sony

[Jeff Kilgroe]

OK, I haven't read every post in this thread, nor have I pulled out a calculator to do any serious thinking. But... It seems to me that if you are trying to quantify enough chroma resolution to adequately represent a 4K image on an S35-sized sensor by using a Q67 approach -- rotate the Bayer pattern 45°. It seems to me that the actual photosite density MUST BE INCREASED by close to 50% or more. So it seems that their higher resolution is mostly accounted for by this alone.

As for easier processing, I see no evidence of that here. Unless they intend to start binning pixels as RGB clusters to be equivalent to one compound or derived RGB pixel space that would be aligned to the normalized imager plane. Binning pixels almost seems like a logical assumption here based on this being Sony's broadcast division and the talk of full 4K resolution output in real time, blah, blah, etc.. The Q67 approach also suggests the possibility of foregoing the use of an OLPF and allowing the demosaic / binning algorithms compensate for aliasing and ringing. I don't know if I'm anywhere close to the mark with that, but so far every way I've looked at this, I fail to see any advantage.

 

[Michael Olsen]

As for easier processing, I see no evidence of that here.

I may have missed this...

I'm certainly no sensor engineer, but I am curious. Why would rotating pixels 45 degrees make processing easier or quicker? Compared to handling standard rows and columns, it actually seems like it would be a bit more complex...unless of course it just processed along a 45 degree as well...in which case it's still doing the same old thing? :001_unsure:

 

[Jeff Kilgroe]

You're not missing a thing, Michael. But some people in this thread seem to think that the Q67 approach will offer easier processing. I think they're assuming that the camera will do all demosaic processing onboard and pass off a 4K RGB file of some sort. Thus eliminating the need for complex debayer processing in post or edit. Either way, I don't see the advantage there... Or at least no advantage that wouldn't include a huge compromise.

 

[JD Holloway]

...I'm also expecting a real-time 4k REDRAY based (compressed) broadcast solution real soon...\

That would be real-time demosaic and REDRAY compression.

Perhaps tied to RED based 4K monitoring.

 

[Eric Lange]

...and inversely proportional to the number of Guinness your 1AC had last night.

..

...

Global shutter is a good thing though.

That’s funny (the Guinness that is).

Yup, I was pretty blown away by the idea of a GLOBAL SHUTTER myself.

This was thought to be a pipe dream or virtually impossible?

SONY seems to be going on a lot about these graphics processors, originally used in the PS3?

These cameras are going to be ridiculously expensive, but for S-3d a true global shutter and synchronized image will be freaking amazing! (If it really is Globally Shuttered rather than just fast).

Jim once said that global shuttering is not going to happen in anyone’s life time (for cinematic res), so I have to admit I’m very intrigued. But beyond that I remember he commented that a global shutter has a “sterile” appearance more like an industrial camera, and has a different feel from a “scanning shutter” from a RED ONE or a film camera.

The math and diagrams of the presentation were pretty dodgy/misleading. They didn’t have to do that, very odd presentation, its tone almost seemed defensive rather than positive and forward looking given the things they are presenting. It really seems in the tone of the presentation that RED IS a concern to them and seem to imply that they (SONY) are more capable than anyone else at providing an entire eco-system for 4k delivery; the implication being why would you go anywhere else?

In other words I found the tone to be one of trying to steer people away from seemingly other (moronic technologies and systems) and that they have the right balance of things, they have the “sensible” “goldilocks” solution of complete acquisition to broadcast and storage solutions. Everybody else is just a moron because they don’t have SONY’s, and I quote, “crazy” algorithm.

If they just said, this is what we are doing, and isn’t it wonderful, rather than really trying to steer people away from other things with slightly misleading data, then a better impression would have been left.

They seem a little worried…and rather desperate that folks in the US get the right end of the stick (as far as they see it), rather than people in “HollyWood” trying to make their own cameras. I.e. the statement that was made was rather patronizing in that somehow “Hollwood” (whoever that really is?) is not really capable of creating a real camera and work flow, and that they should leave it to the “professionals” i.e. SONY. It was made pretty clear that they will make every effort to muscle in on every American Standards committee they can!

Clearly they are worried that those backward hicks…, those Americans (who don’t have proper 4k unlike the Japanese and European sophisticates) are going to push the wrong button and not buy SONY. They know once people learn a different company’s workflow, it’s much harder to make them switch; RED is there. If people are already learning to use RED/EPIC, there is no time to drop everything and switch to SONY systems unless required to do so. This is what they are afraid of, and this does hurt their bottom line (in the US at least).

This was not a feature of projected technical capabilities and workflow, but was really the beginnings of a pissing match to stakeout territory, and obviously influence decision making of potential partners and buyers through cobbled together half facts. This looses credibility, they didn’t need to do that.

They should not openly show such fear… (that was pretty un-cool in my opinion)

 

[Ian Laurie]

Thats IT! I can't take it anymore!

Shreddies are still shreddies wether diamond or square! sorry. as a canuck i just couldn't resist. Honestly though, other than the increased pixel count shown in the presentation, i don't see how it could possibly change anything to put the pixels diagonally. He says this gives you better diagonal definition, but by that same theory it would lower the vertical and horizontal definition. so whats the point here. Am I crazy graham or is this the biggest waste of marketing BS since the diamond shreddie?

 

[Graeme Nattress]

Thats IT! I can't take it anymore!

Shreddies are still shreddies wether diamond or square! sorry. as a canuck i just couldn't resist. Honestly though, other than the increased pixel count shown in the presentation, i don't see how it could possibly change anything to put the pixels diagonally. He says this gives you better diagonal definition, but by that same theory it would lower the vertical and horizontal definition. so whats the point here. Am I crazy graham or is this the biggest waste of marketing BS since the diamond shreddie?

You may say that but couldn't possibly comment.

Graeme

 

[Deanan]

And while you can point to (insert unnamed German camera manufacturer here) as possibly doing this right now, you could also look at Red about 1 or 2 years ago talking about 4K being "all you need" and then quietly working on 5K and beyond.

Not true. We announced 5k as soon as we decided to work on it.
There's was nothing quiet about it.

We also did it to increase the measured resolution to 4k which
we had early on stated 4k bayer pattern measured less than 4k.

 

[Ian Laurie]

You may say that but couldn't possibly comment.

Graeme

OK, so if I dropped the marketing BS line, would my theory about edge anti aliasing hold true? I can't see any reason why it wouldn't. I defer to your knowledge. Also, sorry about my last posts horrible spelling, including your name, I was obviously not paying much attention after having my mind warped by trying to figure out how 45 degrees was beneficial in any way.

 

[Graeme Nattress]

OK, so if I dropped the marketing BS line, would my theory about edge anti aliasing hold true? I can't see any reason why it wouldn't. I defer to your knowledge. Also, sorry about my last posts horrible spelling, including your name, I was obviously not paying much attention after having my mind warped by trying to figure out how 45 degrees was beneficial in any way.

I don't think you're crazy.

Graeme

 

[Ian Laurie]

Good to know. Thanks

 

[Stephen Strangways]

The thing with the diamond Shreddies is that it was a joke. A fun marketing gimmick. They won an advertising award for it, and it translated into a statistically significant sales increase.

Unfortunately, I don't think Sony is just out to have a laugh with Q67. It seems more like FUD.

 

[Eric Lange]

The thing with the diamond Shreddies is that it was a joke. A fun marketing gimmick. They won an advertising award for it, and it translated into a statistically significant sales increase.

Unfortunately, I don't think Sony is just out to have a laugh with Q67. It seems more like FUD.

I’ve watched that SONY presentation about five times now, and sketched out a bunch of stuff to understand this diagonal array, fully expecting that there has to be some advantage in terms of color gamut and accuracy of color signal for high frequency features. But for the life of me I just don’t see it.

For most of the presentation, (just like what Graeme is saying) they are comparing the theoretical performance of two imaging systems where one is almost twice the resolution of the other!

I really liked the slide that shows the girls on the pick-nick blanket. On the left it says BAYER and on the right image its says "Q67". The CTO points out all the moiré patterns on the pick nick basket are from the [“despicable”] BAYER pattern, and then points to the comparable “Q67” image on the right. However, if you look at the small print on the slide it says, “camera: Sinar/P2 4x5, film KODAK E100s, Scanner:ICG/330”. This is all simulated BS. No actual physical examples.

Graeme is absolutely right, if you look at the pattern, it is identical to a BAYER pattern turned 45 degrees. However they try to infer that this grants higher color accuracy and larger gamut. This really is BS. It’s a bit like saying if you take your RED ONE and rotate it about its optical axis by 45 degrees, and then carry out a bi-cubic re-sampling and rotate the resultant image a further 45 degrees that suddenly you have 100% green area response and 50% red and 50% blue i.e. 67% of a true RGB system. This IS BS! Geometrically at whatever scale you want to group or regroup your pixels this is total rubbish if you are comparing the same number of included photosites/unit area. What they are basically saying is higher res is better than low res (give that man a cigar!). The greater colour Gamut can only come from better SNR/(less noise) from using CCD vs CMOS.

I can only think of two reasons to arrange photo sites diagonally; here I speculate,

1.

The physical architecture of The CCD is arranged so that the transmission registers read the photo sites across these 45 degree diagonal lines. This would be faster than having to read scan lines across the whole width of the sensor. I.e. the bucket brigade for transferring charges tavels a much shorter distance than a conventional CCD of a simlar size and numper of "pixels"; it may be that signal integrity is preserved better as a result and hence better SNR, better dynamic range, marginally bigger color gamut etc.etc. Given that they have high frame rates and a global shutter means that the architecture is very different from any other CCD.

2.

The second advantage to a diagonally layed out set of photo sites is that aliasing artifacts from rectilinear structures in the real world would be lessened. (not a major consideration really).

I think to be honest that the real benefits of the diagonal arrangement were not explained at all, and everything else was just to blow smoke. I can only assume they do not want to reveal what is actually going on.

STEVEN I think it is diamond shaped “Shreddies”, but they are not explaining why Shredies turned 45 degrees and wired up in a completely different way is a good idea. Perhaps trawl through the patent literature to find an explanation…


Ta.

Eric

 

[Michael Hastings]

The second advantage to a diagonally layed out set of photo sites is that aliasing artifacts from rectilinear structures in the real world would be lessened. (not a major consideration really).

I think to be honest that the real benefits of the diagonal arrangement were not explained at all, and everything else was just to blow smoke. I can only assume they do not want to reveal what is actually going on.

Eric

why is reduction of real world aliasing artifacts not a major consideration? - seems to be the big bugaboo argument against current DSLRs - and real world is mostly what we care about..

And isn't it somewhat normal (think HDRx) in this industry to hold back some of the particulars? For legitimate competitive or timing reasons?

 

[Andrew M.]

Eric, don’t get excited. They have to differentiate themselves somehow and when technology is pushed to the limits like this sometimes it is difficult to differentiate.

Guess what, I wouldn’t mind to have two boxes of Shreddies on the shelf in the kitchen.
Imagine, you wake up and you look on the shelf with two boxes of Shreddies side by side one square and one diamond.

Which one would you pick if you eat squares Shreddies for past 5 days?

 

[Graeme Nattress]

We use circular zone plates to look at resolution and aliasing in all orientations.

Reducing aliasing is always a concern, but the DSLR aliasing is not because their sensor is wrong or their OLPF wrong, but because they don't read out the data from the sensor correctly (line skip) or they poorly filter before downsampling or use other "shortcuts" in the production of their image.

As mentioned above in argument, from what I can gather about how Q67 is used, it will lead to worse overall luma aliasing than using the same number of pixels in a traditional Bayer arrangement.

Graeme

 

[Eric Lange]

why is reduction of real world aliasing artifacts not a major consideration? - seems to be the big bugaboo argument against current DSLRs - and real world is mostly what we care about..

And isn't it somewhat normal (think HDRx™) in this industry to hold back some of the particulars? For legitimate competitive or timing reasons?

It’s funny you should say that almost 20 years ago I was invited to the SONY HD centre in Los Angeles where HD was being tried out. John Galt (very nice guy) now with Panavision (as I understand) was showing me round. So they had HD footage of a big government (neo-classical) building with fifty steps leading up to the door. As the camera tilted up and down you could see pretty serious aliasing artifacts, as you could see the pixles literally “trickling” over the steps flashing on and off. I pointed this out, and they claimed that they absolutely could not see what I was talking about, we replayed the footage five times, pointed to stuff strobing on screen; I thought this was very funny…

So maybe people do care about this after all.

 

[Michael Hastings]

We use circular zone plates to look at resolution and aliasing in all orientations.

Reducing aliasing is always a concern, but the DSLR aliasing is not because their sensor is wrong or their OLPF wrong, but because they don't read out the data from the sensor correctly (line skip) or they poorly filter before downsampling or use other "shortcuts" in the production of their image.

As mentioned above in argument, from what I can gather about how Q67 is used, it will lead to worse overall luma aliasing than using the same number of pixels in a traditional Bayer arrangement.

Graeme

I understand the reasons DSLR aliasing is different, just criticizing that - whatever the cause - we have been railing against it and now someone is saying "no big deal" when Sony "says" they are addressing it.

Re luma aliasing: since you didn't mention it, do we infer that it might be better with chroma aliasing? And if so, is there a balancing point between the two that might be advantageous?

For example, for the most part DV and dvcpro often looked better than Betacam because, within the confines of SD, the additional sharpness was more noticeable than the slightly better color purity possible with beta cam over DV. But in the brave new world of super high res, maybe there is a balancing act between the two that optimizes for more/most situations.

 

[David Mullen ASC]

My layman's non-technical perspective would say that pointed at a circular zone plate, a Bayer-sensor turned 45 degrees shouldn't show any change in aliasing, but in real-world shooting, there may be a slight advantage since a number of grid patterns encountered (like shooting TV sets) tend to be vertical and horizontal, less often diagonal. But I'd want to test that theory. And it wouldn't help with a grid pattern in a shirt, for example, where the collar and sleeves cause the pattern to turn diagonally.

 

[Eric Lange]

why is reduction of real world aliasing artifacts not a major consideration? - seems to be the big bugaboo argument against current DSLRs - and real world is mostly what we care about..

And isn't it somewhat normal (think HDRx™) in this industry to hold back some of the particulars? For legitimate competitive or timing reasons?

It’s a question of integrity. I hold SONY in very high esteem in that they try as much as they can for large corporation to conduct themselves in an honorable way. They do in fact offer very fair arrangements when working with smaller companies, which is rare in this day and age.

It’s really a question of personal integrity and how that relates to corporate integrity. The presenter and CTO should NOT have presented the things he did in the way he did. In the past I have been forced to give presentations and for reasons of content and integrity once or twice I have had to point blank refuse, and argue very strongly for what should be presented and what should not. The CTO should not under any circumstances be presenting these types of mis-leading half truths. If they just said we believe that the capabilities are going to be X,Y and Z rather than presenting very skewed facts figures and imagery to bogusly dress this all up. There is a difference between holding back information, as opposed to dressing it all up in misleading half truths. It maybe that the CTO had little choice, but more judicious editing and much more carefully chosen words and ideas should have been made. A corporation like SONY has the capability to rise above things in an honorable way, they really don’t need to sink to such a level and present such skewed information (even if it is rotated 45 degrees!).