Give me Cinema DNG any day of the week

As the owner of four BMD cameras, I must admit, I am torn between the 6K Pro and the Sigma FP. The Sigma footage definitely has a feel that reminds me a lot of the footage I’ve loved from my OG Pocket. But it’s back to a fixed screen, worse menus, etc.

I do enjoy my BMPCC4K, but I must admit, there’s something I simply love about the FP color science. And inspiration makes for better work…

This is puzzling..... When people invoke the original BMPCC and its predecessor, you have to wonder, what are they doing in post?

I can only speak to the BMPCC and BMPCC 4K, but it's pretty easy to make the BMPCC 4K footage, Prores or braw, indistinguishable from the BMPCC. Which mostly amounts to degrading the BMPCC 4K footage.

What's stopping anyone?

Excellent observations from Ryan. Definitely the colour “Off” seems a better starting point.

After reading Ryan's message I decided to play a bit ... as I have been shooting in Color Mode Neutral
and really think the color I see from the camera is pretty much what is in the scene. I have used BM Design as color space and BM Film as the gamma ... it gives a profound amount of extra highlight range which the
sensor desperately needs.

So the following is a series of 4 clips looking at a ColorChecker Passport Video ... the first two are REC 709 the third and fourth are using BMD/BMF in the RAW tab. 1 and 3 are color mode off 2 and 4 are color mode neutral. Clips transition noted in the description.

For the BMD/BMF I have a CST converting to Rec 709A.

I did a custom WB with a ExpoDisk2 ... set aperture and ISO .... then made no changes for the 4 clips.
In DR Studio 18.6 I matched ... visually brighness and contrast but did no WB or color shifts. The light source was large window directly behind the camera.

I added two interior scenes shot in Color Mode Neutral with minor lessened Green WB in post for the interior scenes.

Captured as 12bit cDNG at 24 fps ... delivered as ProRes 422HQ.

Thoughts Comments Recommendations?

This is puzzling..... When people invoke the original BMPCC and its predecessor, you have to wonder, what are they doing in post?

I can only speak to the BMPCC and BMPCC 4K, but it's pretty easy to make the BMPCC 4K footage, Prores or braw, indistinguishable from the BMPCC. Which mostly amounts to degrading the BMPCC 4K footage.

What's stopping anyone?

Obviously there are large differences between the image coming out of two completely different sensor technologies and color processing.

If you can’t see that, why would you even purchase a Blackmagic camera?

Obviously there are large differences between the image coming out of two completely different sensor technologies and color processing.

If you can’t see that, why would you even purchase a Blackmagic camera?

"Obviously"? And "large differences"? The color is quite similar between the two, and if the BMPCC is the standard, the 4k can easily be matched to it. You would never be able to tell the difference.

Or is it your claim that you can look at any given production, having gone through post, and reveal what the camera was? Absent obvious defects, that would be an unusual talent..

I'd encourage you to approach Steve Yedlin's demonstrations as blind tests. Come back and tell us how many cameras you've identified from his processed test footage, including 35mm film. His audience of accomplished DPs couldn't do it, but maybe you can improve on their averages.

Bob, one can see some colour differences but Neutral is close to Off so perhaps a matter of personal preferences. A RGB wavelength display would perhaps make the difference easier to judge.

Obviously there are large differences between the image coming out of two completely different sensor technologies and color processing.

If you can’t see that, why would you even purchase a Blackmagic camera?

"Obviously"? And "large differences"? The color is quite similar between the two, and if the BMPCC is the standard, the 4k can easily be matched to it. You would never be able to tell the difference.

Or is it your claim that you can look at any given production, having gone through post, and reveal what the camera was? Absent obvious defects, that would be an unusual talent..

I'd encourage you to approach Steve Yedlin's demonstrations as blind tests. Come back and tell us how many cameras you've identified from his processed test footage, including 35mm film. His audience of accomplished DPs couldn't do it, but maybe you can improve on their averages.

Again, if you can’t tell the differences between cameras, why on Earth would you choose Blackmagic cameras? Just buy a Sony and be done with it. All the features you could dream of…

Just throwing in a thought on what is neutral. I do not know of any camera (sensor or film) that will reproduce colors one to one. In fact I always hear a preference to not have colors reproduced exactly as they exist, but rather to have them as the artist prefers (from artists that is.) Color charts might be useful to compare various cameras, lighting and other variables. Photographing a color chart with intent to mimic the colors found in a scene is useless. For one thing, any color will be perceived differently when next to various others. Even old master painters knew of this in selecting their paint colors.

I happen to like Braw and enjoy the results I get from it. Does that mean I do not like DNG or another, or even film? No, those just can provide different results. I prefer Braw.

Again, if you can’t tell the differences between cameras, why on Earth would you choose Blackmagic cameras? Just buy a Sony and be done with it. All the features you could dream of…

And again, there's such a thing as "color grading", and most log-based footage with 10 bits or more and 422 color sampling or better goes through it. This means you can make the BMPCC 4K look like the BMPCC, if that's what you really want. And because the color science is so similar, it's not much of a stretch.

Of course, if you intend to use the camera as a point and shoot, then you may be more troubled by the differences, and look with longing at a 10-year old camera.

What I didn't do in this discussion was introduce another manufacturer -- amateur graders will likely have trouble making a Sony camera look a BMPCC -- but professional productions that employ their higher-end cameras can make of the footage what they want in post. The amateur Sony footage you may see on the web is beside the point.

Yesterday I started grading the DNG stills from my iPhone 14 Pro using ON1. They’re reasonably okay in ON1 as my deliverables, but it’s almost impossible to know what the untouched DNG really looks like because it seems to me that the Apple Photos app and ON1 both apply some AI to the viewed image before I apply my limited intelligence to grade the so-called ‘raw’ image.

It seems the direction that software is taking in that AI knows best how I can get to the quickest result. Thankfully the Auto buttons never need to be pressed so I don’t ruin the photos.

A

What I didn't do in this discussion was introduce another manufacturer -- amateur graders will likely have trouble making a Sony camera look a BMPCC -- but professional productions that employ their higher-end cameras can make of the footage what they want in post. The amateur Sony footage you may see on the web is beside the point.

The sensors for the two cameras are literally designed and made by different companies. The BMPCC OG sensor was made by Fairchild. The BMPCC4K sensor is made by… wait for it…

Sony.

Just because Sony semi makes a sensor for a manufacturer doesn’t mean it’s going to look like a camera that’s made by Sony.

They actually are truly different companies for one thing. And they make custom orders for any customer. It so happens that some of their customers are also Sony divisions that make cameras. You know that stills and motion are different divisions right?

And in the mythology of raw video somehow being pure many assume it can also be made to look like anything.

And there are many many many choices that can be made that will affect the end output of the camera by those that build them.

When you record a sensor output in a high bit depth codec that also means you have a great deal of room to correct the image to look like and match the image of another camera.

I routinely mix Alexa, Alexa 35 (hey they look totally different don’t you know) and Blackmagic 12k / 4.6k and pocket sensors.

Not many can tell.

I can’t even tell after the grade looking at the image the best it will ever be.

The only way I could tell a 4.6k was that it was slightly sharper than an Alexa Mini.

Any good colourist once they’ve had some time, will integrate those cameras easily.

But

There are differences. Often these show up in more extreme lighting, more challenging lighting scenarios.

When you’re starting to see differences in DR and noise floors you start to not be able to match cameras.

With some high gamut lighting instruments you also get different responses that can make them hard to match. Typically none of them are correct to what the eye sees and they often can’t be easily matched to each other either.

As someone very senior with a lot of the design of these platforms said to me, theirs no such thing as colour science. There’s no science to it.

All workflows have choices baked in from the sensor chose, to the options added or enabled, to the chocies about CFA and reproduction to how those files are handlers, encoded and stored.

Some of them can be undone. Some can’t.

I do scratch my head when I see some longing for the “mojo” of older sensors or claiming that they looked better.

I think it’s really that the user has a workflow and set of adjustments, a way of exposing and lighting tk get results that they like.

When you introduce a new part to that workflow the shots go wonky and they then have to tweak or adjust their workflow to find a result they’re happy with.

I’ve been through this with my Leica cameras. With the older M8 and M9 sensors, both the revered Kodak CCD sensors that are supposed to have better colour, I found a sweet spot for exposure (iso) and a way of processing in C1 that has many steps that were almost automatic, such as setting the level of saturation, how I adjust the curves.

Along comes the new CCD cameras and those muscle memory processing steps to get a satisfying image just make the camera look bad. So I have to find new steps. I did things in a different way and started getting results I was happy with.

Sometimes for fun I’ll go back to those old cameras. Sometimes it because I have to send them damn camera back to Leica for yet another repair and it’s weird to suddenly find that I have to re-remember the recipe for getting the most from the older camera.

The only true advantage that DNG offers that BRAW doesn’t in my view is offering an uncompressed options. Image compression really can be different but not in the way I see typically being posted when BRAW Vs DNG threads flare up.

JB

And again, there's such a thing as "color grading", and most log-based footage with 10 bits or more and 422 color sampling or better goes through it. This means you can make the BMPCC 4K look like the BMPCC, if that's what you really want.

As a colorist I can tell you that you really can't. The footage will fight you. It's a push/pull situation and at the end of the day you can only approximate the looks but not match them. Otherwise let's all just shoot on our iPhones and call it a day.

Just because some people can't tell the difference, it doesn't mean they don't exist. Ultimately as a filmmaker you are influenced and inspired by a number of factors. If one of those factors is the way a certain sensor's footage looks, then why not?

I just witnessed a demonstration of how a $30,000 prized prime looks far worse than a $2000 prime, but I know many cinematographers who still gravitate toward the more expensive one because they feel that it has a certain "mojo" that gives their footage the magic they're after.

I say whatever makes them enjoy their work the best is what they should use.

As a colorist I can tell you that you really can't. The footage will fight you. It's a push/pull situation and at the end of the day you can only approximate the looks but not match them.

Approximations are dandy. An exact match is beside the point, because nobody is looking at A/B comparisons, except in camera tests. And even then, good luck trying to pick which is which, or make a case that the incidentals and anomalies of one are aesthetically superior to the other.

If the nostalgia for the BMPCC reflects anything but reduced resolution and noise ("cinematic") and the s16mm form factor, its proponents have failed so far to make the case, in the same way that the case for cDNG reverts to "feelings" without proofs. Repeating "Fairchild" or "dual gain sensor" doesn't quite do it.... Or "raw", for that matter.

Otherwise let's all just shoot on our iPhones and call it a day..

Am not sure how you made that leap. Color grading does tend to level all cameras of comparable abilities, but who's making a case for iphone footage, whose limitations and defects are pretty obvious? Of course, for no-budget filmmaking, the success of the production won't depend on the camera, so if all you've got is an iphone, go ahead and make the masterpiece. And call it a day.

I’ve spent years intercutting my BMPCC and my BMPCC4K, and no matter how heavy the grade is, you can tell they’re different cameras, and not just because of the resolution. So I always end up pushing the grades AWAY from each other, because trying to get them to match is like buying a pair of black dress pants and a separate black jacket. The blacks don’t match, and you end up looking like you raided a coffin for your wardrobe.

They are different cameras. Their images look different. That’s okay.

That's a completely different question, distinct from the supposed unique aesthetic qualities of the BMPCC. Of course a relatively clean 4K camera is not going to intercut with a much less clean HD camera unless everything is brought down to the level of the weakest image on the timeline. It also depends on the grader. Canon 1D and 7D footage was intercut in at least one mass-market feature film I can think of (and there were others) and nobody noticed.

The BMPCC has been available for 10 years, and it's still easily found on ebay. If it's qualities were really that unique and magical, to this day, where are all the BMPCC movies playing? I've got one in the closet, and it was great for its time, as kind of demonstration of what's possible. But the mythology got out of hand.

John just out of curiosity, what is your camera of choice?

Having used the Micro in DNG for years now, I can also testify that the footage out of it is quite different to the 4K, in some ways better, in some ways worse. Specs-wise it is outperformed by the 4K in all respects, but on a subjective aspect I've never felt compelled to upgrade, which is telling for a camera released years before the 4K. Every time I had my Micro image blown up on a semi-big screen with a simple Rec709 LUT I've always had talent tell me just how great the image looked, as if it was from a much more expensive camera.

But that's just my experience. The BMPCC OG vs 4K IQ debate is a horse that's been beat relentlessly to death, one that never seemed to have settled. I suspect that a good portion of the OG maximalists are indeed those that never really properly learned to color grade, but I also fully reject the notion that the image can be made the same, not even similar.

As for why the OG never caught on in the professional world, it was because for all its lovely image quality, it had bad design elements that made it dead on arrival: flimsy HDMI port, lack of active cooling (burns your hand as you hold it), awful battery life, bad rolling shutter, etc. I would never buy one today.

On a more topic-relevant note, I would indeed loved to have Braw on my Micro to use instead of DNG to save a bit of filespace. But there's no chance of that happening, of course.

We shot BRAW 3:1 on the UMPG2 yesterday. 1.5 TB worth of footage. It looks incredible and gorgeous. I don't think there's a massive difference at that point between CinemaDNG and BRAW other than the BRAW file container is simpler for transferring and then you can get the LUT traveling with the BRAW file.

That's an advantage I think only JB has brought up. And, Sidecar files. If we chose a different LUT today and wanted yesterday's files to have that new LUT we could set it up in Resolve for the updated Sidecar files to travel with yesterday's raw footage and it would then have it. So any time that footage is loaded up it sees the Sidecar and sees that new LUT as opposed to what we shot. That right there is a massive bonus to BRAW.

John just out of curiosity, what is your camera of choice?

That's easy. 35mm film, finished on film (with no digital intermediate).

Among higher end digital cameras, which includes cameras as lowly as the BMPCC 4K when used within their range of best performance, I don't think it matters much, besides convenience to the operator or the DP's or the grader's preference. There will not be enormous differences in the final product.

To hear people saying that it's hopeless trying to make the BMPCC 4K look like the BMPCC is a bit flummoxing. My experience says otherwise, but nobody needs to rely on my experience. Footage from cameras far more different than those two are matched all the time, and nobody says peep.

Ahh that’s cheating…okay, it’s not cheating, but I should have asked about your digital film camera of choice

I have the OG BMPCC and Micro Cinema camera, also BMPCC4k and 6k Pro. The OG and Micro in good lighting for me looks organic with better tonality more gradual but the Pocket 4k and 6k the tonal range looks like it has bigger steps, This may be something to do with dual gain sensor in the older cameras.

If the nostalgia for the BMPCC reflects anything but reduced resolution and noise ("cinematic") and the s16mm form factor, its proponents have failed so far to make the case, in the same way that the case for cDNG reverts to "feelings" without proofs. Repeating "Fairchild" or "dual gain sensor" doesn't quite do it.... Or "raw", for that matter.

I wonder why the people who maintain that they can't tell the difference, or that it doesn't matter, or that shoot down other people's opinions while proclaiming some factual imperative truth that they pulled out of thin air, are also the ones who seem to have the least amount of hands on experience? Hmmmm.

Despite it shoot only 1080p JB still chooses the Micro Cinema camera over any of the 4k or 6k Pocket cameras

I have the OG BMPCC and Micro Cinema camera, also BMPCC4k and 6k Pro. The OG and Micro in good lighting for me looks organic with better tonality more gradual but the Pocket 4k and 6k the tonal range looks like it has bigger steps, This may be something to do with dual gain sensor in the older cameras.

Are you shooting the BMPCC4k in Super 16 mode for the best comparison? I don't think the Pocket 6k has a Super 16 crop but the new 6k cinema camera does, as does the Ursa 12K. Some of the unique visual qualities of OG BMPCC and BMMCC are related to the larger depth of field in Super 16, even when shot wide open; it's not only about the sensor differences. The same fullframe lens is going to look quite different at Super 16 vs. MFT vs. Super 35 vs. fullframe.

i'm simply amazed that compression has the ability to remove artifacts.

Hey all, I didn't want to resurrect an old thread, but I read the whole thing and think I have some insight that might be useful to some folks.

I have always appreciated cDNG because of its open format. Sigma FP, DJI, Achtel 9x7, etc. are all amazing pieces of hardware that were probably more possible because cDNG allowed them to provide comparable quality images without having to pay developers to create a new RAW format, or pay Apple or BM to use their existing codecs. Which makes these amazing pieces of hardware less expensive, and more accessible to the smaller folk.

Also just to clear things up for those who are searching these forums to learn; DNG uses TIFF file spec to "hold" RAW sensor data. Sort of like a container. It is not rastered RGB pixels like a traditional TIFF (or JPG, or PNG). It is, in fact, RAW sensor data. As RAW as it gets.

On the topic of cDNG color rendering: it does in fact, contain many available metadata spaces to detail how it demosaics and displays color. It's just up to the software to support those tags. One example of this: a test I did on my Sigma FP. See topic here https://forum.blackmagicdesign.com/viewtopic.php?t=188556&f=21 Detailed in that topic, Resolve honors embedded Color Matrix tags when demosaicing. Adobe does not, and applies its own matrix. (Sigma FP users: check out that link. It can change the way you decide to shoot on your FP.)

Another benefit of cDNG is its ability to take advantage of new demosiacing algorithms as software progresses. You can even change metadata tags to render color differently if needed. BRAW has this all "baked" in. And in my opinion BRAW is more like a traditional rastered TIFF than cDNG is. So, from that angle, cDNG is far more flexible in the long term. Just my opinion. No one is right or wrong here.

Also cDNG can be losslessly compressed to save storage space. Because of the RED patent, this has to be done in post. Unfortunately it cannot be done in-camera because of the patent. - (Maybe now that Nikon has acquired RED, this might change? Who knows.) - There is a fantastic piece of software for compressing cDNG called SlimRAW here: https://www.slimraw.com Lossless compression can compress - often times - more than half the original uncompressed file size. And lossless compression is just that; lossless. It's also easier on your hardware, making for better playback. You can also, downscale cDNG to half its original resolution if needed too.

Another up-and-coming DNG software is being developed here: https://github.com/dnglab/dnglab is useful for lossless compression. Although still in development. Check it out if interested, and contribute however you can.

In my opinion cDNG is most like shooting on film stock. With film stock, each frame is captured individually, uncompressed, and archival.

Also, not shaming BRAW at all. The codec is amazing at providing color consistency, accuracy, and detail. All with RAW-like control, at a fraction of the storage space and processing power. In fact I shoot BRAW everyday on our production cameras. And if BM makes a FF box camera, then I suppose be shooting BRAW LOL. - It's all down to each person's opinion. Each codec is surely capable of providing more than enough quality for intermediate needs. We are lucky in 2024 to have all these tools. Back in 2005, at the rise of digital cinema, that's a whole other story. Props to the pros who traversed those waters, and paved the way for us younger folks.

Devon, I understand every your good point and I’m agree with you, but…
A good open standard should be standard and have a single sdk to implement in other software.

A good codec should not be interpreted but decoded, every interpretation give you room to error and problems.

Every raw codec is intraframe, shooting in cdng or in raw single file give you the same capture and same result. The problem of motion picture born only in gop codec (which cannot be raw).

The only single advantage of cdng is that if you have a problem to a single frame, you can easily duplicate a frame near file and all ok, on single long file shooting is more dramatic to have a data damage on file.

Anyway, I ever be happy to have ability to record in more format than a single codec, but I want consistency, in the last 27 years I fight with too much codecs problems of custom version of xx coded to be happy a too free flavour of a codec


Inviato dal mio iPhone utilizzando Tapatalk

With Rawlite OLPF installed I would like to have Cinema DNG option.

When I was shooting a concert on the weekend with the UM4.6K, I was seeing very strong horrible moiré patterns on the camera monitor from one of the large grey boxes on the floor near the musicians. The box enables them to monitor their audio levels, sorry I don’t know the proper name. I thought catching moiré was going to be an ugly first for me, but there was nothing I could besides editing the boxes out of the frame in post do to avoid it.

Pleasantly surprised in post, there was zero moiré on the boxes in the recordings. I was recording CinemaDNG lossless!

Edit
Incorrect upon viewing the rendered video in 4K.

Pleasantly surprised in post, there was zero moiré on the boxes in the recordings. I was recording CinemaDNG lossless!

But that moire was obviously an anomaly of the monitoring.... How does that lead to the conclusion that cDNG or lossless somehow eliminates moire?

Given the tributes to the BMPCC OG above, the irony is that false detail and moire is far worse shooting cDNG than Prores on that camera..

Apologies! After viewing the rendered ST2084 video on my 32” monitor, I need to make a correction: the grey on-stage monitors sometimes do show moiré; not as intense and doesn’t ruin the shots as I had feared watching the camera monitor. Sorry about that previous post.

With full respects to the people saying that for CDNG and BRAW, "it's all good," the potential for the topic going off the rails existed because the thread starter baited people with an inflammatory title for the topic, which makes you wonder if he was deciding between that or "I don't need no stinkin' BRAW."

In consideration of all the reasons given hereto, it would have been more accurate to say, "Give me Cinema DNG for those certain times."

There's nothing wrong with options but at 30min per TB@6K it's just not very interesting outside of a few select cases.

Good Luck