ProRes 4444 12bit Recording

Will there be an software upgrade to get DNxHD or ProRes 4:4:4 in the future?

mistype a bit...this is regarding the 4k camera.

viewtopic.php?f=2&t=11311
viewtopic.php?f=2&t=11312
viewtopic.php?f=2&t=11313

Did you think uttering it thrice would make it come true?

(referring to Lewis Carroll's "Hunting of the Snark")

Incorporating ProRes 4444 would require BMD to licence the format from apple - I don't think they'd spend additional money to upgrade already sold cameras.

http://forum.blackmagicdesign.com/viewtopic.php?f=2&t=11311
viewtopic.php?f=2&t=11312
viewtopic.php?f=2&t=11313

Did you think uttering it thrice would make it come true?

(referring to Lewis Carroll's "Hunting of the Snark")

Incorporating ProRes 4444 would require BMD to licence the format from apple - I don't think they'd spend additional money to upgrade already sold cameras.

It also would be pointless. The sensor itself doesn't resolve that. You would want to over sample first. You could in theory do 1080p 4:4:4 ProRes off the 4k

I'd like "ProRes 4444" because it's 12-bit.

"ProRes 422 HQ" is 10-bit.

-

Hopefully BMD will surprise us with compressed RAW and will negate the need for ProRes4444.

Jules

http://forum.blackmagicdesign.com/viewtopic.php?f=2&t=11311
viewtopic.php?f=2&t=11312
viewtopic.php?f=2&t=11313

Did you think uttering it thrice would make it come true?

(referring to Lewis Carroll's "Hunting of the Snark")

Incorporating ProRes 4444 would require BMD to licence the format from apple - I don't think they'd spend additional money to upgrade already sold cameras.

It also would be pointless. The sensor itself doesn't resolve that. You would want to over sample first. You could in theory do 1080p 4:4:4 ProRes off the 4k

u guys work for black black magic ? and i forgot to mention that the question was for the up upcoming 4k production camera..

I'd like "ProRes 4444" because it's 12-bit.

"ProRes 422 HQ" is 10-bit.

-

agree with u on the 12bit and that why i ask ... planning to get the 4k camera n hopefully they upgrade the software for it.

Seem a lot of people here work for BMD and answer my question. But i want some concrete answer from a real BDM employee..not the forum employee sitting behind their comp.

Seem a lot of people here work for BMD and answer my question. But i want some concrete answer from a real BDM employee..not the forum employee sitting behind their comp.

Can you please update your user name to include your full name or add it to your signature. In that way conversations tend to remain civil. It is considered a forum rule around here.

Jules

http://forum.blackmagicdesign.com/viewtopic.php?f=2&t=11311
viewtopic.php?f=2&t=11312
viewtopic.php?f=2&t=11313

Did you think uttering it thrice would make it come true?

(referring to Lewis Carroll's "Hunting of the Snark")

Incorporating ProRes 4444 would require BMD to licence the format from apple - I don't think they'd spend additional money to upgrade already sold cameras.

It also would be pointless. The sensor itself doesn't resolve that. You would want to over sample first. You could in theory do 1080p 4:4:4 ProRes off the 4k

u guys work for black black magic ? and i forgot to mention that the question was for the up upcoming 4k production camera..

No I do not work for Black Magic (not yet at least), the info about sensor is a public knowledge.

I'd like "ProRes 4444" because it's 12-bit.

"ProRes 422 HQ" is 10-bit.

-

Ahh. I see. My bad. So the desire is the quantinization improvement in the dynamic range sampling.

Uncompressed raw is superb. Why compress it if you do not have to ?

I would not want that on my 2.5 k.

I see that on 4k there is media limitations etc and you have to but not on 2.5 k.

Maybe if we could have 444 on sdi would be cool for location recording ?

Uncompressed raw is superb. Why compress it if you do not have to ?

I would not want that on my 2.5 k.

I see that on 4k there is media limitations etc and you have to but not on 2.5 k.

Maybe if we could have 444 on sdi would be cool for location recording ?

Well the problem is that by definition (I know I know, yet again, I love that expression) Bayer pattern raw cannot be 4:4:4. it's raw. RAW has no color. You can debayer raw image into a smaller image size. For it to be 4:4:4 natively each of the pixels should have full color information. Ways to achieve that:

prism split into 3 color components and record onto 3 monochrome chips, mix the data from three chips;

Use a sensor like the one in some sigma cameras, where each pixel records all of the colors, due to different wavelength absorption at different depth of silicon;

Oversample by factor of 4 (area) at recording, then downsample by factor of 2 (dimension) in post.
The only way 4:4:4 would be possible on bm4k if it is for a 1080p recorded ProRes files or for 1080p sdi-HD out, granted it is supported by the sdi-HD protocol and camera's firmware.

this the part from where some of the alexas price comes from ?

Many sensors and optics i mean.

Uncompressed raw is superb. Why compress it if you do not have to ?

I would not want that on my 2.5 k.

You sound like you're assuming that all compression is lossy. If BMD implements lossless compression, then what would you be losing?

Lossless compression is what BMD promised for the BMPCC as well as for the BMCC, so there's no rational reason to oppose compressed raw here.

not at all, this just some purist vision in me talking

In light of the fact that it works as it is (provided that you get fast enough media, of course ) I'd resist lossy compressed raw, but not lossless compression.

i think i can go with both as i have fast io and nice resolve box.

This question is getting out of hand seem a lot of people are posting to my simple question which is a yes or no answer for the update on the software from BDM. Way to much teacher in here. Just an FYI my background is like Grant Petty I'm a post guy (17+ years in the bis start out since Apex and Sony1" was hot). Reason I ask this question is we got a project coming up that will deal with a lot of VFX and green screen work and if we the ability to record to 444 chroma sub @12bit to the 4k cine cam... it would be totally awesome WITH OUT THE TRANSCODE using CinemaDNG ! We have the drive and space for the project. Though toss this question to BDM... if not than we can just settle on ProRess 422 HQ or DNxHD 175xMx and YES CineDNG..... We could use or rent or buy the C500 or the Alexa but the idea of paying 60-70K (camera+external recorder) for something that the 4K cinema camera can pull off for $4k we might as well buy a few of these camera and get some nice lens. The 4k cinema has a 6G SDI output i think it can handle the data path... But yea i guess thanks for all the comment. Now would really like to hear what Grant would say...yes no maybe...that is all..Cheer

p.s and we do own a scarlet but we trying to move away from R3D's.

This question is getting out of hand seem a lot of people are posting to my simple question which is a yes or no answer for the update on the software from BDM. Way to much teacher in here. Just an FYI my background is like Grant Petty I'm a post guy (17+ years in the bis start out since Apex and Sony1" was hot). Reason I ask this question is we got a project coming up that will deal with a lot of VFX and green screen work and if we the ability to record to 444 chroma sub @12bit to the 4k cine cam... it would be totally awesome WITH OUT THE TRANSCODE using CinemaDNG ! We have the drive and space for the project. Though toss this question to BDM... if not than we can just settle on ProRess 422 HQ or DNxHD 175xMx and YES CineDNG..... We could use or rent or buy the C500 or the Alexa but the idea of paying 60-70K (camera+external recorder) for something that the 4K cinema camera can pull off for $4k we might as well buy a few of these camera and get some nice lens. The 4k cinema has a 6G SDI output i think it can handle the data path... But yea i guess thanks for all the comment. Now would really like to hear what Grant would say...yes no maybe...that is all..Cheer

p.s and we do own a scarlet but we trying to move away from R3D's.

You cannot have 4:4:4 in a native resolution from any Bayer pattern sensor, and neither bmcc nor bm4k an exception. If someone tells you otherwise - they are lying. 4:4:4 from a Bayer pattern sensor is only possible with over sampling.
While Black Magic may chose putting a 4:2:2 12-bit image into a ProRes 4444 12 bit file, it will not have any more information than a ProRes 4:2:2 file as far as the color resolution is concerned (though it will have 12 bit sampling rate of course). It'll still be 4:2:2. They of course would have to make huge files (that extra 2:2 for color channels take data bandwidth), because while they do not have any extra information they would still have to encode it as such.

No, c500 cannot give you 4:4:4 at a native resolution, even if it is encoded as such. What it can as per the manufacturer's specs is to give you either 2k or 1080p at 4:4:4 by, wait for it, downsampling the 4k sensor data.
Now, even if they make a ProRes 4444 doesn't mean that the data that written is actually 4:4:4 with an alpha channel. Just like because canon 5dmarkII making 1080p files doesn't mean that there is actually 1080 lines of vertical resolution in there (closer to 600, tops)

Indeed you could have a yes or no answer, but instead you can have an explanation as of why, so next time a sales person tells you that their CMOS Bayer pattern single sensor camera delivers 4:4:4 color at a native resolution, you would be wiser and take your money elsewhere.

This question is getting out of hand seem a lot of people are posting to my simple question which is a yes or no answer for the update on the software from BDM. Way to much teacher in here. Just an FYI my background is like Grant Petty I'm a post guy (17+ years in the bis start out since Apex and Sony1" was hot). Reason I ask this question is we got a project coming up that will deal with a lot of VFX and green screen work and if we the ability to record to 444 chroma sub @12bit to the 4k cine cam... it would be totally awesome WITH OUT THE TRANSCODE using CinemaDNG ! We have the drive and space for the project. Though toss this question to BDM... if not than we can just settle on ProRess 422 HQ or DNxHD 175xMx and YES CineDNG..... We could use or rent or buy the C500 or the Alexa but the idea of paying 60-70K (camera+external recorder) for something that the 4K cinema camera can pull off for $4k we might as well buy a few of these camera and get some nice lens. The 4k cinema has a 6G SDI output i think it can handle the data path... But yea i guess thanks for all the comment. Now would really like to hear what Grant would say...yes no maybe...that is all..Cheer

p.s and we do own a scarlet but we trying to move away from R3D's.

You cannot have 4:4:4 in a native resolution from any Bayer pattern sensor, and neither bmcc nor bm4k an exception? If someone tells you otherwise - they are lying. 4:4:4 from a Bayer pattern sensor is only possible with over sampling. While Black Magic may chose putting a 4:2:2 12-bit image into a ProRes 4444 12 bit file, it will not have any more information than a ProRes 4:2:2 file as far as the color resolution is concerned (though it will have 12 bit sampling rate of course). It'll still be 4:2:2. They of course would have to make huge files (that extra 2:2 for color channels take data bandwidth), because while they do not have any extra information they would still have to encode it as such.

Indeed you could have a yes or no answer, but instead you can have an explanation as of why, so next time a sales person tells you that their CMOS Bayer pattern single sensor camera delivers 4:4:4 color at a native resolution, you would be wiser and take your money elsewhere.

uhh u not reading my question..."yes or no or maybe from BDM" that is all i need to know thanks.

cheer

uhh u not reading my question..."yes or no or maybe from BDM" that is all i need to know thanks.

cheer

This forum is open to all participants, who wish to express their opinion, and, by the way, this is a real name forum.
If you have any personal, secret questions to Black Magic Design and do not wish to have other members of the community to chime in, please use the official contact forms, available on the support pages of the BMD website.

BM very seldom comment on here. So I doubt you'll get an answer to your question. Being rude and obnoxious will win you no friends on here.

The fact is 12bit 444 at 1080p is MAYBE possible. 12bit 444 at 4K is NOT possible because the data is not there after the de-Bayer process.

To get 12bit 444 at 4K you would need a 3CCD camera or a much higher resolution sensor that can be down sampled to produce 12bit 444.

You seem to mistake this user forum for the BMD customer service. If you want an answer from BMD themselves, call them, or send an email, or whatever. But please don't piss the experienced users in this forum who only want to help you.

From the terms you agreed to when you registered:

Registration
Our aim is to have a friendly and respectful forum for Blackmagic Design customers to assist each other with ideas, technical advice and discussions on creativity. To ensure the forums stay friendly and respectful, we ask you to agree to the following terms.

...

You agree to use your real name including your location and job title if applicable. We believe people who post under their own name will be more respectful of others when posting as opposed to posting anonymously.

You agree that the "Blackmagic Forum" is for all levels of creative users and professionals who work in film and video production. Further, you agree to welcome discussions from first time videographers as well as the highest level film professionals.

...

If i was bartman, i'd post a picture of someone running head first into a door labeled "pull". As i am not, just imagine i did.

Btw: BMD offiicials are easily recogniced by their different colored usernames. For a legend look at the "who is online" list at the bottom of the Board Index page.

Well the problem is that by definition (I know I know, yet again, I love that expression) Bayer pattern raw cannot be 4:4:4. it's raw. RAW has no color. You can debayer raw image into a smaller image size. For it to be 4:4:4 natively each of the pixels should have full color information. Ways to achieve that:

prism split into 3 color components and record onto 3 monochrome chips, mix the data from three chips;

Use a sensor like the one in some sigma cameras, where each pixel records all of the colors, due to different wavelength absorption at different depth of silicon;

Oversample by factor of 4 (area) at recording, then downsample by factor of 2 (dimension) in post.
The only way 4:4:4 would be possible on bm4k if it is for a 1080p recorded ProRes files or for 1080p sdi-HD out, granted it is supported by the sdi-HD protocol and camera's firmware.

What Dmitry so expertly explained here comes down to a simple truth (please correct me if i got it wrong):

A 4:2:2 subsampled matrix is enough to store the full amount of information contained in raw bayer pattern data, using a 4:4:4 matrix would just waste storage space and/or bandwidth. 4:4:4 would only be usefull to store 4k images scaled down to 1080.

12 instead of 10 bit would make a difference though, in any resolution.

You cannot have 4:4:4 in a native resolution from any Bayer pattern sensor, and neither bmcc nor bm4k an exception. If someone tells you otherwise - they are lying. 4:4:4 from a Bayer pattern sensor is only possible with over sampling.
While Black Magic may chose putting a 4:2:2 12-bit image into a ProRes 4444 12 bit file, it will not have any more information than a ProRes 4:2:2 file as far as the color resolution is concerned (though it will have 12 bit sampling rate of course). It'll still be 4:2:2. They of course would have to make huge files (that extra 2:2 for color channels take data bandwidth), because while they do not have any extra information they would still have to encode it as such.

No, c500 cannot give you 4:4:4 at a native resolution, even if it is encoded as such. What it can as per the manufacturer's specs is to give you either 2k or 1080p at 4:4:4 by, wait for it, downsampling the 4k sensor data.
Now, even if they make a ProRes 4444 doesn't mean that the data that written is actually 4:4:4 with an alpha channel. Just like because canon 5dmarkII making 1080p files doesn't mean that there is actually 1080 lines of vertical resolution in there (closer to 600, tops)

Indeed you could have a yes or no answer, but instead you can have an explanation as of why, so next time a sales person tells you that their CMOS Bayer pattern single sensor camera delivers 4:4:4 color at a native resolution, you would be wiser and take your money elsewhere.

Chroma sub-sampling of a video file (4:4:4, 4:2:2, 4:2:0 etc) and the design of bayer pattern sensors are two completely different considerations and should not be treated as the same thing.

It is absolutely true to say that the BMCC can deliver a 4:4:4 image after processing the Raw data.

4:4:4 as a description is just saying that no chroma sub-sampling has occurred in the video file. The act of Chroma sub-sampling is uniform for each level.

On the other hand, taking bayer pattern sensor data and debayering it into video data - whereby interpolation has to occur to give each pixel RGB values from each photosite - is NOT uniform and the quality and accuracy of the data is absolutely dependant on the debayering method.

It is completely true to say that if you look at a 2.5k resolution video frame from the BMCC after debayering - each pixel does not contain un-interpoalted RGB values. But the resulting image is not at all comparable to compressing the video with chroma-sub-sampling.

I understand and appreciate the overall point you are trying to make - oversampling the image at sensor level to be downsampled to a lower resolution later on WILL of course result in a higher level of Chroma and luma resolution - rather than from a bayer sensor which has for example 1920x1080 photosites for a 1080p finished video - but you are interchanging two different types of terminology and processes which are not at all comparable.

What Dmitry so expertly explained here comes down to a simple truth (please correct me if i got it wrong):

A 4:2:2 subsampled matrix is enough to store the full amount of information contained in raw bayer pattern data, using a 4:4:4 matrix would just waste storage space and/or bandwidth. 4:4:4 would only be usefull to store 4k images scaled down to 1080.

12 instead of 10 bit would make a difference though, in any resolution.

Not quite.

chroma sub-sampling an image, and debayering to get RGB values via interpolation are not comparable.

The debayering process does yeild a true 4:4:4 image - IE: an image with NO chroma sub-sampling having occurred.

After debayering from any resolution - the each pixel of the video file does contain full RGB data - 2/3 of the data has come from interpolation - the accuracy of the values depends on the debayering method. A 4:2:2 video file however - does NOT contain full RGB data in every pixel - regardless of resolution. The latter is a form of lossy compression.

The more you oversample for a given final resolution - the more resolving power you will have in terms of luma and chroma data (there is a factor of diminishing returns - but that is another issue entirely) but with debayering - there IS data there - just not as accurate data as if you had oversampled. With 4:2:2 - there is not data there, it has been discarded.

As a quick example, look at this image:
http://community.avid.com/cfs-filesyste ... amples.jpg

The top 4:4:4 image could have been derived from a bayer sensor - without any oversampling. Each pixel contains RGB data (some of which was interpolated so not actually detected). The Chroma data for each of those pixels might not be exactly accurate (compared to an oversampled sensor image) - but there is still some data actually there.

Then look at the 4:2:2 image - you can clearly see the effect it has on the image quality. The missing data is not replaced by any interpolated values during playback - it is missing, it has been thrown away.

I hope this clears up the confusion

Tom, I respectfully disagree. Just because it says 4:4:4 it is not so. I understand that it is important not to confuse what is coming from the sensor or camera processor and a compression scheme.
That being said, debayering algorithms are not magic. They are math. If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it. While you might have a full RGB data, does not mean it is a valid data.
Here is an example I did for students once. Lets take a canon DSLR and the video it records. The claim is that the video they record is 1920 by 1080. But that is not what being send to the encoder. I'd be pressed to say we get more than half of it is a true data.
I shot the same scene with a 7d and an old canon xl2 3 ccd SD camera. Did an uprez of the xl2 standard def footage to 1080p, and it looked better, and seemed to have better resolution than "natively" 1080p footage coming from DSLR.
And you can take your 4:2:2 footage coming after being encoded as such, from a Bayer pattern sensor and blur your chroma channels in post, et voila you now have full RGB data for each pixel. We are just shifting interpolation to post.
Now I might be still wrong, so if I am please explain, how does debayering recreates the valid data where there was none.

Tom, I respectfully disagree. Just because it says 4:4:4 it is not so. I understand that it is important not to confuse what is coming from the sensor or camera processor and a compression scheme.
That being said, debayering algorithms are not magic. They are math. If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it. While you might have a full RGB data, does not mean it is a valid data.
Here is an example I did for students once. Lets take a canon DSLR and the video it records. The claim is that the video they record is 1920 by 1080. But that is not what being send to the encoder. I'd be pressed to say we get more than half of it is a true data.
I shot the same scene with a 7d and an old canon xl2 3 ccd SD camera. Did an uprez of the xl2 standard def footage to 1080p, and it looked better, and seemed to have better resolution than "natively" 1080p footage coming from DSLR.
And you can take your 4:2:2 footage coming after being encoded as such, from a Bayer pattern sensor and blur your chroma channels in post, et voila you now have full RGB data for each pixel. We are just shifting interpolation to post.
Now I might be still wrong, so if I am please explain, how does debayering recreates the valid data where there was none.

For clarity's sake, I will break down this response to address each section of your reply (in other words, I am not trying to come across as blunt or rude)

1: "Just because it says 4:4:4 it is not so"
"4:4:4" is a description of chroma sub-sampling - and nothing else. This is not subjective. As the name suggests, sub-sampling is a reductive process, Chroma sub-sampling is a description of reducing the number of chroma samples in a video. It does not refer to any other form of chroma resolution or bayer data or anything else. To describe something as 4:4:4 is to simply say that no chroma sub-sampling has been applied the footage. It is not another way of describing an undersampled image or bayer data or anything else. As a matter of fact - not opinion, non pro-res/DNxHD footage from the BMCC does not have chroma sub-sampling applied, therefore it is true to call it 4:4:4.

This is of course not to say, that the footage has full Raw RGB data for every photosite - it is true that every photosite on the sensor will only contain luminance data for either Red, Green or Blue.

2: "debayering algorithms are not magic. They are math. If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it."
Yes, Debayering algorithms cannot calculate the "true*" missing chroma values for each pixel, but they do calculate something. They calculate an interpolated value which, depending on the quality of the algorithm, might be very close to whatever the "true*" value should be. At the end of the process, there IS RGB data for each pixel - whether or not this data is accurate enough or not is not something I am debating - the fact is, there is data there.

3:" If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it. While you might have a full RGB data, does not mean it is a valid data. "
The interpolation WILL produce data, as I have said, not exactly the same as a non interpolated RGB pixel, but its still chroma data none the less. The validity of this data can be debated, but the main reason chroma sub-sampling is such an efficient way to save space and bandwidth, is because our own sensitivities to colour is less so than luminance - as such, minute mistakes from the interpolation process are unlikely to be spotted without pixel peeping, and even then, the difference is not massive.

4:"how does debayering recreates the valid data where there was none."
The key word here is "valid". The validity can be debated, the point I am making, is that there IS data there. There is full chroma data.

Now that these points have been considered, it follows on to compare this to a 4:2:2 video file.

A 4:2:2 video file does NOT have full RGB data per pixel. Accurate or not, the data is simply not there. The image will appear less sharp overall and will have had a uniform process applied whereby chroma data (if it previously existed, from oversampling OR from interpolation), has been discarded.

You can debayer in different ways to produce good or bad chroma results from a bayer sensor - but the result is full RGB data. On the other hand, if you apply chroma sub-sampling to something, the effect is always the same, and you end up with less chroma data - not even less accurate chroma data, but just less data, data where there previously was at least something.

The overall point here is that while the resolution of the Chroma channel can be affected by many different things, the process of Chroma-subsampling should not be confused with nor interchanged with having a bayer pattern sensor and not having RGB data per photosite. There is quite a difference, both practically and technically.


*True was in quotations because objectively there is no such thing as a true value of an analogue source, they are infinitely divisible and cannot be quantised to an exact value. In this instance I was more referring to what colour value would likely be derived from a different type of sensor which would capture full RGB data at every photosite

Tom, I respectfully disagree. Just because it says 4:4:4 it is not so. I understand that it is important not to confuse what is coming from the sensor or camera processor and a compression scheme.
That being said, debayering algorithms are not magic. They are math. If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it. While you might have a full RGB data, does not mean it is a valid data.
Here is an example I did for students once. Lets take a canon DSLR and the video it records. The claim is that the video they record is 1920 by 1080. But that is not what being send to the encoder. I'd be pressed to say we get more than half of it is a true data.
I shot the same scene with a 7d and an old canon xl2 3 ccd SD camera. Did an uprez of the xl2 standard def footage to 1080p, and it looked better, and seemed to have better resolution than "natively" 1080p footage coming from DSLR.
And you can take your 4:2:2 footage coming after being encoded as such, from a Bayer pattern sensor and blur your chroma channels in post, et voila you now have full RGB data for each pixel. We are just shifting interpolation to post.
Now I might be still wrong, so if I am please explain, how does debayering recreates the valid data where there was none.

For clarity's sake, I will break down this response to address each section of your reply (in other words, I am not trying to come across as blunt or rude)

1: "Just because it says 4:4:4 it is not so"
"4:4:4" is a description of chroma sub-sampling - and nothing else. This is not subjective. As the name suggests, sub-sampling is a reductive process, Chroma sub-sampling is a description of reducing the number of chroma samples in a video. It does not refer to any other form of chroma resolution or bayer data or anything else. To describe something as 4:4:4 is to simply say that no chroma sub-sampling has been applied the footage. It is not another way of describing an undersampled image or bayer data or anything else. As a matter of fact - not opinion, non pro-res/DNxHD footage from the BMCC does not have chroma sub-sampling applied, therefore it is true to call it 4:4:4.

This is of course not to say, that the footage has full Raw RGB data for every photosite - it is true that every photosite on the sensor will only contain luminance data for either Red, Green or Blue.

2: "debayering algorithms are not magic. They are math. If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it."
Yes, Debayering algorithms cannot calculate the "true*" missing chroma values for each pixel, but they do calculate something. They calculate an interpolated value which, depending on the quality of the algorithm, might be very close to whatever the "true*" value should be. At the end of the process, there IS RGB data for each pixel - whether or not this data is accurate enough or not is not something I am debating - the fact is, there is data there.

3:" If there is no information in a blue pixel about the green channel, no amount of interpolation will recreate it. While you might have a full RGB data, does not mean it is a valid data. "
The interpolation WILL produce data, as I have said, not exactly the same as a non interpolated RGB pixel, but its still chroma data none the less. The validity of this data can be debated, but the main reason chroma sub-sampling is such an efficient way to save space and bandwidth, is because our own sensitivities to colour is less so than luminance - as such, minute mistakes from the interpolation process are unlikely to be spotted without pixel peeping, and even then, the difference is not massive.

4:"how does debayering recreates the valid data where there was none."
The key word here is "valid". The validity can be debated, the point I am making, is that there IS data there. There is full chroma data.

Now that these points have been considered, it follows on to compare this to a 4:2:2 video file.

A 4:2:2 video file does NOT have full RGB data per pixel. Accurate or not, the data is simply not there. The image will appear less sharp overall and will have had a uniform process applied whereby chroma data (if it previously existed, from oversampling OR from interpolation), has been discarded.

You can debayer in different ways to produce good or bad chroma results from a bayer sensor - but the result is full RGB data. On the other hand, if you apply chroma sub-sampling to something, the effect is always the same, and you end up with less chroma data - not even less accurate chroma data, but just less data, data where there previously was at least something.

The overall point here is that while the resolution of the Chroma channel can be affected by many different things, the process of Chroma-subsampling should not be confused with nor interchanged with having a bayer pattern sensor and not having RGB data per photosite. There is quite a difference, both practically and technically.


*True was in quotations because objectively there is no such thing as a true value of an analogue source, they are infinitely divisible and cannot be quantised to an exact value. In this instance I was more referring to what colour value would likely be derived from a different type of sensor which would capture full RGB data at every photosite

Tom, do not worry about being rude. Any new information is always welcomed and pointing out misconceptions and misunderstanding will never be perceived as a personal attack. I still disagree with some of your points, however.

Tom, do not worry about being rude. Any new information is always welcomed and pointing out misconceptions and misunderstanding will never be perceived as a personal attack. I still disagree with some of your points, however.

I greatly appreciate your understanding and commend your attitude

In my post I tried to refrain from editorialising any of my points and attempted to stick to facts - as such I do not see how you can disagree with anything I have said.

As far as I know, nothing I have said is untrue or inaccurate, I also appreciate and welcome new information, I would never be so arrogant to assume I am never wrong, so if you wish to do so - please do highlight any areas you feel are not true.

Perhaps this is a clearer explanation (not as thorough by easily more to the point)
http://www.reduser.net/forum/archive/in ... 20528.html

"Demosaicing and subsampling are performed in completely different colour spaces, GRGB and Y’CbCr, respectively. They are also two very different processes. Subsampling reduces bandwidth by reducing resolution. Information is lost in that process. Demosaicing, on the other hand, is a method of interpolation in which none of the existing data is compromised."

I believe the different understandings revolve around the concepts of "data" and "information" contained therein.

Let me try to summarize what i think i've learned from both of you: debayering is a process that creates additional data, but it cannot create additional information. Subsampling 4:2:2 is a process that destroys data (and thus information). In theory the 4:2:2 subsampled images hold as much or more data as the raw footage; yet in praxis it might not hold as much information as raw, because the data lost in subsampling isn't neccessarily just the data created by debayering...

I believe the different understandings revolve around the concepts of "data" and "information" contained therein.

Let me try to summarize what i think i've learned from both of you: debayering is a process that creates additional data, but it cannot create additional information. Subsampling 4:2:2 is a process that destroys data (and thus information). In theory the 4:2:2 subsampled images hold as much or more data as the raw footage; yet in praxis it might not hold as much information as raw, because the data lost in subsampling isn't neccessarily just the data created by debayering...

My points werent based on semantics, but the differences between two completely different things which were being interchanged in this discussion thread.

One should not describe a bayer sensors chroma resolution using terminology which exclusivly describes a compression method. The two are seperate and their impact and influence in the image are not the same either.

Anyone who has ever compared the visual difference between a 4:4:4 video file and 4:2:2 video file, will know that in reality there is not a big difference - such is the efficiency of chroma sub-sampling. (the only exception here is when using chroma key - getting a 4:4:4 alpha compared to 4:2:2 is more noticable)

To try and better display the points I raised in my previous posts - I have produce this photoshop file.

I imported a raw 2.5k sequence from my BMCC into Resolve - applied a basic colour grade (contrast and sat) - exported it at 1080p in both a 4:4:4 cineform video file, and a 4:2:2 cineform video file. Both set to the highest quality setting.

I have blown up a section of the image where one layer is from the 4:2:2 video file and the other is from the 4:4:4 video file.

The difference is not huge, but it is there.

This clearly demonstrates that the process of chroma subsampling the bmcc raw image does discard colour data - and thus is not comparable to the interpolation from a bayer sensor -ie: not having RGB data per photosite.

http://we.tl/XHJCJrEYu1 <---PSD download file


So how big is the difference when the original source is from a camera which DOES have an oversampled image sensor?

Here is the same demonstration but this time from an Arri Alexa ProRes 4:4:4 frame of video:
http://we.tl/6x97tV3XkA

The relative difference is roughly the same.

It's really wrong to describe a bayer sensor has having only 4:2:2 worth of colour information. It's an oversimplification and an assumption about the ratio.

A lot of people like to talk about the fact that a de-bayered sensor is like a 4:2:2 file because of the ratio of red green and blue pixels, which happens to be the same.

It's really not correct though to extrapolate that out to compare them, however convenient the leap and correlation to the ratio is.

4:2:2 relates to "encoded" video. This number only applies to actual encoded video, like what we get with ProRes.

A bayer sensor image is created through an entirely different mathematical process that then you can create the encoded video from. But many things affect the creation of that image. The de-mosiac algorithm, the array or pattern of the pixels not to mention the bit depth, the processing pathways, the way the file is stored, gamma encoding used, compression....

It's just really really simplistic to extrapolate the 4:2:2 number of encoded video with the ratio of red to green to blue pixels.

jb

It's really wrong to describe a bayer sensor has having only 4:2:2 worth of colour information. It's an oversimplification and an assumption about the ratio.

A lot of people like to talk about the fact that a de-bayered sensor is like a 4:2:2 file because of the ratio of red green and blue pixels, which happens to be the same.

It's really not correct though to extrapolate that out to compare them, however convenient the leap and correlation to the ratio is.

4:2:2 relates to "encoded" video. This number only applies to actual encoded video, like what we get with ProRes.

A bayer sensor image is created through an entirely different mathematical process that then you can create the encoded video from. But many things affect the creation of that image. The de-mosiac algorithm, the array or pattern of the pixels not to mention the bit depth, the processing pathways, the way the file is stored, gamma encoding used, compression....

It's just really really simplistic to extrapolate the 4:2:2 number of encoded video with the ratio of red to green to blue pixels.

jb

Thanks John,

In one single post you have better explained the point I have been trying to express

Thanks Tom and John.

This has been an interesting read. I was also under the misconnection that the de-bater process threw away data the same was as the 4:2:2 compression process.

Live and learn.

Thanks Tom and John.

This has been an interesting read. I was also under the misconnection that the de-bater process threw away data the same was as the 4:2:2 compression process.

Live and learn.

4:2:2 refers to chroma subsampling.

Which is NOT the same as de-mosiacing.

jb

Basically, deBayering without down-sampling produces an RGB (4:4:4) image with less actual measurable resolution than the pixel count might suggest. But it is not less resolution in exactly the same way as 4:2:2 chroma sub-sampling would cause.

So by staying 4:4:4 after deBayer you are preserving a better image than a 4:2:2 sub-sampled image could hold, but you are "wasting" some storage.

So by staying 4:4:4 after deBayer you are preserving a better image than a 4:2:2 sub-sampled image could hold, but you are "wasting" some storage.

It's a matter of opinion as to whether the data is valuable or a waste.

It's a matter of opinion as to whether the data is valuable or a waste.

Hence the quotes!

You may not be storing quite as much information as the pixel count would suggest, but there is still a good case for keeping everything you do have, and not throwing any of it away.

I think you and I are in agreement on this.

It's a matter of opinion as to whether the data is valuable or a waste.

Hence the quotes!

You may not be storing quite as much information as the pixel count would suggest, but there is still a good case for keeping everything you do have, and not throwing any of it away.

I think you and I are in agreement on this.

My apologies - we both are in agreement

Thanks Tom and John.

This has been an interesting read. I was also under the misconnection that the de-bater process threw away data the same was as the 4:2:2 compression process.

Live and learn.

4:2:2 refers to chroma subsampling.

Which is NOT the same as de-mosiacing.

jb

Yeah... As I said, I understand that now.

Cheers

Thanks Tom and John.

This has been an interesting read. I was also under the misconnection that the de-bater process threw away data the same was as the 4:2:2 compression process.

Live and learn.

4:2:2 refers to chroma subsampling.

Which is NOT the same as de-mosiacing.

jb

Yeah... As I said, I understand that now.

Cheers

Ditto! Thanks, Tom and John for being so patient and such gentlemen about this.

I was guilty of assuming that the debayer process say of a 1920x1080 frame represented the same loss as the 4:2:2 chroma subsampling of the same size and it could be rectified by debayering from 3840x2160 down to 1929x1980. Now I understand that I was wrong so apologies for my erroneous posts earlier.

Now I understand why some people generated ProRes 4:4:4:4 from the raw files rather than something less than that. It seems to me if one is going to shoot raw, it is a false economy to be doing your post work in ProRes 4:2:2 HQ. I think if you are shooting raw, suck it up and prepare to use more storage space than just for your original raw footage. This argument assumes you are not going back to Resolve to render your final deliverables based upon your original raw footage. If you do go back to Resolve, sure you may not have a benefit from doing 4:4:4:4 in post NLE etc. except for green screen and special effects.

Rick Lang
Sent from my iPad using Tapatalk HD

My apologies - we both are in agreement

No problem.

It seems to be one of those things about the internet, that everything appears more confrontational than it is. Maybe all discussions should be had face to face over a beer

My apologies - we both are in agreement

No problem.

It seems to be one of those things about the internet, that everything appears more confrontational than it is. Maybe all discussions should be had face to face over a beer

I'll drink to that!

BM very seldom comment on here. So I doubt you'll get an answer to your question. Being rude and obnoxious will win you no friends on here.

The fact is 12bit 444 at 1080p is MAYBE possible. 12bit 444 at 4K is NOT possible because the data is not there after the de-Bayer process.

To get 12bit 444 at 4K you would need a 3CCD camera or a much higher resolution sensor that can be down sampled to produce 12bit 444.

LOL rude and obnoxious?

really now hahah!! thanks and not looking for a friend just looking for A SIMPLE answer.

Thanks Tom and John.

This has been an interesting read. I was also under the misconnection that the de-bater process threw away data the same was as the 4:2:2 compression process.

Live and learn.

4:2:2 refers to chroma subsampling.

Gratz on enlightening the people about this. I can't believe my simple post got many people argue over this simple question. And (LMAO) calling me rude and stuff..got to love that! Being real isn't rude people.... Live with ET!

Which is NOT the same as de-mosiacing.


BTW John thanks for passing out the BMD Pocket camera files. Got to play with it pretty neat camera. We order a few still waiting for it. Hopefully the RAW update is available by then. Taking it to China to film some Kung Fu grandmasters.

Keep it up bro!

Cheer
jb

My apologies - we both are in agreement

No problem.

It seems to be one of those things about the internet, that everything appears more confrontational than it is. Maybe all discussions should be had face to face over a beer

Hey Nick you in here too!! haha Stumble on ur site way back then. Love reading it...and try some of your LUT pretty neat stuff.

Cheer Bro!

[quote="Tom"]Anyone who has ever compared the visual difference between a 4:4:4 video file and 4:2:2 video file, will know that in reality there is not a big difference - such is the efficiency of chroma sub-sampling. (the only exception here is when using chroma key - getting a 4:4:4 alpha compared to 4:2:2 is more noticable)

here a portion from a quote from my earlier post that many people didn't read and just start passing out or start schooling people on this thread.

"Reason I ask this question is we got a project coming up that will deal with a lot of VFX and green screen work and we would like the ability to record to 444 chroma sub "

I just want to get a cleaner and easier pull of all the stuff we are doing. If we can have the extra data then we are going to be happy if not going just stick at what we have n do our best.

THanks for pointing this out to the people that don't know what it is dude!! lol now maybe people will get it..and enough about the vodoo magic on the sensor this..sensor that...and stuff.

Cheer man!

[quote="adamroberts"]BM very seldom comment on here. So I doubt you'll get an answer to your question. Being rude and obnoxious will win you no friends on here.

The fact is 12bit 444 at 1080p is MAYBE possible. 12bit 444 at 4K is NOT possible because the data is not there after the de-Bayer process.

[To get 12bit 444 at 4K you would need a 3CCD camera or a much higher resolution sensor that can be down sampled to produce 12bit 444.

"Reason I ask this question is we got a project coming up that will deal with a lot of VFX and green screen work and if we the ability to record to 444 chroma sub @12bit to the 4k cine cam

lmao dude read that quote which part did it say RECORD 12bit 444 "@4K res"? it clear say record 444 chroma sub @12bit "TO" again i repeat "TO" the 4K cine cam...not record AT 4K res....way off the subject here buddy.

cheer