Ursa 12k Color Shift when shooting in 8k

I thought I was noticing a white balance shift when switching from 12k to 8k but I did some tests and it seems like it is mostly in the blue channel. It looks like when going from 12k to 8k (and 4k) you see some change in the blue channel like maybe some of the red channel is getting added into it.

Is this consistent with what other Ursa 12k cameras are doing? Maybe it is just a bug in the internal downscaling.

Here are some vectorscope images from Resolve. The red and green are also slightly different but Blue is definitely shifted to red/magenta.

8k Vectorscope

8K.jpg (98.6 KiB) Viewed 1746 times

12k Vectorscope

12.jpg (98.75 KiB) Viewed 1746 times

12k vs 8k Vectorscope

12k vs 8k.jpg (112.06 KiB) Viewed 1746 times

Here are 2 frames if anyone wants to download and check my work.

https://elvisripley.synology.me:5001/sharing/gQxs4W5IT

I noticed this too, back when I got my 12k and did some initial tests. Going from 12k to 8k results in a overall cooler image and it stays the same in 4k. I think the camera boots into a different firmware when exiting 12k readout. For me I never felt like this is a problem as I rarely mix 12k and 8k footage, I usually decide beforehand what I want to shoot for a given project.

I might have to start doing that. I was doing some product shots at normal speed (12k) and then was doing some 120fps 8k so it became an issue.

There is a difference.

You’re using different parts of the sensor because of the way the photosites are arranged.

In 8k and 4k the pattern is the same. You’ll notice rolling shutter time changes between 12k and 8k but not 8k and 4k as well.

It shouldn’t be a huge difference though. If you’re seeing something that you consider unacceptable then you should reach out to BMD with examples and ask them to look at it.

JB

Anybody knows if this color shift also happens with the Cine or Pyxis 12k?

Anybody knows if this color shift also happens with the Cine or Pyxis 12k?

I think this might be an Ursa 12K only problem. Both the Cine and Pyxis 12k use a newer sensor.

Hello I started this post.

It is a weirdness with the Ursa Mini 12k only with mostly Blue. It was just that sensor.

Hello I started this post.

It is a weirdness with the Ursa Mini 12k only with mostly Blue. It was just that sensor.

Have you tried other 12K cameras?

Anybody knows if this color shift also happens with the Cine or Pyxis 12k?

I think this might be an Ursa 12K only problem. Both the Cine and Pyxis 12k use a newer sensor.

The newer sensors are also RGBWWW, no?

Anybody knows if this color shift also happens with the Cine or Pyxis 12k?

I think this might be an Ursa 12K only problem. Both the Cine and Pyxis 12k use a newer sensor.

The newer sensors are also RGBWWW, no?

The URSA 12K is the V1 sensor.

The Cine 12K & Pyxis 12K is the V2 sensor.

The URSA Mini Pro 12K uses a sensor with a 2.2 micron pitch that covers Super35. The Cine 12K and Pyxis 12K use a new sensor with 2.9 micron pitch that covers ‘full frame’ as you know.

I think we are all aware that it's not the exact same sensor (S35 vs FF vs 65), however BM's proprietary RGBW technology is the common thread between them. I guess until someone tests and shows graphs from the Cine 12k we will not know for sure (and the Pyxis 12k is not out yet).

But yeah, sometimes it is a bit painful to match our UMP between 12k and other resolutions, where we need the exact colors of the products we shoot.

What exactly is the difference between the V1 and V2 sensor that prevents this?

Is the V2 sensor not the case, as is with the V1 (UMP12k), that less photosites are sampled with lower resolutions? Thus leading to image quality degradation?

Does the V2 sensor not operate in this way?

My guess for the color shift with the UMP 12K is that it is due to the algorithm used to calculate pixels. The 4K and 8K being similar and differing from that for 12K... and then there are the super-16 modes.

I have and still suggest that a better algorithm can be created... not only for the resolution changes, but for a better quality over all. I suspect that the current algorithm allows for more speed versus quality and yet is not a simple 'binning'.

The camera is still very usable and the sensor better than any previous design (single sensor.) And I would not give up the 240 fps at 4K super-16. I would still like to see a super-16 of this sensor in a 6K pocket.