What are the lines of resolution for 3K Scarlet

[Jagger Christian]

What are the lines of resolution for the Scarlet at 3K?

What aspect ratio would that be?

 

[Peter Majtan]

RED seams to be making their sensors 16:9. Now considering that Red One (4K) has Active Pixel Array of 4520 x 2540, You can expect Scarlet to be app. 3390 x 1904, but that is only a speculation to give You an idea.

"Officially" 3K (in 16:9) would mean 3072 x 1728, so the final specs should be between this and the above...

Hope this helps...

 

[Barry Green]

You cannot even begin to speculate as to how it will resolve. The sensor is different than Red One's, the pixel density is different, and we don't have any idea what the lens will be.

The recorded format will be 3072x1728 @ 16:9 (other aspect ratios may be included) but that doesn't imply what the resolved resolution will be.

 

[Peter Majtan]

I think HardRockCA was interested in the basic aspect ratio (which is most likely going to be 16:9) and we already know (thanks to Graeme) that the sweet spot of resolving the RED's Bayer is 78%. I think this is a fair assumption and valid speculation...

What buffles me is that You have used the words "will be" in regards to the recorded format. Given that 4K Red One was not 4096, but rather 4520 (h), how can You be so sure 3K Scarlet is going to be 3072? What we do know that it is going to be that, or more - unless You know more then we do...

 

[Peter Majtan]

PS: Assuming the minimum of 3072x1728 and the 78% efficiency of RED's Bayer decoding You will end up with at least 2396x1346 resolved RGB resolution...

 

[Dan Hudgins]

Lens will probably limit resolution.

Lens will probably limit resolution.

What are the lines of resolution for the Scarlet at 3K?

What aspect ratio would that be?

Since it will have a built in lens, the MTF curve of the lens will probably dominate the apperent resolution, and the focus will vary over the zoom range so the apperent resolution will change at different zoom settings...

For digital cameras with Bayer filters you want the lens circle of confusion about 4 pixels area, but lenses do not make neet spots so there is some comma that will extend 6 to 8 pixels or more off center.

To bad you can not swap off the low cost zoom for a good prime, even a simple 6 element prime lens can be sharper than a 12 element zoom, and give higher contrast.

 

[Justin K Phillips]

RED refers to 4520 as "4.5k." 4096 is 4k, 3072 is 3k. If Scarlet was 3.5k, they would call it that.

 

[Barry Green]

RED refers to 4520 as "4.5k." 4096 is 4k, 3072 is 3k. If Scarlet was 3.5k, they would call it that.

Exactly. And on the Red One it's 3072, using RedCode -- and the Scarlet has already been announced as using RedCode. So while nobody outside Red knows for sure (and there's questions as to whether anyone inside Red knows "for sure" since "everything is subject to change" and we're told to "count on it") it seems extremely likely that the Scarlet 3K recording mode will be 3072 wide; as for the # of pixels tall it depends on the aspect ratio (16:9, 2:1, etc).

 

[Barry Green]

I think HardRockCA was interested in the basic aspect ratio (which is most likely going to be 16:9) and we already know (thanks to Graeme) that the sweet spot of resolving the RED's Bayer is 78%. I think this is a fair assumption and valid speculation...

But -- and the reason I say "not so fast" -- is that it all depends on the lens too. There are many, many things that go into the ultimate final image, which include the sharpness of the lens, the performance of the OLPF, the sensor design, the demosaic process, all sorts of things. Red One can be given more of a flat-rate "78%" figure because you can mount the finest lens in the world onto it, and thus optimize that portion of the imaging chain. With Scarlet we're going to be using a fixed lens, so that 78% figure may not hold for Scarlet -- it's potentially limited by the sharpness of the fixed lens. Perhaps Scarlet will have a sharp-enough lens to give us a full resolved 78%, perhaps not. We don't know and won't know for a year.

 

[Barry Green]

What buffles me is that You have used the words "will be" in regards to the recorded format.

The 3K recorded format is already a known quantity, given that the Red One uses it. I guess it's possible they would change that for Scarlet, but frankly the thought had never crossed my mind; 3K should be 3K as far as a standardized recording format goes.

(then again, "everything" is subject to change...)

Given that 4K Red One was not 4096, but rather 4520 (h), how can You be so sure 3K Scarlet is going to be 3072?

Red One's SENSOR was 4520, but its recorded format is 4096.

 

[Peter Majtan]

I am not trying to fight with You Barry - You are absolutely right about the lens being part of a bigger picture, etc...

I just don't think HardRockCA was asking for so much "detail" in the first place...

But since we are where we are - I have few questions myself:

I am sure RED ONE has a "true" 4K recording mode (i.e. 4096), but how do You explain this "official" specs:

Full Pixel Array: 4900 (h) x 2580 (v)
Active Pixel Array: 4520 (h) x 2540 (v)

(this on is for UM3K as well):

I am aware of RED referring to RED ONE in some interviews (etc...) as 4.5K just to make the point that the actual ACTIVE resolution of the sensor is more then 4K (4096) (mind You in Bayer mosaic). But in general literature and elsewhere on net (and this forum) RED always call RED ONE 4K (not 4.5K) knowing fully well that the "true" RGB resolution is not even 4096. That is why they don't try to push too hard the 4.5K image. Correct me if I am wrong...

Red-Code on its own has nothing to do with resolution. You could have Red-Code at any resolution You would want (given some basic coding limitation, such as it might be multiple of 16, or similar...) Using the codec name to confirm resolution is just not right...

AND FINALLY:

I have never said it won't be 3K (3072). In fact I have specifically said it is going to be 3K (3072) OR more... A very fair assumption indeed...

Anyway - the original questions were:

What are the lines of resolution for the Scarlet at 3K?

Notice he asked "Scarlet" not the "f/2.8 zoom lens". If we are talking here "recorded" format of Scarlet it is fair to assume for it to be the 3072x1728 "true" 3K. How will it resolve to RGB? Graeme went to extensive lengths explaining the 78% efficiency of the "de-mosaicking" algorithm (not the sensor HW) as being the sweet spot. I think it is fair to assume Scarlet's Bayer is going to be the "same" Bayer as it is in RED ONE (Greame must be getting headache by now :innocent: ).

What aspect ratio would that be?

I think we all agree here that it is most-likely going to be 16:9 native sensor


HardRockCA - feel free to correct me if I have misunderstood what You are asking for...

 

[Jagger Christian]

HardRockCA - feel free to correct me if I have misunderstood what You are asking for...

That's pretty much it Pete! I just wanted to know what I could expect as a final full rez product from the camera, if I used the Scarlet at it's full potential and what aspect ratio it would be at.

 

[Dj Joofa]

I have few questions myself:

I am sure RED ONE has a "true" 4K recording mode (i.e. 4096), but how do You explain this "official" specs:

Full Pixel Array: 4900 (h) x 2580 (v)

Some of the pixels on typical sensors are used for black level monitoring, so it is not surprising that 4900 > 4520.

Active Pixel Array: 4520 (h) x 2540 (v)

Even for active pixel grid, typically space is allocated for issues such as vignetting, etc., so it is not surprising that 4520 > 4096.

 

[Peter Majtan]

That's pretty much it Pete! I just wanted to know what I could expect as a final full rez product from the camera, if I used the Scarlet at it's full potential and what aspect ratio it would be at.

I don't want to confuse You with all the tech-mambo-jumbo we have hurdled at You. If you ask any RED member they would answer You in a very simple way:

Scarlet is targeted for 2K delivery (2048x1152) in 16:9 format after accounting for all the bayer processing and other things You might "do to it" in the post. This will allow for direct scale to HD master (1920x1080) or "cropped" DCI master (2048x1080) without any loss of final quality. 2K is the "sweet" spot of 3K bayer sensor and besides the only delivery format larger then 2K-DCI or HD is 4K-DCI - which is clearly not intended Scarlet's playground.

 

[Barry Green]

I am not trying to fight with You Barry - You are absolutely right about the lens being part of a bigger picture, etc...

No fighting, I'm just trying to point out that (as in so many things) there is no easy stock pat answer. The answer, as almost always, is "it depends."

I am sure RED ONE has a "true" 4K recording mode (i.e. 4096), but how do You explain this "official" specs:

Full Pixel Array: 4900 (h) x 2580 (v)
Active Pixel Array: 4520 (h) x 2540 (v)

Okay, case in point. Here's the deal: SENSOR PIXELS HAVE NO RELATIONSHIP TO RECORDED FORMAT PIXELS. It's unfortunate that we even use the same term to describe them; it'd probably be less confusing if we used "photosites" to describe what's on the sensors, and "pixels" to describe what's in the recorded footage.

But they are not the same. They have no 1:1 relationship. Your sensor can have more or fewer pixels than your recording format. You have to take some effort to wrap your head around it, but it's a vital distinction.

There are three things at play here:
a) sensor resolution
b) recording format resolution
c) resolved detail

Those are not the same thing, although frequently people want to claim that they're all equivalent. The resolution of the sensor isn't directly tied to your recording format, and (in three chip systems) sometimes the sensor res can be much lower and still deliver great-looking images (because, in a case like that, you have to actually multiply each chip's pixel count by 3 to get a fair idea and be more comparable to a Bayer single-chip system). Look at the Genesis, its sensor resolution is 5760x2160, but its recorded format is 1920x1080. No similarity.

In the case of the Red, the sensor is slightly larger (in photosites) than the recording format. This is not a conflict. Most systems are like that. Sensor photosites do not correspond to frame pixels. In fact, in the case of a Bayer system, sensor photosites pretty much directly cannot correspond to frame pixels, because the sensor is black and white (with a color filter) whereas the recorded frame is an RGB color pixel. The sensor photosites need to be processed in context with their surrounding pixels before a final frame pixel can be synthesized.

So, for clarity, understand that sensor pixel dimensions are not related to recorded frame dimensions.

Okay, now on to the second thing: "resolution" vs. the recorded frame. Those aren't the same. They could be, and in the case of Red shooting 4K and delivering 2K they probably are the same thing (meaning, the 2K recorded frame holds 2K worth of resolved detail) but in full frame size, they are definitely NOT the same thing. For example, when shooting 4K, you'll have a 4K+ sensor imaging into a 4K frame size, but you'll get about 3.2K worth of resolved detail.

When someone's asking "lines of resolution" they are (assumedly) asking how much resolved detail there will be in the frame. This we cannot answer, not until we get our hands on finished units and test them to see what all the factors are that contribute to making the resolved detail softer than the recorded frame size. I would suspect, on a smaller/denser chip with a fixed lens, that we won't be hitting the full 78% ratio that has been observed for the Red One. I also think that the post saying "count on a great 2K finished frame" is probably spot-on.

Long-winded, and potentially confusing, but I hope it addresses your question.

Red-Code on its own has nothing to do with resolution.

Correct, but it does have everything to do with frame size. Which is why I said that the frame size would likely be the same between the Red One (in 3K mode) and the Scarlet in 3K mode.

I have never said it won't be 3K (3072). In fact I have specifically said it is going to be 3K (3072) OR more... A very fair assumption indeed...

For recorded frame size, yes. For resolved detail, no. That's all I'm saying.

Notice he asked "Scarlet" not the "f/2.8 zoom lens".

Right. Scarlet is not a chip, it's not a lens, it's not a recording format, it's the grand summation of all those things. And every one of those is going to influence the final resolved detail. Which is why I say, we cannot know what the actual resolved resolution will be. Not until we can test an operational Scarlet and see what it actually produces.

 

[Barry Green]

Scarlet is targeted for 2K delivery (2048x1152) in 16:9 format after accounting for all the bayer processing and other things You might "do to it" in the post. This will allow for direct scale to HD master (1920x1080) or "cropped" DCI master (2048x1080) without any loss of final quality. 2K is the "sweet" spot of 3K bayer sensor and besides the only delivery format larger then 2K-DCI or HD is 4K-DCI - which is clearly not intended Scarlet's playground.

I would agree with all of that, I think that's a great summary.

 

[Emanuel A.]

Ipsis verbis.

 

[Peter Majtan]

Ipsis verbis.

My point exactly. Without "hiding" (no offense Barry) behind ton of technical mumbo jumbo - which does confuse the hell out of "normal" (read "not like us nerd engineers") users - we were talking about the same thing. I was trying to explain to HardRockCA in lemons terms what can he expect from Scarlet. I do wrap my head around it quite easy (I was for the last two years Head of R&D and Special Projects for EYEdentify Inc. in Tokyo and for over a decade Technical Director for DERYL Group), but what I have found most difficult in our industry is to be able to explain complex technical issues to "normal" users in a way that they can actually understand (no offense HardRockCA - for all I know You could have 30 years of R&D and engineering experience making me the "trainee") and that will help them to get the best out of their gear...

But I do have to disagree on one point You are defending: "Here's the deal: SENSOR PIXELS HAVE NO RELATIONSHIP TO RECORDED FORMAT PIXELS."

They do. I have never said it was 1:1, but the relationship is there. The actual ratio directly depend on the algorithm used to de-mosaic the sensor pattern (be it Bayer or anything else). But the relationship is there. This is precisely why You have a FIXED selection of recording formats. Otherwise You could have a "variable" recording format and choose anything You wanted in 1-px increments. Hell You could even (according to Your logic) "RAMP" the recorded image resolution. But this is not the case. Of course You could argue that in POST You can do anything to the image (I have seen 8 MPx image printed on 400 x 200 cm photograph!), but that is in the POST. Inside the camera Your recording options are limited and FIXED based precisely on the EXISTING relationship between the sensor pixels and de-mosaicking algorithm...

I hope this makes it clear without the mumbo-jumbo...

:innocent:

 

[Peter Majtan]

Man - I must have a suicidal tendencies going around and "poking" Jedi Masters. My apology Barry if I have came across a little harsh - this was not my intention. I just wanted to clear this before a "simple" miss-information could evolve into yet another urban myth regardless of our original intentions... We do have the responsibility to post correct info - especially when talking mambo-jumbo on forum such as this one. Otherwise I can hire any SONY rep to replace me as the Technical Director and Head of R&D... :biggrin:

 

[Barry Green]

Peter, I totally understand, and all is fine. When I go into those long-winded explanations I am conscious that I'm not writing for an audience of one, this is there for all to see, so I try to explain it roundly and fully. In no way was I addressing you personally.

The final summation of Scarlet having a 3K-sized recording frame, and making wonderful 2K finished products, is probably very accurate. My point was that we don't know what the total resolution will be until we test a finished product.

As far as pixels equating to end-frame pixels: I can't agree, because they don't. There is no direct correspondence at all. Genesis uses 12 million photosites in its sensor, and 2 million pixels in its recorded frame. Red uses 8 million pixels in its sensor, and 8 million pixels in its recorded frame. Panasonic uses 1 or 2 million pixels in each of its sensors, *3 = 3 million to 6 million sensor photosites, but 2 million pixels in their recorded frame. In the SD days you had a recorded frame of 345,000 pixels, yet you could have a Canon XL1 with 235,000 sensor photosites (per sensor) or you could have a Sony PDX10 with over a million photosites (per sensor) all recording to that same 345,000-pixel frame.

You cannot directly track the presence or existence of any particular photosite throughout the chain of imaging. There is no way to say that any particular sensor pixel "becomes" any final recorded frame pixel. Especially in a single-sensor system, where basically all the sensor photosites get thrown into a big "soup" of pixels, and new (entirely different) pixels are synthesized and extracted.

Further compounding it is that CCD sensors are analog devices. There is no way to access any individual sensor pixel. The CCD outputs an analog signal that needs to be sampled back into the digital domain -- it's more like a microphone in that aspect (and nobody ever asks if your microphone is "16-bit", right? Just because we're recording 16-bit digital audio, we don't go around shopping for "16-bit" microphones, do we? We all understand that microphones output an analog voltage signal, which needs to be "digitized" into 16-bit samples. Well, a CCD does the same thing.)

I'm just pointing out that sensor pixels are decoupled from recorded-frame pixels. In the case of the Red One, you'll have a front end where 2 million blue pixels, 2 million red pixels, and 4 million green pixels all get poured into the mix, stuck in the oven and baked, and out the other end will come 8 million 12-bit full-color RGB pixels. What happens in the sensor pixels influences what happens in the final pixels, obviously, but it's not like the sensor pixels are directly represented in the final recorded frame.