is just for the controls you find in the RAW panel, the SDK (and applications using it) can't recreate all the complex colour grading operations you can do in Resolve. 

I've also attached a sample Blackmagic RAW frame with an embedded 3DLUT. If you view this in Resolve, go to the RAW panel and change to clip decode and you will see you can enable/disable the 3DLUT, and change from the embedded 3DLUT to the sidecar LUT. This gives you more options for making sure when passing around clips the intention of how you want it to look is preserved non-destructively in any app that supports SDK 1.3+ (when released) which i think is pretty cool!
 
sample.sidecar sample.braw

By that definition you have probably never used raw then, as I haven't come across any camera that supports uncompressed/lossless raw that doesn't alter the sensor data in a process referred to as "calibration" - where every pixel is altered. That can include noise reduction and other processing that's already applied before the raw file is created. Some sensors do this on chip before the ADC! I've also never seen a camera that doesn't do pixel remapping as all sensors suffer hot/dead pixels. Some even offer user initiated pixel remapping to address this as it can occur over time. Which means lots of pixel data is CREATED because the sensor didn't provide useful information. So yeah, no camera from any manufacturer I have seen or used would qualify as offering "raw" to your standard.

People seem to have a romantic and idealised notion about what raw is.

By that definition you have probably never used raw then, as I haven't come across any camera that supports uncompressed/lossless raw that doesn't alter the sensor data in a process referred to as "calibration" - where every pixel is altered. That can include noise reduction and other processing that's already applied before the raw file is created. Some sensors do this on chip before the ADC! I've also never seen a camera that doesn't do pixel remapping as all sensors suffer hot/dead pixels. Some even offer user initiated pixel remapping to address this as it can occur over time. Which means lots of pixel data is CREATED because the sensor didn't provide useful information. So yeah, no camera from any manufacturer I have seen or used would qualify as offering "raw" to your standard.

People seem to have a romantic and idealised notion about what raw is.

By that definition you have probably never used raw then, as I haven't come across any camera that supports uncompressed/lossless raw that doesn't alter the sensor data in a process referred to as "calibration" - where every pixel is altered. That can include noise reduction and other processing that's already applied before the raw file is created. Some sensors do this on chip before the ADC! I've also never seen a camera that doesn't do pixel remapping as all sensors suffer hot/dead pixels. Some even offer user initiated pixel remapping to address this as it can occur over time. Which means lots of pixel data is CREATED because the sensor didn't provide useful information. So yeah, no camera from any manufacturer I have seen or used would qualify as offering "raw" to your standard.

People seem to have a romantic and idealised notion about what raw is.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.

We write the file headers/close the file every 5 secs during record, our cameras since the BMCC have always periodically written the headers/closed the file so at most if you lose power you might lose the last 5 secs of the clip. If you lose power before 5 secs of recording, then it could be an issue.