Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Dragon 6k not eligible for UHD "Tier 1" camera?

I bett they've measured sharpness, but called it a resolution. :rofl:

Resolution is just pure pixelcount, right? :biggrin:
 
Graeme Nattress said:
Resolution and sharpness are indeed not the same thing. (snip) Our perception of sharpness is proxied by MTF50, the point at which our contrast has dropped to 50% of which it started with. (snip) Graeme
Click to expand...

To me, this is the crux of the biscuit. My theory as to why the EBU Epic Dragon test showed lower "sharpness" than one might expect, is the interdependent relationship between contrast and measurable line pairs. I further contend that were the R3Ds of the chart processed with an S-curve (or other contrast enhancing settings) in the construction of the RGB image to be measured (not unlike internal DSPs do on many cameras) the EBU would get results more like Graeme's.

IMO, technological change often outpaces testing methodology.

Cheers - #19
 
I don't know...
When I hear sharpness, I'm thinking aha, lens...lines. When I hear resolution, I'm thinking aha, sensor, display...pixels (or subpixels).
I'm also thinking sharpness, resolved by so and so many pixels, thank You Mr. Nyquist! :biggrin:
 
gotcha, thanks for the correction

Graeme Nattress said:
Your math is off.....

A CFA with 8192x4320 (8k) has:

R: 4096x2160
G1: 4096x2160
G2: 4096x2160
B: 4096x2160

I think you divided by 4 but when considering linear dimensions you divide by 2.

Graeme
Click to expand...
 
Measuring a log encoded signal rather than a gamma encoded signal does make a difference Blair. I just did a measure and on the chart I was using it altered the MTF50 % mark by about 6%. The lower contrast of the log encoding did look like lower resolution even though we know the actual resolution would be identical.

Graeme
 
Graeme Nattress said:
Measuring a log encoded signal rather than a gamma encoded signal does make a difference Blair. I just did a measure and on the chart I was using it altered the MTF50 % mark by about 6%. The lower contrast of the log encoding did look like lower resolution even though we know the actual resolution would be identical. Graeme
Click to expand...

Thanks for running that test Graeme. As long as we're digging in a bit, how does one account for the pixel pitch of the display used to judge moiré and aliasing on the zone plate? I get that one can use an instrument to peg MTF50, but what about other variables?

I'm particularly curious about the specifics of any scaling operations employed during the EBU test.

In any case, thanks again, curiosity never sleeps ;-)

Cheers - #19
 
Yes, with any image display you have to ensure it's not actually creating moire. You can usually do that just by doing a zoom in on the image. Obviously, zooming out would be problematic.... When you plot a scan-line you can also see the effects of aliasing in it, but ultimately, moire is going to be a bit subjective. You can also look at a very high resolution zone that you know you can't resolve and see how well aliases in it are suppressed.

Graeme
 
I've had a number of people report aliasing issues over the years, probably a handful of cases. Almost all were actually display issues and there wasn't actually aliasing in the raw data or a full-demosaic viewed 1:1. I've seen at most a couple of real-world examples of actual real aliasing, and it's been on the mild-side - nothing like I've seen on other cameras. In other words we strike the right balance between resolution and aliasing. Of course, you can provoke aliasing via a properly focussed and highly detailed test chart. Sampling theory guarantees that will happen, but you've got to try hard, and again, these charts indicate we get the balance between aliasing and resolution correct. We strive for minimal aliasing. Striving for zero is not a practical approach, but neither is omitting the OLPF and getting some pretty ugly aliasing coming through that needs extensive fixing or a re-shoot.

Graeme
 
As Graeme says, it's not possible to completely kill aliasing and moiré, but it can be minimized. I've almost never seen it in my R3D footage, but from time to time the stars anti-align, and it appears. Check out the engineer's right sleeve and right side of chest. The first of the 8 frames shows minimal, but the next few frames show it rearing up, before again disappearing again:

https://www.dropbox.com/s/0vfkdyqh1png6to/W021_C001_0216SE_001.R3D?dl=0

As I said, its very rare in my R3D experience. I see it much, much less often than the average episode of Downton Abbey (whose marvelous costumes excite aliasing and sometimes moiré in almost every episode).
 
Yeah, that's about the worst I've seen, and it's certainly quite mild. Also, just turning on the D.E.B control in RedCine-X almost entirely eliminates it....

Graeme
 
Thanks for the example. It's impressive to see that only a RCX denoise or a DEB can eliminate this. Wich is good news! If only other camera alisaing could be "switched" off like this!
I wouldn't tell this shirt has a texture, it must be the weawing of the thread that makes the alisaing.

Thanks for sharing.

Pat
 
Phil Holland said:
Again, I have no issue with their Tier definitions. I have issue with a specific test that clearly missed a few things and an issue with the interviewed BBC fellow who is classifying cameras to specific Tiers that they themselves don't adhere to.

All of which has produced weird information which clashes against industry tests elsewhere.
Click to expand...

Yes, well stuff like these seems to emerge every now and then, while the average viewer could care less at these levels of detail. My post wasn't aiming at your comments, but the atmosphere emerging from this topic. Which tends to lead to some of us later defending this community to those who also use generalization as a preferred reasoning process but from a differing perspective.

Seems like some principles are still stuck from the days of SD and early HD. RGB 3x pixel count principle inherited from 3 chip broadcast cameras being one of them. The sense of that approach in 2016 itself is debatable, let alone conflicting measurements.

While we are at questionable content acquisition tool classifications and conditioning of content creators, we can also mention the "4K minimum" itself. Pixel count factor made sense in SD to HD switch due to drastic visual differences but in UHD realm pixel peeping in the context of mass consumption criteria is becoming ridiculous. That being said, the notion of production gear lobbying comes to mind because the notion of content re-sellers and office folks rapidly developing resolution sensibility doesn't exactly shine with reason.

It is a factor and competitive advantage for the tool manufacturers, post production factor and generally specialized professional niches element. Defining overall visual aesthetic as well. Theoretical or realistic pixel count stopped being a visible technical quality limitation after digital had finally reached 35 mm film acuity.

In just a few years we have gone from "their eyes won't allow them to see the difference" to "minimum 4K imager" dictated by the programmer. As if the re-seller is qualified or should get to decide on the very aesthetics of the content. Which acuity affects. In case of Alexa range acuity is the closest to those of a 35 mm analog medium. And someone allowed themselves to decide that is suddenly not good enough today and dictate that criteria to content creators.

In times of most stupefying trash in history of television fed to mass psyche, with 16 fold resolution increase for home delivery practically within decade and a half, some programmer's content criteria includes the number of dots of the imager used.

For ultra detailed images which will be re-processed and compressed by the content provider and again re-processed on 99.99% of UHD TV sets, automatically set and left on "ultra-turbo-extra-hyper" anyway.

Forest for the trees. Or in this case - forest for the bark of trees under microscope. Wonder what 21st. century's pearls of wisdom are next.

I'm awaiting skyscrapers tied by a string and bottle corks on forks to prevent self injury and law suits, with frequent family neurologist counselling after latest 4K 60p HDR action thriller experience.
 
Wow you won RU and the Internet with that one!
 
Phil Holland said:
I actually emailed Alan regarding this test a while ago as this was brought up in the past (his test was published Jan 2015), but my notes were mainly:

- They didn't shoot 6K FF (6144x3160), but rather 6K HD (5568x3132).
- They were critical that the specific zoom lens they used. Alan in email said to me they used "about 70mm prime IIRC".
- If they do consider the Alexa65 and F65 a Tier 1 camera, considering optics as well for that matter, any current RED 6K camera should also apply. Especially Weapon.

Alan did do a rather nice job on his lighting notes. Andy is referencing that specific test paper and honestly optics did come into play and the test should be redone.
Click to expand...

Can you provide links for the two references you mention?
Thx
 
Mike Finegan said:
Can you provide links for the two references you mention?
Thx
Click to expand...

I'm not posting my email conversation between Alan and myself out of respect of privacy. But I will say, this test was structurally flawed and hasn't actually had any effect on what cameras BBC is using as Epic and Weapon Dragon 6K cameras have been used. More recently 8K cameras to that point. This to me is a non-issue despite the paper.

I'll quote this:

"It wasn't a zoom. We found that the zooms vignette rather badly so we used a short prime, about 70mm IIRC."

This isn't in the EBU doc, and that's part of the several things I have issues with. I do testing for other entities and have my own conclusions which has greatly effected their choices. If we were to rate cameras in tiers in pure resolving power and general image quality RED has been leading the way for about 3 years now.

What I will link to is the EBU paper:
https://tech.ebu.ch/docs/tech/tech3335_s16.pdf[

Since my results are rather different (different methods, higher quality optics), I can't really say much from there.
 
Phil -
Understood completely.
Thanks for the EBU link and quick response!

Would be more valuable if they provided same report for other cameras.
All sensors will have some degree of aliasing unless they exceed Nyquist (2x oversampling).
 
You must log in or register to reply here.
Back
Top