Blackmagic Forum

I got my hands on both cameras and decided to do a quick, crude test.
please excuse the messy basement! We moved earlier this year.

The Komodo is always on TOP and the 12k is always on BOTTOM.
Komodo specs: 6k 17:9 R3D MQ 800 ISO
12k specs: 8k 17:9 BRAW 8:1 800 ISO (highlight recovery ON)
(it was around the same data rate, to judge 'fairly', but it was more to see how they handle card space)

detail and highlight colored by Adam Langdon, on Flickr

detail and highlight by Adam Langdon, on Flickr

lowlight mixed light by Adam Langdon, on Flickr

ll ml zoomed in by Adam Langdon, on Flickr

zoomed in 2 by Adam Langdon, on Flickr

zoomed in highlights by Adam Langdon, on Flickr


CONCLUSIONS:
the Komodo has a better highlight roll-off, but it's really close and only in extreme scenarios do you see the advantage.
The 12k has way more shadow detail and a 'brighter' image, perhaps 'less stretched' than the RED in Log.
both can match incredibly well!
The noise pattern is pretty close to even, when you turn OFF the Chroma Noise Reduction in the RAW tab for the Komodo.

There is a lot more highlight detail available in the U12K, and numerous ways to recover it, and not just the checkbox. I exposed and graded in extended video in the following video, but I could have made the job of revealing the obscured highlights simpler if I had graded in 'film'. Either way, it is possible to bring highlight details out because they exist within the recording. The other issue I have with your experiment is 8:1 compression will sacrifice detail in broad flat areas. Q5 is much better and Q3 or 3:1 is better still.

Tom Roper wrote:There is a lot more highlight detail available in the U12K, and numerous ways to recover it, and not just the checkbox. I exposed and graded in extended video in the following video, but I could have made the job of revealing the obscured highlights simpler if I had graded in 'film'. Either way, it is possible to bring highlight details out because they exist within the recording. The other issue I have with your experiment is 8:1 compression will sacrifice detail in broad flat areas. Q5 is much better and Q3 or 3:1 is better still.

I'm saying, from the raw images, the Komodo seems to have a more pleasant roll-off. I didn't say the 12k doesn't have a lot of recoverable highlights. There's times when I'm amazed at how much info I can recover from BRAW.

Adam Langdon wrote:I'm saying, from the raw images, the Komodo seems to have a more pleasant roll-off. I didn't say the 12k doesn't have a lot of recoverable highlights. There's times when I'm amazed at how much info I can recover from BRAW.

Pleasant roll-off is a subjective interpretation which is your sole right to own but I am using the clinical interpretation of highlights as data points to be mined and shaped into the image.

I'm not convinced on the concept of canned highlight rolloff benefit, or that it even exists outside of the log transfer. Look at the Log-C curves for Arri. There's no rolloff. Arri even states specifially there is no rolloff except at ISO 2500, yet people marvel at the "rolloff." The fact of the matter is that Arri simply captures lots of highlight data off the sensor, and the perceived rolloff is simply having highlight data to present in the first place. But because 16 bit linear data has to fit in a 12 bit box, there is compression at the top (and bottom) so the issue becomes, do your highlights fall into the narrow band of IRE's where such a rolloff exists? In other words the highlights depend on how the scene was exposed as to whether they fall into the rolled off range. That's why in my opinion, the best rolloff is what you do with them in your grade and the important thing is that they are captured as data points, which the U12K does, not class leading but adequately.

Thanks for sharing this! My G2 is having some issues so I've been looking at replacement / backup options and had been looking at both the 12k and komodo, though obviously they're very different bits of kit.

I see what you're saying about highlights particularly in that lamp shot — the transition between clipped and not seems a bit nicer on the komodo. Did you have the recover highlights thing checked in the raw panel? Also curious if you brought the exposure on the komodo up a little to get shadow detail closer to that of the 12k if it'd lose its highlights advantage.

We *are* agreed that it's impossible to access image quality, including highlight roll-off, from ungraded log? And that log is not a display format, does not aspire to look good and does not represent or indicate the recoverable data in an image?

Certainly solidifies my view that the 12K is my next camera upgrade. But I'll still wait for that next generation 12K in hopefully an updated body with other new features. After all, sometimes it's worth upgrading every other generation so that you're not constantly buying new gear. Can't wait to see what 2023 brings.

Very impressed with the 12K sensor. Not too impressed by the Komodo. But I'm also boycotting RED now due to their latest patent BS.

John Paines wrote:We *are* agreed that it's impossible to access image quality, including highlight roll-off, from ungraded log? And that log is not a display format, does not aspire to look good and does not represent or indicate the recoverable data in an image?

+1