New Blackmagic Camera July 16th?
- Thread starter Teddy_Dem
- Start date
Joshua Cadmium
Well-known member
That is so great to hear. The sensor appears to me to be more of a generational leap than anything I've seen out there, so the fact that it was not a repurposing of existing tech means it is not going to be bogged down by other extraneous features designed for other fields.
Yeah, that Canon sensor has the same 2.2um pixel pitch, but it lists a max framerate (albeit at full 120mp resolution) of only 9.4 fps. Also, it looks nothing like what it shows in the 12K promo shots.
I was fascinated by Grant Petty’s mention that the new sensor has equal amounts of red, green and blue photosites. How is this possible? Well, if you arrange the colors in a 3 x 3 square you can achieve just that. Here’s what it looks like.
Each 3 x 3 square contains 9 photosites with equal amounts of red, green and blue.
Now the question is, how do we make a picture from this arrangement? And here it gets interesting. There are multiple ways to do it, but the most obvious is to take a 3 x 3 square and create one image pixel from it. In other words, 9 photosites would make one finished pixel with equal amounts of red, green and blue in it. In a sense, this is the holy grail of color imaging – full color resolution in every pixel.
So, if it takes 9 photosites to make one pixel, the next question is how many total pixels are we left with? Obviously, it will be 1/9 th the amount of total photosites.
Total photosites = 79,626,240 (12,288 x 6480)
79,626,240 divided by 9 = 8,847,360 final pixels
And guess what size image has 8,847,360 pixels in it? 4K at 4096 x 2160 !
These numbers are right off the specs sheet for the Ursa Mini Pro 12K.
Now we can see why Blackmagic went to a 12K sensor. It was to produce the ultimate 4K image with full RGB information in every pixel!
Each 3 x 3 square contains 9 photosites with equal amounts of red, green and blue.
Now the question is, how do we make a picture from this arrangement? And here it gets interesting. There are multiple ways to do it, but the most obvious is to take a 3 x 3 square and create one image pixel from it. In other words, 9 photosites would make one finished pixel with equal amounts of red, green and blue in it. In a sense, this is the holy grail of color imaging – full color resolution in every pixel.
So, if it takes 9 photosites to make one pixel, the next question is how many total pixels are we left with? Obviously, it will be 1/9 th the amount of total photosites.
Total photosites = 79,626,240 (12,288 x 6480)
79,626,240 divided by 9 = 8,847,360 final pixels
And guess what size image has 8,847,360 pixels in it? 4K at 4096 x 2160 !
These numbers are right off the specs sheet for the Ursa Mini Pro 12K.
Now we can see why Blackmagic went to a 12K sensor. It was to produce the ultimate 4K image with full RGB information in every pixel!
puredrifting
Major Contributor
If the camera can only record BM RAW and not ProRes, that's not a good move for them. Few pro editing shops are on Resolve. And even few Premiere users even know about the BM RAW Plug In. FCP will never support it. AVID? Who knows?
puredrifting
Major Contributor
You haven't been here very long, have you?
ahalpert
Major Contributor
Finally some sensor details:
"This is how Phil Holland explains it, ‘With Blackmagic’s new sensor instead of having the Bayer 2×2 grid (GRBG) it has a 6×6 grid. There are 6G, 6B, and 6R, plus 18W pixels. The W are clear or “white” pixels. This overcomes the reduced sensitivity issue of having a very small 2.2 micron pitch.’
The problem with having a very small 2.2 micron pitch sensor is that low light performance could very well be compromised, however, the high resolution may help to ease some of those worries. The white pixels are just clear and are there to deal with the light loss of using such small photosites with tiny microlenses. As a comparison the RED Monstro is 5 micron pixel, the RED Komodo is a 4.4 micron pixel."
"This is how Phil Holland explains it, ‘With Blackmagic’s new sensor instead of having the Bayer 2×2 grid (GRBG) it has a 6×6 grid. There are 6G, 6B, and 6R, plus 18W pixels. The W are clear or “white” pixels. This overcomes the reduced sensitivity issue of having a very small 2.2 micron pitch.’
The problem with having a very small 2.2 micron pitch sensor is that low light performance could very well be compromised, however, the high resolution may help to ease some of those worries. The white pixels are just clear and are there to deal with the light loss of using such small photosites with tiny microlenses. As a comparison the RED Monstro is 5 micron pixel, the RED Komodo is a 4.4 micron pixel."
John Brawley
Well-known member
You're kind of on the right path. You're missing one part though.
There's a grid, but it's 6x6 pixels.
It's an equal number of red, green and blue pixels, 6G, 6B and 6R. So that leaves 18 unaccounted pixels. Those pixels are there but there's no CFA on them. They're clear, or monochrome, or white pixels.
So it's RGBW in a 6x6 grid.
JB
HKO
Well-known member
he he he. .
John Brawley
Well-known member
And Alexa has 8 micron pixels which is why they haven't been able to make a 4K s35 camera.
He's a bit right, but what's happening with the W pixels is that they are combined like a dual gain sensor, with the white being high sensitivity and the coloured pixels combined being the low sensitivity, so you get a kind of HDR process.
JB
John Brawley
Well-known member
I think you can.
You might not be able to render it at more than HD or do noise reduction...
JB
Last edited:
Many years ago Kodak experimented with an RGBW sensor. It was a simple 2x2 array with one photosite for each color + monochrome. They ultimately abandoned it because they couldn't get the colors saturated enough. I wonder what's different now?
ahalpert
Major Contributor
Much higher resolution might help?
I thought Brawley's test that he posted looked a little desaturated in places although the color quality and variation on the skin tones was exceptional. But I chalked the saturation up to grading choices.
I mean it's true that we're dealing with a 50% monochrome sensor. But there are still 40 million chroma pixels scattered across it
CaptainHook
Member
Hey Joshua, long time - hope you're well!
The 4K S16 crop is another "special" mode (like the 8/4K full FOV modes) that is actually the same field of view as the 6K S16 crop - but with the higher 220fps read out.
That's why they both have the "Super 16 crop" description - they are the same FOV.
Mitch Gross
Veteran
Open them up and put them side by side.
Mitch Gross
Veteran
That’s like asking the price of a Bugatti. If you have to ask it’s too much.
Open them up and put them side by side.
Time to Vimeo Milvus lens tests? Would love some info! Saw a post on FB about CP.3s being very similar to Supremes as well.
Mitch Gross
Veteran
It’s almost as if 4 * 3 = 12 !!
But that’s not what’s happening here. There are 2K of red, 2K of green, 2K of blue, and then 6K of clear (no color filter). The clear photo sites are much more sensitive to light because of the lack of chemical color dye sitting on top of them. So each color photo site is paired with a clear photo site, one providing color info and the other providing sensitivity info. Fairly brilliant and only really works well when you get into the high resolutions such as this.