Blackmagic Forum

Ellory Yu wrote: As I said, you can have as much Ks as you want but it boils down to what gets projected on the screen - a 2K DCP package unnoticeable to the viewership, or a 4K UHD at best streamed via Netflix or

I've never bought this argument.

I've been admiring the screensaver on Apple TV.

Many of those were shot using three stitched 8k cameras, and it's clear that something special is going on with those screen savers, even watching them in "just" 1920.

You know that a lot of TV drama was shot on 35mm even when it was NTSC and it was only the "soaps" that shot video...?

There are advantages to supersampling on the acquisition side even for lower resolution delivery.

JB

Que Thompson wrote:I don't think it will be a disaster because it's a proprietary sensor.

According to JB, they've been working on it the pass 3 years. They will probably use it in any new camera they make.

Although that may mean no new M43 cameras... But who knows, maybe this thing that output the M43 size as well.

JB also said they shot for 14 hours and used just over 1TB at 12K.

Stop thinking about the "camera". The camera is just a shell around the SENSOR.

Also, being. new sensor, the initial yield is low and the costs are high. It makes perfect sense to place the natural limitation of a lower volume high end camera like the 12K.

I presume and hope this is the beginning of a new sensor family, and other version will appear. But the initial cost of each sensor is high right now. Once that comes down then you'll start to see it in other cameras.

If they did this sensor as a Pocket camera with a pocket price you'd be on a waiting list for 2 years.....there would be a stampede that they could never meet. It makes a lot of sense to build this production line over time...

It also doesn't stop them from developing other cameras in that time ?

JB

Ryan Earl wrote:Are the details available for the BRAW compression settings on the (3) sample clips? Could you point me to them if they've already been discussed? I am getting smooth 24 fps playback on a 2015 27" iMac in the timeline and color tab. (4K timeline, not 12k)

Again, clips look stunning in 5K on the iMac.

METADATA in the edit page and under camera you can choose "camera" and it will tell you down the bottom.

Oh. wait....

Hit CMD I (get info) on the file in the finder and it's right there...blew my mind when Hook told me that shortcut...

Also listed is resolution and frame rate and WB. all in the INFO of the file ion the finder...(this particular clip was Q0)

Love BRAW...

JB

I found this presentation on misconceptions about RGBW color filters enlightening:
http://image-sensors-world.blogspot.com ... ilter.html

Grand tip, thanks John. Did you have fun shooting with the new 12K camera?
Cheers

Denny Smith wrote:Grand tip, thanks John. Did you have fun shooting with the new 12K camera?
Cheers

Hell yes...more so when I got to grade and edit...

I just got my final version of the hardware, so looking to really go out and flog it now...

JB

Que Thompson wrote:Good job not answering the question.

What, this?

Que Thompson wrote:Edit:
If it was Sony... They would have made you pay for it through firmware updates. 6K yesterday... 8K next year for $2500... and 12K in 2024 for another $5,000. You want 110fps at 8K? That's another $1,000.

You buy what you need for the time it is needed and get a 10X return on investment before and each time you buy the next model. Otherwise, you're either a asinine businessperson or a hobbyist. Then the increase in value/performance becomes irrelevant. That should answer your question enough to make you think first.

John Brawley wrote:I've never bought this argument.

I've been admiring the screensaver on Apple TV.

Many of those were shot using three stitched 8k cameras, and it's clear that something special is going on with those screen savers, even watching them in "just" 1920.

Not expecting everyone to buy into it. Apple TV is a unique use case for now. Until then, it's not the norm.

John Brawley wrote:
You know that a lot of TV drama was shot on 35mm even when it was NTSC and it was only the "soaps" that shot video...?

Yes I know… and your point was?

John Brawley wrote:
There are advantages to supersampling on the acquisition side even for lower resolution delivery.
JB

With all due respect, if this was your point, I agree there are advantages especially then. There's a point of diminishing return in what is technology capable and what the human eye sees. I rather err on the human eyes first. As I said, its toy vs. practicality. If I have a good story and shot it today on an OG BMPCC and Netflix picked it up, does it really matter that it was shot on an OG BMPCC? Do I really have to shoot it in 12K and deliver it in 4K and that will make the story visually better? Maybe it is a small percentage of viewers which does not even matter when you think in terms of volumes. That's my perspective and I subscribe to it. I think we can agree to disagree here and it is okay.

Ellory Yu wrote:

John Brawley wrote:I've never bought this argument.

I've been admiring the screensaver on Apple TV.

Many of those were shot using three stitched 8k cameras, and it's clear that something special is going on with those screen savers, even watching them in "just" 1920.

Not expecting everyone to buy into it. Apple TV is a unique use case for now. Until then, it's not the norm.

John Brawley wrote:
You know that a lot of TV drama was shot on 35mm even when it was NTSC and it was only the "soaps" that shot video...?

Yes I know… and your point was?

John Brawley wrote:
There are advantages to supersampling on the acquisition side even for lower resolution delivery.
JB

With all due respect, if this was your point, I agree there are advantages especially then. There's a point of diminishing return in what is technology capable and what the human eye sees..

I just gave you an example that probably everyone today is seeing.

I think those screensavers are beautiful.

And I learned later they were created using super high resolution shots.

This demonstrates a counter point, that's easily verifiable to everyone out there reading this thread. That's all.

JB

He just wants to argue because BM made a camera that he thinks nobody needs.

All he's looking at is "12K", then shutting down. That's it. It's exhausting reading many of these posts and posts on the web. All they say is "Why do I need 12K?" It's such a closed minded view.

He still didn't answer if he would have been happier if they released it as 6K and gradually up'd the Ks in new cameras using the same sensor.

I bet people would have said things like... "The sensor in the 8K renders the highlights much better than the 6K sensor..." Just talking...

BUT BM JUST GAVE IT ALL TO YOU. 12K!!!!!

This is version 1. What's the max DR this sensor can achieve? What improvements can be made to this sensor other than resolution? How will this affect our BM ecosystem/workflow in the future? These questions are more interesting to me than... "What am I going to do with 12K?" "Who delivers anything in 12K?"

Don't you think a COMPANY knows more than a customer/user in regards to what is going to be profitable? You act like BM is a bunch of people in a garage making stuff and kicking it under the door when it's done.

Users wake up and look at something for 10 seconds and instantly know everything about it's future. All most of these people understand is Ks, DR, FPS... And by "understand", I mean the concept of... Some have a more technical background and can digest information beyond that... Yet, most want to just give their surface level opinion with absolute minimal thought. It's so annoying. As creators, you should understand this. If I looked at your work and told you it sucked within 10 seconds, you'd have a serious problem with me. "That sucks, it won't sell much." Meanwhile, the same people are complaining about features in a $1200 camera that nearly rivals the INDUSTRY STANDARD (more than 50X the price). Wanting to use it as a &%@* WEBCAM!!

I can't say what I really want to say.... So I won't.

John Brawley wrote:... If they did this sensor as a Pocket camera with a pocket price you'd be on a waiting list for 2 years.....there would be a stampede that they could never meet. It makes a lot of sense to build this production line over time...
JB

You likely realize that putting the 12K technology into the BMPCC4K, you create a BMPCC8K camera. 8192 pixels x 2.2 microns = 18.0224mm; less than the current 18.96mm for the 4K sensor.

I think you’re right about the waiting list. I’m putting in an order for the BMPCC8K now! Good thing I already have two PL->mFT adapters.

I'm surprised by the responses here, though I shouldn't be, I've been here since the OG Pocket. People love to nitpick.

My initial thoughts were, "Wow!" Is this camera for me? Nope, beyond my price range for my purposes. Am I super-excited about this camera? Yep! I'm not reading specs so much as listening to John talk about what the images look like. I take what he's saying seriously because what's exciting to me is not specs, but the actual image. I think it's awesome that BMD worked on their own sensor. Part of this is the possibility of a future that has some of the magic of those initial sensors. I think this is a great step forward that shows that BMD are doing serious R&D continuing to make their own mark. I'm excited about future trickle down to other product lines. The nature of BRAW and the efficiency with editing shows that they're looking at the entirety of their users workflow. Even if I was just focused on a singular product I would say that the camera is a great deal for its price point. But I'm more focused on what this signals for the future. Great job Blackmagic!

I think the Oscar for managing to keep a secret for this long will go to Blackmagic. Our rumor mill here missed that one. Though we're ahead of the curve on the 16K MicroPocket with 18 stops of dynamic range and full resolution output to the VideoAssist that will be launching next month.

Ellory Yu wrote:With all due respect, if this was your point, I agree there are advantages especially then. There's a point of diminishing return in what is technology capable and what the human eye sees. I rather err on the human eyes first.

Even though there are diminishing returns on the spatial resolution that the human eye can see, that only applies to the end product. There's whole bunch of advantages to shooting at higher resolutions than the end product that are beyond spatial resolution.

For example, noise is something that would be recognized in the end product. A 12K source will have much finer noise and it's much more likely for that noise to be of a higher frequency than the highest frequency parts of the image. So applying noise reduction to 12K source before downsampling to 4K will result in a cleaner image than if you were grabbing an noise profile from the 4K source.

Moire has far less to do with spatial resolution of the final image and far more to do with spatial resolution of the recorded image. Down-sampling in camera or in post does a lot of to eliminate moire and avoiding de-mosaicing will at least get rid of color moire.

Then there's the VFX world where higher resolutions mean much higher render time and memory usage but they also mean better tracking, better keys, more isolation. A 12K image down-sampled to 4K is an excellent way to get more of that useful data into a shot without increasing render time and memory usage. It also allows the source footage to share the same sub-pixel accuracy that the computer generated assets have which will help them integrate with each other.

Ellory Yu wrote:As I said, its toy vs. practicality. If I have a good story and shot it today on an OG BMPCC and Netflix picked it up, does it really matter that it was shot on an OG BMPCC? Do I really have to shoot it in 12K and deliver it in 4K and that will make the story visually better? Maybe it is a small percentage of viewers which does not even matter when you think in terms of volumes. That's my perspective and I subscribe to it. I think we can agree to disagree here and it is okay.

This is a tech discussion, not a film-making one. You can tell a good story with pretty much any camera. Hell, 28 Days Later was shot on a XL1 which had horrible resolution and three 1/3" sensors. That doesn't make this camera useless in comparison.

This isn't about the viewer much like the lens, makeup, dolly, NLE, or anything like that is for the viewer. There's a bunch of things that go into the production of films and TV shows that viewers don't know about and if you told them about them, they probably wouldn't care. This is about the process and what things make that process easier for us.

Btw, I have the OG BMPCC. It can shoot beautiful images but those images can also be noisy and it gets noisier as it heats up. Its also very, very prone to moire and aliasing to the point that people bought low-pass filters (that cost nearly as much as they payed for the camera) to try to fix that. It's also so light that when I had lay it on the floor and tilt it up with my, my heartbeat was causing my hands to shake the camera. All these problems that seem little in the context of a conversation on a forum, realistically result in a lot of wasted time, money, or both to try to compensate.

So yea, the viewer may never know that you ran all your shots through NeatVideo, that you had to take breaks while filming to let the camera cool down, that you needed to swap between 3 batteries for a 3 hour shoot, or that you had to put the camera on a wonky rig to support a larger external battery but it does all matter to you and the people you're working with.

So think about it the other. The viewer won't know what went into creating each shot in the end so why not use the tools that let you get that shot in 5 minutes instead of 20 minutes?

Mark Grgurev wrote:For example, noise is something that would be recognized in the end product. A 12K source will have much finer noise and it's much more likely for that noise to be of a higher frequency than the highest frequency parts of the image. So applying noise reduction to 12K source before downsampling to 4K will result in a cleaner image than if you were grabbing an noise profile from the 4K source.

Playing with some of John's test footage, this advantage became very clear. I blew up one of the shots 6X on a 4K timeline and the in focus image detail was several pixels wide compared to 1 pixel noise. The noise was easily removed without even coming close to impacting the actual image data. When there is a 1-1 ratio of pixel to real image detail, this just isn't possible. Downscaled 6x, the noise was effectively removed for free.

Guys can we start the conversation about color resolution, effects of high high resolution and small pixels on fine or complicated gradations, and lens selection yet?

=<

Just kidding. Continue.

Kholi Hicks wrote:Guys can we start the conversation about color resolution, effects of high high resolution and small pixels on fine or complicated gradations, and lens selection yet?

=<

Just kidding. Continue.

I bet if you start it, they’ll come.

No one knows the right questions to ask. It’s just been speculation based on older technology.

Que Thompson wrote:

Kholi Hicks wrote:Guys can we start the conversation about color resolution, effects of high high resolution and small pixels on fine or complicated gradations, and lens selection yet?

=<

Just kidding. Continue.

I bet if you start it, they’ll come.

No one knows the right questions to ask. It’s just been speculation based on older technology.

Didn't I start something in a thread about 12k and 8k falling below human colour resolution and evening things out, already?

joe12south wrote:

Mark Grgurev wrote:For example, noise is something that would be recognized in the end product. A 12K source will have much finer noise and it's much more likely for that noise to be of a higher frequency than the highest frequency parts of the image. So applying noise reduction to 12K source before downsampling to 4K will result in a cleaner image than if you were grabbing an noise profile from the 4K source.

Playing with some of John's test footage, this advantage became very clear. I blew up one of the shots 6X on a 4K timeline and the in focus image detail was several pixels wide compared to 1 pixel noise. The noise was easily removed without even coming close to impacting the actual image data. When there is a 1-1 ratio of pixel to real image detail, this just isn't possible. Downscaled 6x, the noise was effectively removed for free.

Not this again, so you are saying that the 12k camera only produces effective detail resolution below 2k. I thought Braw sorted that out!

What's the real situation of those details Joseph, are they smooth or lumpy?

Robert Niessner wrote:I found this presentation on misconceptions about RGBW color filters enlightend:
http://image-sensors-world.blogspot.com ... ilter.html

Thanks Robert for quoting that fir people.

Now, on dynamic range.

I've seen an article that described all sorts of dynamic range extending techniques, but don't have a link unfortunately, so I'll explain. Somebody made an attempt on it somewhere recently as well.

At full 12k it might be 14 stops, but if they used a very high end way of defining stops as being with little to no noise, to match the camera market, then those stops could still be equivalent to the G2, and if I looked at a test chart, I would probably find details at 16 or 17 stops in noise (which if you look at Dmitry's threads you'll see my recommendations for how to clean noise and extend dynamic range. Which I think predates the same advice from a BM engineer). Silicon these dad's us getting 17 or so stops naturally, but that is only useful for electronic news gathering unless you can clean noise over frames track real detail to determine what is noise, which is my ancient recommendation from the 1990's maybe, but I don't say for IP reasons.

So, when you clean noise you probably will get extra stops of ussable image from that 17 stops natural silicon capacity at 12k. But when you down res, you even out noise a bit, maybe getting extra ussable stops, maybe even more so, if you do the clean properly described above. But there are limits where there is so much noise you would just get error in value down scaled. Lets just say that the two techniques get 1-2 stops more.

Now, back when this whole 8k is possible thing kicked off (and guys, did the camera overheat or couldn't process the image above 6k, or cost like an Red, no, not even at $10k custom instead of 8k pocket based on Sony chip for $2k. Maybe I should work out a tutorial one day on how to make pocket 4k footage to look more like the 12k or G2, or Alexa) I mentioned how downscaling to 4k from 8k, might give you an extra bit or two pixel depth. This pixel depth is also potentially ussable with stops (but not fully in my fore mind yet). See, you have 12 bits, which aren't btoo much in HDR processing terms, but when you stack 4 pixels together on a down scale you times the light value potentially collected by 4, that's 2 bits worth, and your lowest value is just one of those pixels being lit, and highest, all of them being fully lit, but just one 12 bit pixel fully lit and the rest dark, is a quarter of the value. But this is more about value resolution, then dynamic range, as wherever one or all, it is still measuring the same peak brightness in the dynamic range, and the sane bottom brightness. But surely there is a way to squeeze some DR out of that, I'm not seeing.

Now, we come to highlight recovery, where each different type of color can be used to reconstruct other clipped colours, to restore pixel values. Every coloured filter tile is like a different ND filter, but just for one colour range, along with a clear tile. That's 4 different levels of filtering over varying strengths that don't evenly line up. So, up to 2 bits. Highlight recovery on this chip might be very interesting given the clear tiles.

Is Sony the new sensor foundry, as there is no dual gain anymore? If so, they have intra hdr technology that is useful, that doesn't require dual exposure.

So, covert the footage to 16 bit+ and play. See how much more real world dynamic range you an squeeze out of it. Of course 16 stops to start with would get you that without much effort, or a lot more range on top of that with effort.

Suggestion: BM should do a guide with the camera and in resolve, on how to squeeze the maximum dynamic range from the cameras, with auto options to use these things?

Hey JB,

Could you do some tests with the 4K and 6K full sensor recordings or link us to some tests? I am very interested in using this camera for 4K/6K acquisition if we are artifact free (no moire) at those resolutions. That would sell me on it!

Thank you!

Kholi Hicks wrote:Guys can we start the conversation about color resolution, effects of high high resolution and small pixels on fine or complicated gradations, and lens selection yet?

=<

Just kidding. Continue.

I don’t know enough of the science to take on most of those topics, but I am very interested in Lens selection for the new sensor and the tiny photosites. Kholi if you wanted to kick off a thread with that topic alone, I’m sure we’d like to here your experience so far. There are a lot of lenses that claim to be 8K ready, but I’ve never seen anyone talk about what 12K in 2.2 micron pitch means for lenses.

Since I’ve already ordered the BMPCC8K that will also have this pitch, over a smaller sensor, I’m hoping my SLR Magic APO and Tokina CinemaATX lenses survive the cut.

There might be a message when you see the lenses featured in the promotional materials for the UMP12K.

Blackmagic's pricing for this in their home country is interesting. In the USA it's $9,995, but here in Australia the price is AU$19,745! Now if the Australian Dollar was only worth around 50 US Cents that would be reasonable, but it's actually worth around 70 US Cents at the moment which would equate to just over AU$14,200. So what's the other AU$5,500 for exactly Blackmagic? Even Apple don't push it this far.

Comments anyone?

Steve

Wow.

Kim Janson wrote:But do you necessary want the sharpest lens. It sounds like the 12k sensor also has a colour filter, so small amount of blur might be a good thing, and there is also: CA, Contrast, Flare,... and how accurate one can focus.

Thinking about focus, if they would have put the clear photosites just tiny amount different level, maybe using the colour filters, this would have been awsome for continuous autofocus. Does the Ursa Mini Pro 12k have any usable autofocus?

What I really would want to see (well actually develope, but I do not have the resources) is external focusing devices working together with in camera focusing. The external focusing device could be sonar, time of flight, 3D sensor, manual focus, and I have some other ideas too. I think this kind of resolutions will need better focusing than normal manual focusing can provide, to make meaningfull use of the resolution.

Check out the NVIDIA development boards, and of you can get a phone sensor array module that can hook up to it. Zcam Probably still uses this, and a lot of people do AI cabe a stuff on the boards. The old one had six sensor channels to do realtime drone flight AI. It was the size of a credit card. And more aggregate data input than would be needed to do 8k. They haven't done much that small lately, but have an even smaller embedded board now, much cheaper maybe? Looking up what others are doing, their code and shops with camera kit for it, maybe a start.

This is literally grey import jungle territory. They might be banking on the Aus dollar crashing and the US serging, but maybe it will be the reverse. Uncertain times require price adjustments on the fly.

Both prices are off the website so should include taxes. In Australia there is 10% GST and I guess possibly 10% Import duty, but I can't see that it would make that much difference. It was mentioned on these Forums recently that the BMPCC 4K had gone up from around AU$1800 to AU$2500 here in Aus as well so perhaps it's across the board? It would be cheaper to fly to the USA and buy one and come back, even with the import duty!
Crazy times indeed.

Steve

Steve Golding wrote:Both prices are off the website so should include taxes. In Australia there is 10% GST and I guess possibly 10% Import duty, but I can't see that it would make that much difference. It was mentioned on these Forums recently that the BMPCC 4K had gone up from around AU$1800 to AU$2500 here in Aus as well so perhaps it's across the board? It would be cheaper to fly to the USA and buy one and come back, even with the import duty!
Crazy times indeed.

Steve

USD prices are quoted without tax because different states charge different taxes.

Even if you added both taxes to the $14,000 price, you would still be nearly $3,000 shy.

Steve

Kim Janson wrote:About the focusing.

What I meant is a interface with the in camera automatic fokusing

It poroviding cordinates of the areas that are in focus

displaying on viefinder what the user, via the external focusing device wants to focus

etc.

I.e. the in camera automatic focusing and external focusing working together. The external part we can do no problem, but there is not the camera interface to communicate with the in camera automatic focusing system and that we would need.

Ask for a developer function to send a bitmap of focused points, maybe even a 4-8 bit value to define how focused a point/infront-behind.

Canadian prices are $14,695 CAN. That is retail so there’s possibly 12% taxes plus a small recycling tax and shipping to add to that. Maybe $16,500. I pay in USD though.

Ellory Yu wrote:You're missing the point. You're eating Grant's dog food and you're gaga over it. I'm not. I'm looking at the prism of what makes good business sense for me irrelevant of the vendor - this camera is not, at least not until it can be visibly seen there is consumer demand for images that needs the Ks. Doesn't matter if, for example, you can shoot 12K to get a superb 4K because of a new sensor. A superb 4K image is just a 4K image. Right now, that's so irrelevant. As I said, you can have as much Ks as you want but it boils down to what gets projected on the screen - a 2K DCP package unnoticeable to the viewership, or a 4K UHD at best streamed via Netflix or YouTube to your TV set that, for 90% of the viewing public, is 1080p right now.
I rather see the engineering effort put on functionality that improved usability and productivity of the UMP 4.6K G2, stable products, well speaking for myself. New toys are just toys to me until there is payback - unless you're a hobbyist.

Yes, it does all boils down to what gets on the screen but that doesn't mean that it makes sense to shoot low-bit rate h.264 for release on YouTube Plus. Is RAW irrelevant? Is 12-bit irrelevant? No. This is that but with resolution.

I've noticed that you haven't responded to any of my other responses to you. Is it maybe because you don't know what you're talking about? Is it maybe because the whole 4 years that you're been making stuff hasn't made you an expert on anything?

Ellory Yu wrote:I rather see the engineering effort put on functionality that improved usability and productivity of the UMP 4.6K G2, stable products, well speaking for myself. New toys are just toys to me until there is payback - unless you're a hobbyist.

I can't think of anything worse that Blackmagic could do then to intentionally be at or behind the curve. The UMP 4.6K G2 may be a decent improvement over the G1 but it's using a tweaked version of the sensor used the Ursa Mini 4.6K that was introduced 5 years ago but with support for higher frame rates.

People have been asking for a new 8K camera though and that's what Blackmagic is giving them in the form of a 12K camera that can shoot 12K, 8K, or 4K using the whole sensor while still having excellent readout speed. Using a higher resolution sensor to shoot a better low resolution image is nothing new but blackmagic is doing it to a larger extent and still letting you shoot with the full sensor.

Ellory Yu wrote:BTW, I have nothing against a new 12K camera.

You sure have a strange way of showing it. You've basically written the equivalent of a chapter in a book saying it's a useless, dead-in-the-water camera that nobody should buy and is just Blackmagic trying to show off despite that fact that Blackmagic is clearly justified in the choices they made with this camera. You said as much within the post I'm currently responding to.

Ellory Yu wrote:I just wished that BMD does better in supporting older gen cameras while putting time on their new ones. Today, their customers do not have their new products [yet], they however need enhancements to their existing investment. They have to do both and not expect people to jump to a new ship each time. Plan obsolescent is futile to their image if this keeps on going - and that is true for any manufacturer. Think about this - the Pocket 6K is less than a year out yet it feels like its obsolete - true for the 4K and the UMP G2 as well. And what about those who still are using their G1? Yes, camera evolve and new products are necessary. Quality, support, and service is however paramount to succeed.

The UMP G2 isn't being discontinued because of this camera. It's still going to get updates in the future but even if it doesn't, you already have a great camera. Just use it.

The Pocket 6K may be just a year old but it's 1/5th the price of the UMP 12K. They're not competing and this does not make the 6K obsolete.

So, we can all agree now that an 8k camera would be a great idea

Mobile phones are going 16k 1 inch+:shortly, and you could probably get those sensors for less than $20 (even $10) in bulk, and maybe even get a wide 16k video feed out of it. That's like a toy is getting 16k. I'm waiting to see the new micro, but GH6 and maybe another Sigma, are coming up. If the present Sigma had descent full res internal raw compression, it would have been nice at the size.

Mark Grgurev wrote:I've noticed that you haven't responded to any of my other responses to you. Is it maybe because you don't know what you're talking about? Is it maybe because the whole 4 years that you're been making stuff hasn't made you an expert on anything?

Well, I'm flattered that you've been following my post for the past 4 years. That's a dedicated follower mind you. If that hasn't made me an expert, then that makes you stupid for following the last 4 years. I'm just going to respond to you this once just to let you know that the reason I ignore you (and people like you) is because there's nothing to say to someone whose head is buried in the ground... and that's a very polite way to describe it, and best way to insult someone's lack of intelligence.


You, taken with BMD UMP 12K BRAW Q5

Is there any more footage out there to look at?
Any low-light?
Any bright sunny day footage?

Yes, I want more than still shots of people with their heads in the ground.?

What 'Camera Conspiracies' thinks about this 12k camera

"Blackmagic URSA mini Pro shoots 12k video now. So much for the Canon R5 8k, that's so yesterday. Is this the greatest cinema camera ever made? It sounds like the 12k files aren't too hard on the computer, but is it good enough for vlogging and youtube?"

They do seem to be the only people on the planet still earning an outrageous income during the pandemic and can pickup a UMP12K to make beautiful pictures or just to make fun of it. More power to them though as it all helps in the long run.

There are people here way more tech-savy and into the engineering side than I am, so I figured I'd ask this question. Since we're talking about oversampling, what would yield better results.... shooting 12K and downsampling in an 8K or 4K resolve timeline or downsampling in camera (in the sensor) and shooting 8K/4K? My understanding is that doing it in the software is better and I know it also depends on how good the camera does it in realtime, but in general, what would be better?

I'm also aware that there might not be a definitive answer for this yet since the camera is brand new. I'm still trying to wrap my head completely around the new tech.

PaulDelVecchio wrote:There are people here way more tech-savy and into the engineering side than I am, so I figured I'd ask this question. Since we're talking about oversampling, what would yield better results.... shooting 12K and downsampling in an 8K or 4K resolve timeline or downsampling in camera (in the sensor) and shooting 8K/4K? My understanding is that doing it in the software is better and I know it also depends on how good the camera does it in realtime, but in general, what would be better?

Not sure how different the results would be but I can speculate. Doing it in-camera does the resizing before compression so doing it in camera might result in a slightly softer detail at lower bit-rates but still a noticeable improvement over having a native sensor of that resolution. Resizing in post should avoid that and you'd have the ability to choose the down-sampling algorithm you use.

Ellory Yu wrote:Well, I'm flattered that you've been following my post for the past 4 years.

All I did was search your name on Google. Found your website with the blurry, noisy shot of you editing in Resolve and followed the link to your IMDB that shows you're earliest stuff being in 2016. I assumed you at least knew about search engines if you have a website. Guess I was wrong.

Ellory Yu wrote:I'm just going to respond to you this once just to let you know that the reason I ignore you (and people like you) is because there's nothing to say to someone whose head is buried in the ground

Such a beautiful way of saying "Actually my temper tantrums are just because I'm smarter than you."

PaulDelVecchio wrote:There are people here way more tech-savy and into the engineering side than I am, so I figured I'd ask this question. Since we're talking about oversampling, what would yield better results.... shooting 12K and downsampling in an 8K or 4K resolve timeline or downsampling in camera (in the sensor) and shooting 8K/4K? My understanding is that doing it in the software is better and I know it also depends on how good the camera does it in realtime, but in general, what would be better?

I'm also aware that there might not be a definitive answer for this yet since the camera is brand new. I'm still trying to wrap my head completely around the new tech.

I don’t know for sure... I’d think it would be like upscaling HD to 4K. Do you let your streaming device upscale, or your tv or your Blu-ray player? Who does it best? Or, do you upscale in Resolve?

It all comes down to software and in this case BM would control both ends (in camera & davinci). I’d guess the hardware (embedded software) would do it better, but I don’t know.

With the BMPCC 4K, the difference isn't subtle, at least not if you're looking for it. Shooting either braw or Prores full-sensor UHD and downsampling on an HD timeline in Resolve provides noticeably more detail than shooting full sensor Prores HD, with the camera doing the downsampling in real time. (BTW, Prores HQ is more detailed than the higher compression varieties of braw, like Q5 or 8:1.)

Resolve also gives you control over how the downsampling occurs (sharper, smoother, etc), which is not available in-camera.

Good point, Mark. Our answers of course would be based on our prior experience where it was generally understood you could get better results using the resources of Resolve. However we’ve been told that BRAW 2 was designed for this new sensor, moreso than current sensors, to take advantage of the symmetrical array, including the amazing ability to record 12K or 8K or 4K from full sensor readouts, as it produces an output data stream, it will likely mean it is very good doing the work in camera. BMD is now clear that noise reduction is applied in camera and may be further enhanced at 8K and 4K. If I owned the machine, I’d likely output 4K widescreen from 12K readouts. Up to 120 fps is fine. Just trying to keep RAID storage manageable as I am somewhat limited there.

Honestly though the choice is yours given these 12K streams play so easily in Resolve without optimized media. If you owned the camera, you could answer this question after a day of shooting and a day of grading. It’s good to test and either way won’t challenge your resources.

Whether it’s 12K or 8K or 4K, it’s still 12-bit raw that isn’t going to break.

rick.lang wrote:BMD is now clear that noise reduction is applied in camera and may be further enhanced at 8K and 4K.

Did they say that? I didn't catch that.

It’s on their website and perhaps it’s near the top of the first page on the new camera.

Ah it says "Noise management, sensor profiling and new edge reconstruction algorithms are used to produce high quality cinematic images with incredible depth, crisp details and beautiful image separation."

But I don't think that's anything new. I remember hearing about that when BRAW was announced and I got the impression I had to do more with preventing noise that's introduced during de-mosaicking and maybe trying to cancel out fixed-pattern noise. I'm not sure that there all that much of it going on but I could be wrong.

I think they’ve perfected what they can do in BRAW 2 with the 12K symmetrical pixels from the new CFA. So the words might be similar to what they did for the original BRAW Q0, the execution has changed dramatically (and subtly). The in camera 12K to 8K and 4K also helps greatly, but the sample 12K output we see is very nice. Just expect 8K and 4K to look even better.

Mark Grgurev wrote:

Ellory Yu wrote:Well, I'm flattered that you've been following my post for the past 4 years.

All I did was search your name on Google. Found your website with the blurry, noisy shot of you editing in Resolve and followed the link to your IMDB that shows you're earliest stuff being in 2016. I assumed you at least knew about search engines if you have a website. Guess I was wrong.

Ellory Yu wrote:I'm just going to respond to you this once just to let you know that the reason I ignore you (and people like you) is because there's nothing to say to someone whose head is buried in the ground

Such a beautiful way of saying "Actually my temper tantrums are just because I'm smarter than you."

Well, that's good that you're smarter than I am. Ask me if I care. And that... IMDB. For the record, those projects posted on IMDB where just as of 2016 by the project owners and being that you are smarter than me, you know that those credits are automatically generated when the production posted their films. I've been making pictures since 1998, way before IMDB existed. But given that you're just basing what's on IMDB in the last 4 years, let's just say that in that span of years I have edited 2 well paid recent feature films and one short, directed by yours truly, that got a silver award from an international festival. That's more than the 5 home grown shorts you made since 2012. Take that to the bank! And by the way, all my IMDB entries are credits from the production companies who made them, not self posted. Yours were likely self posted and self credited, which I'm sure you're good at because your smarter. Come back and brag when you get your first feature notch in your belt. Best of luck pumpkin.

Mark Grgurev wrote:Ah it says "Noise management, sensor profiling and new edge reconstruction algorithms are used to produce high quality cinematic images with incredible depth, crisp details and beautiful image separation."

But I don't think that's anything new. I remember hearing about that when BRAW was announced and I got the impression I had to do more with preventing noise that's introduced during de-mosaicking and maybe trying to cancel out fixed-pattern noise. I'm not sure that there all that much of it going on but I could be wrong.

Ahh, so they are taking advice from my playbook, good. I was suspicious of that pattern and mentioned need for reconstruction, as I have mentioned for many many years. So, that should be getting a grade higher. Maybe that's how they are going higher compression ratios. But have they adopted the latest edition of my universal staggered image preservation scheme yet, mentioned here last year or three?


We figure out, what others make.