What BRAW Quality should I shoot my Senior Capstone in?

Hello everyone!

I will be recording my senior capstone which I plan to be around 10-12 minutes.
Will be doing so on a Panasonic S5II attached to a BM Video Assist 12G 5" with a 2TB Samsung T7.
I'll be shooting at 6K at 23.98 fps.
Usually when shooting BRAW, I use Q5 if unsure what quality to pick but this time I have 4TB total for the short (including music, rendered VFX, audio, and deliverables.)

I plan on using a lot of VFX for the project, mainly film grain and trippy visuals, though no green screen.

Would Q5 be good enough or do you all recommend something else? I don't want to shoot too high or too low of a bitrate. Either constant bitrate or constant quality would be fine.

From what you’ve described, Q5 will work just fine.

Bugger hell, I had to look up what a Senior Capstone was https://en.wikipedia.org/wiki/Capstone_course

Non the wiser still......................

Bugger hell, I had to look up what a Senior Capstone was https://en.wikipedia.org/wiki/Capstone_course

Non the wiser still......................

It's a thesis based on real-world skills. Whoever wrote that wiki article needs to lay off the wank button.

Bugger hell, I had to look up what a Senior Capstone was https://en.wikipedia.org/wiki/Capstone_course

Non the wiser still......................

To be fair, I had no idea what one was until senior year.

Based on the math I’ve seen on BMD’s website, Q5 is between 6:1 and 15:1. In the context of a student film with a bunch of grain added in post, you’ll be fine.

Based on the math I’ve seen on BMD’s website, Q5 is between 6:1 and 15:1. In the context of a student film with a bunch of grain added in post, you’ll be fine.

Maybe I went a bit overkill getting 4TB for the film, huh. I suppose that I'll use some of that space then for beta renders, different deliverables, and maybe a poster.

I tend to favor Q3 as a default, dropping to Q5 only if I need more recording time on the media.

Agree with Jim, but all Q values give good results.

For reference:

Q0: 2.0:1 - 4.0:1
Q1: 2.5:1 - 6.0:1
Q3: 3.5:1 - 9.0:1
Q5: 6.0:1 - 15:1

Totally possible that these values are tweaked between camera bodies and software updates but this is what I calculated from their website.

For reference:

Q0: 2.0:1 - 4.0:1
Q1: 2.5:1 - 6.0:1
Q3: 3.5:1 - 9.0:1
Q5: 6.0:1 - 15:1

Totally possible that these values are tweaked between camera bodies and software updates but this is what I calculated from their website.

This is a great chart.

Agree with Jim, but all Q values give good results.

Good to hear. Looks like it's not a situation this time like where 12:1 gave me leaves that looked like blocks while being larger in file size than my old H.264 10-bit 4:2:2.

I believe Braw is great in file size compared to ProRes but not comparable in efficiency to H.265 or 4. This generally gives it better quality though and of course full 444 12-bit color.

I tend to favor Q3 as a default, dropping to Q5 only if I need more recording time on the media.

Sounds like a good idea to follow.

I believe Braw is great in file size compared to ProRes but not comparable in efficiency to H.265 or 4.

Of course not, H.264/5 are GOP codecs, while BRAW and ProRes are I-frame only. But it's easier on the machine if you don't have specific hardware, and the quality doesn't suffer from fast motion.

I believe Braw is great in file size compared to ProRes but not comparable in efficiency to H.265 or 4.

Of course not, H.264/5 are GOP codecs, while BRAW and ProRes are I-frame only. But it's easier on the machine if you don't have specific hardware, and the quality doesn't suffer from fast motion.

I wonder if Blackmagic could create a GOP version of BRAW to allow higher compression ratios, but with a warning that it would be harder on the computer. Something like 12:1G, 18:1G, 20:1G, 22:1G. Just an idea. It would make 12:1 more useful as currently I find it only useful for interviews with very blurry backgrounds and clear skies. Leaves turn to blocky mush. Until then (which could easily be never) I'll stick to Q5 or higher.

I understand storage is rather cheap these days but on cameras that take advantage of BRAW, the solution to save space is either increasing the ratio or lowering resolution (cropping)...

I believe Braw is great in file size compared to ProRes but not comparable in efficiency to H.265 or 4.

Of course not, H.264/5 are GOP codecs, while BRAW and ProRes are I-frame only. But it's easier on the machine if you don't have specific hardware, and the quality doesn't suffer from fast motion.

For the same bitrate, an interframe encoding will provide a far better quality than an all-intra encoding.

I believe Braw is great in file size compared to ProRes but not comparable in efficiency to H.265 or 4.

Of course not, H.264/5 are GOP codecs, while BRAW and ProRes are I-frame only. But it's easier on the machine if you don't have specific hardware, and the quality doesn't suffer from fast motion.

For the same bitrate, an interframe encoding will provide a far better quality than an all-intra encoding.

Agreed. That is why to my knowledge All-I uses very high bitrates like in ProRes. I wonder how they got BRAW to be higher quality than ProRes HQ while being smaller. It's fascinating. Q5 BRAW at 6K is smaller than even ProRes 422 at 4K from the shoots I remember.

I feel like it is time for ProRes to become more effieciant like when DNxHD switched to DNxHR. The only major update Apple has seemed to do was ProRes Raw, which doesn't have the file size choice of BRAW anyway.

BRAW is only partially debayered. That helps to keep the amount of data lower.
Think of it: RAW from a Bayer sensor is only greyscale information. The debayering algorithm knows where the filters are and is 'inventing' color information for the in-between ones. So, any debayering is increasing the amount of information for RGB. Yes, I know, converting to YUV 4:2:2 would reduce it again.

Debayering actually reduces information.

Explain, please. Are you referring to bit depth?

Explain, please. Are you referring to bit depth?

Spatial and color information will be lost also artifacts and additional noise can be introduced.

Here is some fancy argumentation (not written by me):

Resolution Diminution: Inherent to the debayering methodology is the necessity to conjecture the duo of unrecorded color values for each pixel, as the pixel itself records a singular color component among three. This conjectural process, inherently flawed, may culminate in the degradation of image detail and overall resolution. Subtleties such as intricate patterns or textures are susceptible to becoming obscured or manifesting moiré patterns, an outcome of the inaccuracies born from interpolation.

Fidelity of Color: Interpolation algorithms operate under presuppositions regarding the chromatic data of the scene, derived from adjacent pixels. These presuppositions are not invariably precise, engendering the possibility of chromatic distortions or inaccuracies, particularly conspicuous in regions populated with delicate color details or where abrupt color transitions occur.

Generation of Artifacts: The act of debayering is capable of engendering artifacts, including, but not limited to, unnatural colors or the appearance of zippering—jagged perimeters along areas of high contrast. While sophisticated demosaicing algorithms endeavor to mitigate these flaws, their manifestation remains a possibility, particularly within the confines of processing systems of diminished quality.

Augmentation of Noise: The debayering process, by its nature, may exacerbate the noise intrinsic to the original sensor data. This exacerbation is a consequence of the interpolation process, which amalgamates the values of several pixels, thereby potentially proliferating and intensifying noise. This phenomenon is particularly pronounced under conditions of minimal illumination or within sensors characterized by diminutive pixels.

Bit-depth is not a factor unless there are not enough bits to fully capture the dynamic range in linear light. Linear representation is just about the worst way of encoding video, a far better representation is log.

Well, the first point is a good reason to use an OLPF.
The last point is counteracted by some noise reduction BM is quite obviously doing for BRAW recording.
Finally, nobody is recording RAW in linear, even Arri is using log, the only data reduction allowed without Red's consent, it seems. Well most productions with Arri cameras use ProRes anyway. You may even argue, that there is actually no recording of RAW these days. Ever seen RAW from a Thompson Viper?

So, since nobody is giving us detail about their debayering process and the whole pipeline until anything is getting to storage, what to do? You need to compare debayered ProRes, HEVC, and BRAW by testing, testing, and testing again under changing conditions. To get back to the initial question: depending on image content, Q5 can both be pretty small and good image quality. And H.265 can be anything ProRes 444 is, like: all-I, 12-bit, even RGB, but I doubt it would be that much smaller than BRAW.

Well, the first point is a good reason to use an OLPF.
The last point is counteracted by some noise reduction BM is quite obviously doing for BRAW recording.
Finally, nobody is recording RAW in linear, even Arri is using log, the only data reduction allowed without Red's consent, it seems. Well most productions with Arri cameras use ProRes anyway. You may even argue, that there is actually no recording of RAW these days. Ever seen RAW from a Thompson Viper?

So, since nobody is giving us detail about their debayering process and the whole pipeline until anything is getting to storage, what to do? You need to compare debayered ProRes, HEVC, and BRAW by testing, testing, and testing again under changing conditions. To get back to the initial question: depending on image content, Q5 can both be pretty small and good image quality. And H.265 can be anything ProRes 444 is, like: all-I, 12-bit, even RGB, but I doubt it would be that much smaller than BRAW.

I do agree with that. I just believe that experimentation can be done with BRAW by the Blackmagic team to give other options which would work better on crazy high resolutions like their 12K camera, but without reducing quality significantly like 18:1 All-I probably does. I don't see it as necessary, but it could be helpful.

Just some perspective 2:1 to 5:1 (depending on the frame contents) can be lossless, above that will have to be lossy. Above 10:1 artifacts will begin to show. For interframe compression, it is at least double those ratios and often much more.

When BRAW was introduced, the ‘raw’ quality options were either Q0 or Q5. I shot Q0 to maximize the retention of information subject to recording limitations in a shoot. After Q1 and Q3 were introduced, I usually shoot Q1 to ease the media burden and still feel that’s a safe route to visually lossless.

Fortunately this approach seems to have avoided moiré, but that could also be due to material choices of the theatre’s costume department. When shooting a wedding, I guess I’ve just been lucky while shooting with the highest resolution available on the camera. For someone shooting corporate interviews, hopefully you have some influence over materials used.

When BRAW was introduced, the ‘raw’ quality options were either Q0 or Q5. I shot Q0 to maximize the retention of information subject to recording limitations in a shoot. After Q1 and Q3 were introduced, I usually shoot Q1 to ease the media burden and still feel that’s a safe route to visually lossless.

How can camera original be "visually lossless"? It is what it is, a representation, full of seen and unseen artifacts and distortion.

We could refer to those tell-tale traces visible with magnification, at higher compression, as "artifacts", but it would be more accurate to say they're additional artifacts. Enlarge any frame, at any compression level, and you will see plenty of artifacts, in fact that's all you'll see. It's not reality you're looking at, it's artifacts....

And is anyone honestly claiming to be able to distinguish Q1 or 3 from Q5 or 12:1 on the vast, vast majority of material you shoot, without stopping the frame and enlarging it? At what point do these preferences amount to fetish, at least for the event productions described above?

And is anyone honestly claiming to be able to distinguish Q1 or 3 from Q5 or 12:1 on the vast, vast majority of material you shoot, without stopping the frame and enlarging it? At what point do these preferences amount to fetish, at least for the event productions described above?

Unless you're keying hair and such fine detail for compositing, it's going to be very difficult to see a difference, IMO. And Steve Yedlin's resolution tests bear that out, as does the fact that Roger Deakins wanted the 4K Skyfall softened because it was too sharp. And that was uprezzed from 2K.

Unless you plan on printing frames to put in fine art galleries, don't worry about it. Just focus on the art; no one will notice what compression ratio you recorded in at the screening, but they WILL notice if the art direction, lighting, and/or action are bad.

And if it sounds bad, they'll think it looks bad.

So true, the last one can’t be stressed enough.

It's human physiology: poor image quality can actually be absorbing, up to a point, which makes the resolution Olympics all the more silly. The more detail, the less engaging dramatic material tends to be -- at least once a threshold is satisfied.

But everybody recoils instantly at distorted, low res and/or badly recorded sound.

The resolution factor is silly because it's so irrelevant for motion pictures, and because digital sensors have advanced so far that even a cell phone can capture video of high enough qualify for a regular theater.

You hear a lot of newbies and craft-free independents proclaim X is more cinematic!

Yet they can't tell you what it's more cinematic than, nor what makes it so because they have no idea.

In the end "cinematic" means "tells a story" and that's really it.

How much resolution and detail you have in a shot has next to nothing to do with how a viewer will read it... when I learned photography I learned using the f/64 club techniques -- nigh infinite depth of field, manipulating the focus plane to get practically everything in the shot onto the plane of focus and shooting at anywhere between f/32 and f/64.

Learn how viewers' attention will be drawn through the frame, and you can use any amount of sharpness and detail that you want and still make a compelling image.

Use light, shadow, lines, frames within frames, etc.

That is not to say that shallow depth of field is bad, though going overboard so far that the show anonymizes the location is pointless and silly, but rather shallow depth of field is another tool to wield when it fits the moment.

Hello all! I shot my capstone in Q1 BRAW and it only came out to about 850GB. Hopefully that gives me good room to work with in terms of minor VFX doused in heavy grain. Thanks for all your recommendations and have a wonderful day.

Careful with heavy grain if your movie is ever shown over the internet and not only in cinema.

Hello all! I shot my capstone in Q1 BRAW and it only came out to about 850GB. Hopefully that gives me good room to work with in terms of minor VFX doused in heavy grain. Thanks for all your recommendations and have a wonderful day.

I definitely agree with Uli... it would be very easy to overdo the grain and ruin the film. It usually needs to be quite subtle to be effective.

Hello all! I shot my capstone in Q1 BRAW and it only came out to about 850GB. Hopefully that gives me good room to work with in terms of minor VFX doused in heavy grain. Thanks for all your recommendations and have a wonderful day.

I definitely agree with Uli... it would be very easy to overdo the grain and ruin the film. It usually needs to be quite subtle to be effective.

I'm doing it pretty light (Resolve's 35mm 400T preset) but then heavy as a stylistic choice when the villain shows up or is about to.

Careful with heavy grain if your movie is ever shown over the internet and not only in cinema.

Thanks low streaming bitrate

Even 400 T used to be pretty heavy grain in analog times
Definitely try a sample in your final format to see what compression does to your grain, it may get very blocky. Why not define the difference by grading choices?

Even 400 T used to be pretty heavy grain in analog times
Definitely try a sample in your final format to see what compression does to your grain, it may get very blocky. Why not define the difference by grading choices?

I'll probably define it by both. A solution to the compression problem is rendering in 8K, but that would lead to subsampling of the 6K footage and an insane processing time increase. I may submit it to film festivals, but may hold off again since it's probably not worth submitting.

How does rendering in 8K solve the problem of limited bandwidth allocated to web streaming which necessitates high compression? If anything, it only compounds the problem. Lack of bandwidth is the issue, not lack of frame size. HD/2K is plenty of resolution to render pleasing film grain as long as it isn’t compressed to hell.

How does rendering in 8K solve the problem of limited bandwidth allocated to web streaming which necessitates high compression? If anything, it only compounds the problem.

It solves it on youtube because it allocates a higher bitrate to higher resolution files. It's what some YouTubers who still shoot in 1080p do, where they upload 1440p or 4K to not get the heavy 1080p compression.

That is not the case any more if you don't pay them. Just test it!

That is not the case any more if you don't pay them. Just test it!

Ugh that's unfortunate. I suppose that's why so many filmmakers pay for Vimeo to host original files...

If you have enough followers, YT will improve.
But how do you get enough followers with grainy, blocky films?
Go Vimeo.

If you have enough followers, YT will improve.
But how do you get enough followers with grainy, blocky films?
Go Vimeo.

True. The fun part is that YouTube doesn't tell you what follow count to be at for that to happen...

They also didn't tell you when they reduced quality. They only care for the money you make (for them), not your film. You can still get decent quality, but not with visible grain. It ruins the compression.
Maybe one day, there will be compression to recognise the difference between grain and detail. But maybe then your films are made by AI anyway.

Grain is (semi) random, and that means it cannot be compressed which means that too much grain destroys a compressed video.

Personally, I do not understand this urge to add grain "so it looks like film". It's as silly as making an audio recording and adding scratches so "it sounds like a vinyl record".

The only time I would want to add some light grain is if the source is too "cartoonish" so adding grain can add some perceived resolution.

This young guy explains it pretty well (technically, grain is just like snow or confetti):

Personally, I do not understand this urge to add grain "so it looks like film". It's as silly as making an audio recording and adding scratches so "it sounds like a vinyl record".

It's not always about trying to make it look like film; sometimes it's simply about adding texture. When the documentary photographer Sebastião Salgado lost an entire project—60 rolls of Tri-X—to airport scanners, he switched to a digital camera. But in his editing he and his team added the grain he was used to seeing from film, since he felt the images seemed naked without it. I don't think he was trying to make his digital images look like film; instead he felt the images needed that texture to look right to his eyes. It all depends on what you're used to, I guess. But even if you didn't grow up shooting film (or looking at film images), you might like the effect of grain.

Audio is often deliberately degraded for the sake of beauty or texture or variety. Why do electric guitarists use distortion and effects pedals?

Just like Sebastião Salgado won't present his work on cellphones, added grain can look good in cinema, which is using far better compression and higher bitrates.
That's why I asked about the distribution channel.

… Audio is often deliberately degraded for the sake of beauty or texture or variety. Why do electric guitarists use distortion and effects pedals?

Agree with everything you said prior to this statement, Brad. But I must go off on a brief tangent and take exception to your language, specifically use of the pejorative term ‘degraded’ is an unfortunate choice.

Rather we could say the audio can be ‘enhanced’ for the sake of emotive beauty as in this piece of guitar work beginning around 22” in this clip which unaltered wouldn’t suggest any of the grieving felt at the loss of the 29 lives on the Edmund Fitzgerald:

I must go off on a brief tangent and take exception to your language, specifically use of the pejorative term ‘degraded’ is an unfortunate choice.

Rather we could say the audio can be ‘enhanced’...

If you play piano, the correct way to express this is to say that the piano has been “prepared”.

I wanted to try this, but my piano teacher took exception to me preparing her piano.

David Greilsammer demonstrates a prepared piano

How to Sound Like John Cage

Wikipedia: https://en.wikipedia.org/wiki/Prepared_piano