Externally recorded BRAW and feature films

Hi all, it would be great to get unbiased comments for the below. Reading online articles and watching videos on BRAW vs Prores RAW, the general consensus, it seems, is that BRAW recorded externally e.g. using the BM Video Assist monitor on a LUMIX S1H (or any non-Blackmagic camera), is inferior to BRAW recorded internally in a BM camera and also inferior to Prores RAW recorded externally e.g. using an Atomos Ninja on a LUMIX S1H. For e.g. it is said that BRAW is not true RAW compared to Prores RAW as BRAW is partially debayered in camera.

The million dollar question 1 is, if a feature film was shot on a LUMIX S1H via externally recorded BRAW, post-produced and color graded etc. and if the same feature film was shot on a LUMIX S1H via externally recorded Prores RAW, post-produced and color graded etc. - will you be able to tell the difference between both movies and will the BRAW shot film look less quality/inferior to the Prores RAW shot film?

Question 2 - do you know of any feature film[s] that was shot using externally recorded BRAW?

BM = Blackmagic Design

You could pixel peep anything to death. In the end, it doesn't really matter. What matters in a feature is a good script and then good audio. I know that doesn't answer your question as posed but who here wouldn't prefer watching a Jordan Peele movie shot on a Nokia cell phone over watching a poorly scripted feature shot on IMAX cameras? In the end, you have a proper camera. I'd just try shooting a clip of each and see which one YOU like best.

There's a widely quoted youtube test, conducted with the SH1, which deals with this very issue, which you must have seen by now(?) All else being equal -- which is never the case -- some may prefer Prores raw over braw. But other tests, performed by an actual engineer, Adam Wilt, did not appear to come to the same conclusion.

In practice, proresraw is a lot less convenient than braw, and any conceivable difference will be overwhelmed by the lighting, cinematography and quality of the color grade. You could also shoot internally on the SH1, and that won't make any difference either on a very low budget production. These are movies after all, and of some presumed human interest, not pixel analysis exercises.

If you're inclined to worry yourself to death at 3 a.m., think of it this way: there's going to be so much else wrong with my feature film, that the difference between two quasi-raw formats is tiny, invisible to 99.7% of viewers (and half of the other .3% are probably lying.)

OTOH, to return to senseless technical obsessions, a feature film is a large undertaking. Surely it's worth a test? If you consider yourself the typical viewer, can you see any difference or not? And what does the colorist tell you?

The other question you might explore is to what extent is external braw used in professional productions -- and are you up to *their* standard? JB could be of help here.

At the risk of sounding like my cynical self -- if you're asking about shooting a feature film on an S1H and recording externally to a 3rd party device...there's a pretty good chance that your film will end up on some streaming platform at best, being watched in glorious 8-bit highly compressed h.264 or h.265.

My point is that any slightly differences that you might see between formats will be obliterated in the end.

Use what's most convenient to you, and don't sweat the small stuff.

Here's the link to Adam Wilt's test referenced above:

https://www.cinematography.net/2021-raw-tests/index.html

Adam Wilt is as objective and unbiased as there is.

But on the point of BRAW not being a "real" raw format because it is partially debayered in-camera, consider that:

1.) The qualifying definition of true raw as 100% debayered externally was arbitrary. There were good reasons for doing a partial debayer in-camera as Grant explained in one of his videos, namely that it makes sense to leverage the hardware built in and braw together because they are "in" the camera, and doing so relieves your decoder of a burdensome drawdown of resources. The in-camera hardware decoding is built into and already in the camera, so it makes sense to do a partial debayer there for resource conservation, while the partial debayer that happens in the decoder avails to you all of the same controls in the Resolve raw panel (and more).

2.) The not "real" raw myth of partial debayer has been used by the brand shills to try and marginalize BRAW when the strictest interpretation of "true" raw is sensor voltages, but all raws from Sony, BMD, Red, Apple are compressed raws, either DCT or wavelets and as such, are codecs not raw.

3.) You can obsess which looks best but it is a fool's errand. What's not is the speed, flexibility and convenience differences.

Great post Tom. I agree.

It’s funny. For me now when I see phrases like not real raw or pure pixels they’re all kind of made up phrases.

How can something be more raw when it’s compressed.

Or how can it really be raw if the underlying data is pre-processed for noise reduction. Or even dead pixel correction. Or sensor calibrations.

It’s not a great logic.

I also watched this very recently and though it’s a long watch, it does explain the underlying process of raw video very well. You may or may not be offended by the political messaging about a certain camera brand but I do think he explains how raw video at a sensor level works fairly well.

JB

Addending a comment to Kay's post above.

Walter Volpatto commented in an hour long online presentation, that he'd tested a number of film grain emulation procedures. The test included both in-theater projection and over-TV viewing on large TVs with a Hollywood crowd audience.

From "real" film overlays to simply Resolve's noise/grain OFX.

And no one could see a difference. Because, primarily, the compression of the deliverable typically specified for each type of presentation, and the translation of the viewing system back to a visual image.

I'd love to go through that whole presentation again of course. But this part really hit home:

*Don't waste time on what will never make it to the viewer's screen.*

Sent from my SM-S908U using Tapatalk

I also watched this very recently and though it’s a long watch, it does explain the underlying process of raw video very well...but I do think he explains how raw video at a sensor level works fairly well.

JB

I agree. Although it appears the Jinni Tech's purpose is to refute claims of no prior art, that clearly did exist as prior art within the demosaic's of Cineform; the Jpeg2000 encoder/decoders embedded into Xilinx FPGA's; VLSI wavelet compression chips from Analog Devices; the use of such chips in video boards from Atomix; in cameras the SiliconImaging SI-2K; even the social media discussions of DVInfo.net and David Newman and Cineform Raw that I remember well. Leaving that aside, John you are right that the underlying process of raw video is explained at the sensor level very well. Compressed raw is not raw; all the modern raw codecs are compressed; claiming that a brand of raw is not "true raw" and the arbitrary partial in-cam demosaic myth along with it; flipping a speed enhancing, resource conserving feature of partial in-cam demosaic as a disqualifier for truth in raw is quite patently stupid.

Hi all, original poster here.

This is to say thank you for all your great and interesting replies and links etc. It seems it is all in others' heads when they try to talk down BRAW in some way, looking at the facts and tests done so far by others.

Here is wishing all of you a great 2023!

Hi all, it would be great to get unbiased comments for the below. Reading online articles and watching videos on BRAW vs Prores RAW, the general consensus, it seems, is that BRAW recorded externally e.g. using the BM Video Assist monitor on a LUMIX S1H (or any non-Blackmagic camera), is inferior to BRAW recorded internally in a BM camera and also inferior to Prores RAW recorded externally e.g. using an Atomos Ninja on a LUMIX S1H.

If it were me, I'd just tell the client to record using 10- or 12-bit ProRes 444 and be done with it. That will still give you the range to correct what you need to correct.

I'm not saying that Raw is useless -- I'm saying with a $3000 camera, it kind of doesn't matter. All the intangibles, like the lighting and the art direction and the composition and the exposure, all matter a lot more. Get all those things right, and you can make a fine-looking show or film just with ProRes 444. Many people shot with Alexa on ProRes 444 for the last 10 years for network TV shows, and they were fine. Would we prefer the Raw? Sure, but it's not a gigantic and unsubtle improvement over the 444 files. Do a test for yourself and see if there's a visible difference.

Question 2 - do you know of any feature film[s] that was shot using externally recorded BRAW?

https://www.amazon.com/Angel-Mountain-D ... op?ie=UTF8

[
I'm not saying that Raw is useless -- I'm saying with a $3000 camera, it kind of doesn't matter. All the intangibles, like the lighting and the art direction and the composition and the exposure, all matter a lot more. Get all those things right, and you can make a fine-looking show or film just with ProRes 444. Many people shot with Alexa on ProRes 444 for the last 10 years for network TV shows, and they were fine. Would we prefer the Raw? Sure, but it's not a gigantic and unsubtle improvement over the 444 files. Do a test for yourself and see if there's a visible difference.

The reason for “raw” capture is that for the same K count and compression ratio per channel as an RGB codec it is roughly 1/3rd the data rate because only one value per photosite needs to be recorded verses in an RGB codec three values per pixel need to be recorded. As long as the decode isn’t too processor intensive (and with a codec like BRAW it is in some cases less intensive than even ProRes), “raw” recording means less compression is required at the same data rates as 12bit ProRes4444. Regardless of the price of the camera, less compression is always better. Also, at lower price point cameras, 444 codecs are not an option, so “raw” recording avoids 422/420 subsampling.

Also, at lower price point cameras, 444 codecs are not an option, so “raw” recording avoids 422/420 subsampling.

Absolutely true. As a Nikon photographer, the ability to record raw video externally with the Nikon Z6/Z7 series cameras with the BMD VA or Ninja V makes a huge difference in quality. The quality of the internal H.264 in these cameras is poor and the jump in quality to raw (ProRes and BRAW) is massive. Even compared to 10 bit HEVC, the raw video is significantly better when it comes to grading. I would have had to change systems when I got into video if not for the ability to shoot raw. Add the ability to modify white balance easily and cleanly in post and it's a no-brainer.

In relation to the original question, I think BRAW on the VA 12G just edges it over ProRes Raw on the Ninja V in terms of ability to recover highlight detail (less prone to clipping all else being equal).

Forget 444. Most people can't see the difference between 442 and raw, and even at the best of times differences are subtle. And these are moving pictures, after all. They're not meant for close study, even assuming they're seen in a theater on not on somebody's phone.

It's true that reduced storage needs of (say) braw Q5 versus Prores are significant, but that's less so with other raw formats and still less against the varieties of 10 bit 422 h.264 recording, as available on the S1H, which are real options today for impecunious productions.

Even people who know better worry more about bit rates than their tone-deaf dialogue, no matter how low the budget, so it's plain that these lunatic obsessions are compelling. But best not encouraged.....

Forget 444. Most people can't see the difference between 442 and raw, and even at the best of times differences are subtle.

Yup! I get into argument with people when they ask for delivery of ProRes4444XQ vs a perfectly acceptable ProResHQ. Especially when it comes to VFX pulls, I think ProRes4444 is overkill except for green/blue screen shots where there might be a marginally better amount of data for the keyer to work with.

On my own projects I send out all my VFX pulls as ProResHQ which works splendidly. Of course if I'm working for another client I'll give them what they ask for despite how overkill I think it is.

Happy New Year everyone. Santa brought me a BMPCC 4K which I'm thrilled to own at long last.

I'm just about to start a small project which depends on green screen, so any advice on the best format to use will be much appreciated. I'm guessing BRAW at highest quality so as to get the best key via Fusion?

(I owned the original pocket cinema camera back in 2015, and the quality was beautiful, but just couldn't find a way to use it easily, so sadly sold it. The footage from the 4k is stunning, I love it!)

Rob

I hear what a lot of people are saying about the differences between the CODEC's is negligible by the time it reaches the viewers. The reason I use braw is to have greater flexibility when editing, and that goes hand in hand with the flexibility of node based editing on Resolve's Colour Page.

Given sufficient bit depth any codec can have the flexibility of raw and this is partially the principle behind Braw, being 12bit. The biggest advantage with Prores 444, for high end cinema capture, outside of critical green screen work, is perhaps being 12bit also as opposed to 10bit of 422HQ. But in all practical purposes and with well exposed material, the qualitative differences may be hard to notice or justify, as has been mentioned. For example, on a 6K sensor like the BMD ones 10bit shooting in log may be sufficient to capture the full DR of the sensor, whereas it might not be with something like the Arri Alexa. The only theoretical advantage of a true 'raw' codec may be being able to take advantage of future improved demosaicing algorithms, where that might not be fully possible with something like Braw.

If the film story, acting and/or production design sucks, no will will watch the film. At this point, it doesn’t matter what codec or Raw you used other than for the technical reasons already discussed here. Also, regular viewers, who are the main audiences, will not pixel peep or even know have to discern the image quality when on screen. It is a moving picture after all.

The only theoretical advantage of a true 'raw' codec may be being able to take advantage of future improved demosaicing algorithms, where that might not be fully possible with something like Braw.

It's not only possible with BRAW, it's been done with BRAW in the 12K. First of all, it's a unique RWGWBW 6x6 sensor matrix, and secondly, BMD have already made demosaic alterations through 12k camera firmware updates.

It's not only possible with BRAW, it's been done with BRAW in the 12K. First of all, it's a unique RWGWBW 6x6 sensor matrix, and secondly, BMD have already made demosaic alterations through 12k camera firmware updates.

Have you any 'white paper' proof for such a theoretical assertion?

Not theoretical and better than white papers, I demonstrated an anomaly in my 12k footage related to the demosaic which caused the involvement of BMD engineers at the highest level, and privy to beta test of what was then an experimental firmware that was incorporated into later releases.

Not theoretical and better than white papers, I demonstrated an anomaly in my 12k footage related to the demosaic which caused the involvement of BMD engineers at the highest level, and privy to beta test of what was then an experimental firmware that was incorporated into later releases.

Thanks Tom but I'm not sure this proves what you're saying. I'm talking about taking raw sensor data and later re-processing that with better demosiac algorithms than were perhaps originally unavailable, theoretically speaking of course, which as I alluded was of perhaps little practical merit. I understand you can't talk because of your involvement at the 'highest level' but I still don't follow. Thanks again.

If the film story, acting and/or production design sucks, no will will watch the film. At this point, it doesn’t matter what codec or Raw you used other than for the technical reasons already discussed here. Also, regular viewers, who are the main audiences, will not pixel peep or even know have to discern the image quality when on screen. It is a moving picture after all.

Avatar was shooted in FHD betacam SR (h264 part 10, 10 bit 4:2:2), they key without problem the chroma key and this high compression not be a problem to be the biggest box office of movie history (it win over EndGame this summer with the re-release).
if you record a good prores, braw you are over of original JC shooting, i think you can dedicate the rest of efforts on lighting, framing, screenplay and more ;-D

Thanks Tom but I'm not sure this proves what you're saying. I'm talking about taking raw sensor data and later re-processing that with better demosiac algorithms than were perhaps originally unavailable, theoretically speaking of course, which as I alluded was of perhaps little practical merit. I understand you can't talk because of your involvement at the 'highest level' but I still don't follow. Thanks again.

I respect the involvement of BMD at the highest level and will not disclose; but BMD demosaic changes are published among the updates for Blackmagic Camera Setup 7.7 for the 12K. I am not aware of reciprocal examples where this has been done for so called "true" raws that you assert could have future improved demosaicing algorithms. That seems unlikely in view of the legacy 422 bayer sensor designs upon which such futuristic demosaicing algorithms would be based.

URSA 12K 7.7 FW Testing:

I am not aware of reciprocal examples where this has been done for so called "true" raws that you assert could have future improved demosaicing algorithms. That seems unlikely in view of the legacy 422 bayer sensor designs upon which such futuristic demosaicing algorithms would be based.

Interesting, but you're talking camera firmware and I was talking about the codec demosiacing algorithms used in post, when these cameras may have long since disappeared. It's not really a reason to insist upon 'real' raw anyhow as I alluded and used the liberal may be and might throughout, though I do maintain it is very possible. Red are such an example of such continued refined raw codec development, for example.

If the film story, acting and/or production design sucks, no will will watch the film. At this point, it doesn’t matter what codec or Raw you used other than for the technical reasons already discussed here. Also, regular viewers, who are the main audiences, will not pixel peep or even know have to discern the image quality when on screen. It is a moving picture after all.

Avatar was shooted in FHD betacam SR (h264 part 10, 10 bit 4:2:2), they key without problem the chroma key and this high compression not be a problem to be the biggest box office of movie history (it win over EndGame this summer with the re-release).
if you record a good prores, braw you are over of original JC shooting, i think you can dedicate the rest of efforts on lighting, framing, screenplay and more ;-D

Exactly my point. However I am not reducing the need for the technical abilities that will be needed and understand it is often pretty complicated.

We seem to have detracted from the OP’s original inquiry.

Steve is correct in that philosophically, raw recording of the unmosaiced information, would allow later re-transcoding with a maybe better algorithm to improve the de-mosaic.

In reality, the improvement would be at best marginal and since I started shooting raw video, I’ve never once “gone back” to the negative for any show.

And therein is the issue. The footage has to have also been stored and kept along with all the source footage, and the editorial timeline to “reconstruct” the project, presumably also in some editing software project file that is many generations old. Who can open a V4 Resolve project with colour corrections all saved today?

Even in conventional film preservation and restoration, they typically do not use the camera originals, but use the interneg masters, the final delivery negative. If you only have to “pay” to archive the delivery masters then it’s a lot less expensive that archiving everything.

And most delivery contracts I’ve seen only specify keeping delivery items, not source materials.

So yes in theory, this is one of the advantages of shooting raw, but I can’t name a single projected where this has been done, and I’ve certainly never done it and I doubt that there would be more than a few % difference in image improvement anyway. Maybe in 50 years someone will try to go back and restore something I shot, but there are so many other problems with doing that before you even get to the camera originals and most restoration workflows are geared towards working with final delivery masters. Nobody in restoration is going back to camera negatives wholly, usually because they’re not typically saved, even studio movies….

At least with BRAW, you could export a BRAW truncated master of all the clips used in the final delivery as the source files. It still doesn’t save the colour decisions but at least you have the “camera originals” in a better than master delivery format that can be used in the same workflow.

JB

*It’s amazing that a lot of film preservation has happened because of bootleg copies of film prints. Many films get forgotten or don’t become cult films till many years later and source material is often lost forever.

It’s only “illegal” bootleg prints that have saved many films, often kept by collectors or projectionists.

I have a friend who is a film programmer and she was trying to find a film print of Tim Burton’s Batman and was told one doesn’t exist. The studio doens’t have any! So unless they strike a new print also wearing out their precious interneg doing so there’s no way to see the film that was shot and released on film in its original form. It’s not THAT old a film, it’s by a significant director and yet you can’t watch it as intended.

Even that bootleg process has gone today because of “the cloud”. One of the most annoying things I find is this fallacy we’ve been sold on the cloud…except the cloud doesn’t follow you in different countries (geoblocking) and the content can also disappear an then no one has a copy of it…

Totally agree, John! And that is why I said 'theoretical'. I don't think it it is a reason to choose so called 'true' raw. Braw, Prores Raw, Any raw, they are just as good as each other for all practical purposes and no one has proven Braw is not up to snuff. Thank God it is free from the patent thing and we are very lucky to have it.

*It’s amazing that a lot of film preservation has happened because of bootleg copies of film prints. Many films get forgotten or don’t become cult films till many years later and source material is often lost forever.

It’s only “illegal” bootleg prints that have saved many films, often kept by collectors or projectionists.

100%. And not only with film. I remember interviewing the head of the Internet Archive for a documentary project a few years ago, and he had boiled down their basic philosophy to the realization that, as he put it, "access drives preservation". He said that locking things away will cause them to eventually disappear, and the best way to protect media for the future is to instead open it up and share it as widely as possible.

He said that locking things away will cause them to eventually disappear, and the best way to protect media for the future is to instead open it up and share it as widely as possible.

Could YouTube, Vimeo, and the likes be the preservers for the next century? A lot of copies, as well as copies of all sorts of digital films, are uploaded to them.

. He said that locking things away will cause them to eventually disappear, and the best way to protect media for the future is to instead open it up and share it as widely as possible.

The problem is rights holders.

Many films or shows are sold to multiple rights holders who then hold those rights in different geographic territories, often for different terms. They sometimes release different versions of films too (depending on local compliance with classification requirements and even political censorship.

Many times the original rights holders cant be found. Or can’t be found to renew, don’t want to respond or the ownership can’t even be established.

There’s a fantastic documentary called Demon Lover Diary about the making of a movie. It’s near impossible to find a copy of it. There are a couple of prints floating in collections and I saw it as a film print about ten years ago.

I’m a producer on a movie and even I don’t know the details of all the sales made because I’m two steps removed from the sales agents and distributors of the film. It’s hard to even know what the rights status is for the movie in various territories. Local rights holders change offices, change ownership, change names. They still technically hold those rights but essentially warehouse the film.

The rights are in limbo. This is the problem for more obscure films in a digital streaming age. Like with bootlegging prints, piracy is sometimes the only way to find these films online.

JB

I'm just about to start a small project which depends on green screen, so any advice on the best format to use will be much appreciated.

To be quite honest, I haven't been thrilled with the quality of the green screen footage from the Pocket 4K due to unexpected noise even at low ISO's and proper lighting. The best advice that I can give you for keying green screen from that camera is to noise-reduce all the footage before you attempt to get a key.

You could pixel peep anything to death. In the end, it doesn't really matter. What matters in a feature is a good script and then good audio. I know that doesn't answer your question as posed but who here wouldn't prefer watching a Jordan Peele movie shot on a Nokia cell phone over watching a poorly scripted feature shot on IMAX cameras? In the end, you have a proper camera. I'd just try shooting a clip of each and see which one YOU like best.

Why would you prefer to watch a Jordan Peeler movie on anything?

Sure BRAW has its quirks (forced noise reduction and sharpening) but with some gaussian blur (170 radius to blur sharpening halos away) and texture pop filter in Resolve, the image using my S1H still has two big advantages over shooting 10 bit H264/H265 internally:

• 12 bit color
• more color information in dark areas and overall more consistency throughout the exposure range

6K BRAW actually decodes slower than 6K H265 on my M1 Max Macbook, so there's no performance advantage.

Speaking of feature films, nothing yet, but I did a few short films last year on the S1H using external BRAW and tons of filtering:

Sure BRAW has its quirks (forced noise reduction and sharpening) but with some gaussian blur (170 radius to blur sharpening halos away) and texture pop filter in Resolve, the image using my S1H still has two big advantages over shooting 10 bit H264/H265 internally:

• 12 bit color
• more color information in dark areas and overall more consistency throughout the exposure range

6K BRAW actually decodes slower than 6K H265 on my M1 Max Macbook, so there's no performance advantage.

Speaking of feature films, nothing yet, but I did a few short films last year on the S1H using external BRAW and tons of filtering:

I agree. Why would I shoot 10 bit. When I can shoot 12 bit.
I have moved to an S1h and a video assist. And am getting amazing results.