Blackmagic Forum

Howard Roll wrote:Map the 12K image 1:1 on a 100" diagonal monitor and it becomes a retina screen at 25" or greater.

If your eye was a s35mm sensor you'd need a <6mm lens to have a wide enough FOV to resolve the full image.

The reality is that our eye can only resolve about 3 or 4K and that's probably being generous.

To see the threshhold of diffraction limiting on the 12K you need 2-4X zoom, that factor will change the perceptual threshhold. Meaning if you can only see 3K the effects of diffraction won't become perceptually noticeable until ~F8.

Fun Fact: The UM12k has more color receptors than the human eye.

Fun Fact #2: Every 2/3" 4K Broadcast camera has the ~same pixel pitch as the 12k and has since the dawn of 4K.

Good Luck

Depends on FOV, you need something you can't resolve, to not see pixels, like I do on my 4k TV. I didn't think it would be that bad in any situation, but there are some times where I do, when well enough. 24k for the full field of view, but above 70 degree just gets more and more useless in standard study telling.

I advocate 4k for standard delivery, 8k for special projects, maybe even 16k fur a wedding, as obsessed people are going to walk up to a video wall to check things out one day (they can pay extra extra for that). Few older people can see as well as me on mega nutrients and a $1000 glasses I get really well scripted. And maybe not even half can see 4k. Lots more younger people though. Anyway, for those who can't see 8k or 4k, 2k is probably ok.

rick.lang wrote:For those who questioned the value of having a 12K sensor, that was so yesterday. Seems Canon will release a 19K sensor this month. Presumably for stills though as maximum frame rate may be only 5 fps:

https://ymcinema.com/2020/10/20/canon-introduces-a-new-19k-250-mp-aps-h-cmos-sensor/

Hmm, pocket canwra?

Re-edit:

Rick,

https://ymcinema.com/2020/08/28/canon-a ... ops-of-dr/

A lot of people want 2k, this is only 1/1.8 inches. 20 stops and can do low light. At around s16, it could do 4k and ve excellent for a micro like product. But I'm still interested in that little camera phone I posted, but they took that down, pretty interesting little thing, just bought a 128GB card for it.

URSA 12K 60fps HDR | Test Footage

Canon now use dual gain sensors (similar to Arri sensor and Fairchild sensors in early BMD cameras) so dynamic range and quality in deep shadow noise will increase from generation to generation.

Same time BMD moved to 4K 6K and 12K resolution, small pixels, high compression, blurry but fast BRAW codec with some amount of moire, not too pretty noise structure, no factory OLPF filter...

I personally prefer pixel perfect moire-free HD (or 4K max) straight out of camera than 8K or 12K that produce huge file sizes, require noise reduction post processing, extremely fast computer, additional OLPF, and may produce surprises in deep shadows due codec compression and small pixels.

Dmitry Shijan wrote:I personally prefer pixel perfect moire-free HD (or 4K max) straight out of camera

I can't get my Ursa 12k to produce any significant moire even on test charts designed to do exactly that. From what I've seen, moire only comes from a poor post workflow. Not out of the camera. Also.. a native 4k camera doesn't resolve at 4k. If you have a 4k display, you'll get a better image oversampled.

You may say I'm a bit biased since I own the 12k. But it's just a tool for a job. I just think it's important to be fair and not spread bad info.

My guess is it more like very small amount of moire mixed with shadow chroma noise generated on contrast edges and more-less amplified depending of BRAW compression, scaling and project settings. Too many variables with this sensor and 12K workflow options...
There is some noise problems discussed in other thread. Some 8K and 12K clips shared shared Eugene Belsky and Michael Hoffman here
https://drive.google.com/drive/folders/ ... ffJCESixzM
https://drive.google.com/drive/folders/ ... 1_56ASPwcN

Noise is almost gone when debayer quality set to 2K. In 4K and higher it became visible at 1:1 zoom ("Center Crop with no resize" project setting) as well as in downscaled to HD project.
At HD resolution image is perfect and mind blowing. Color science gen5 is great, colors and dynamic range looks nice in one click, image looks crisp and there is no any visible noise.
At 4K image is a little bit sharper and noise is became visible.






On Michael Hoffman's examples a lot of noise visible at the dark purple-magenta cloth. That color patch on ColorChecker usually the noisest with any other BM camera as well.



So i think this is more like perfect HD/4K camera. I guess same image quality with dramatically less file sizes and processing power could be produced with HD/4K dual gain sensors with large pixel size. But participation in megapixel race instead of quality of pixels is probably some marketing strategy...

Here is you can see how moire and color noise strength changes depending of debayer resolution. Project setting scaling set to "Center Crop with no resize".
8K seems like optical resolution limit of the lens, because there is no visible difference between 8K and 12K

Camera 7.2 came out today as a software update and has quite a lot of image related improvements in it.

Is colour science gen 5 comming to Pocket 4k? I like to be able to monitor using gen 5.

I have a 12k rented and doing some test shots. I'm a little concerned about this these color dots appearing on this car. Is this aliasing? They show up in both 12k and 8k, extended video and film. This is a 4096 timeline. I'm really happy with the resolution but not with this artifact.

Scott Pultz wrote:I have a 12k rented and doing some test shots. I'm a little concerned about this these color dots appearing on this car. Is this aliasing? They show up in both 12k and 8k, extended video and film. This is a 4096 timeline. I'm really happy with the resolution but not with this artifact.

dots.png

I’m guessing a prismatic effect from sun being transmitted through the drops of water ?

Aliasing tends to be more when you have super fine detail that can’t be resolved accurately.

JB

Is it rendered or are you just watching it in Resolve? With Performance Mode activated the resolution might be decreased and a faster, low quality resizing algorithm might be used during preview. Best way to be sure is to render to the resolution of your monitor and watch the result.

Seems extremely high resolution will not help to 100% fix problems with moire. And seems in reality higher resolution also may produce a lot of moire and aliasing but just at different details level. And all those rainbow moire patterns will be there after image resize. Here is example from known online chart camera test https://www.dpreview.com/reviews/image- ... 8554630656

roger.magnusson wrote:Is it rendered or are you just watching it in Resolve? With Performance Mode activated the resolution might be decreased and a faster, low quality resizing algorithm might be used during preview. Best way to be sure is to render to the resolution of your monitor and watch the result.

It's in a still that I exported as PNG from resolve. I haven't had time to do a rendering yet.