Scarlet & Epic 5K resolution only 1667 pix?

[Ken Martini]

Difference between 1080 X 1920 3 chip camera and Epic 5K 1 chip:
Divide 5K by 3 and you get a 1667 pix Epic camera.
The 3 chip delivers 1920 pix, a higher resolution.

Is my math wrong, or did Red make up new ways of measuring pixel count?
The truth will set you free, I think

The main compelling reasons it seems to be in this camera system is the frame size, PL mount, and cost. You would get less artifacts from 3 chips verses 1 chip.

 

[Eki Halkka]

Is my math wrong, or did Red make up new ways of measuring pixel count?

Your math is wrong.

 

[Liam Hall]

Very wrong, you are measuring apples and oranges.

RED is RAW.

Look up debayering and converting to RGB and you'll find your answer.

 

[GPSchnyder]

Your math is wrong.

Damn wrong to be precise...

 

[Fredrik Harreschou]

 

[Josh Negrin]

Where did you get those numbers? Also note, the RED is a single chip camera. The RED's sensor alone is far more advanced than most pro three chip cameras on the market and all the prosumer three chip cameras.

Also, there are many other factors to consider when purchasing a camera; the sensor is a big one, recording format (filetype, bitrate, etc.), and workflow, to name a few more.

You have much to learn, young grasshopper. But that's why we have this forum and we're here to help.

 

[J. Eric Camp]

Wow, everything you said was skewed.

As was said, look up Bayer pattern CMOS sensors versus 3 chip CCD's

The 3 chip 1920 camera will deliver a 1920 horizontal image.

The 5k Epic will deliver a roughly 5000 horizontal image, (with the MTF of roughly 4k)

On a per photosite count 3 1920 x 1080 chips have 6,220,800 photo sites to deliver an image that has accurate color and sharpness with 2,073,600 pixels. Ergo 33% effective.

The 4k Red one, a 1 chip CMOS with active recording area of 4096 x 2304 has 9,437,184 photo sites. This delivers an image with 9,437,184 pixels. Now since it is a bayer pattern it's accuracy lands somewhere around a 3.2k image which would have somewhere around 7,361,003 pixels, or 78% effective.

To start....

 

[Daniel Browning]

The 3 chip delivers 1920 pix, a higher resolution... less artifacts

Yes. Also, the earth is flat, snake oil cures all ailments, tinfoil hats stop aliens, and I have a bridge to sell you.

 

[Michael Olsen]

Though it's a bit off topic, I'd at least like to make sure that my understanding of things is correct and I think it might help the OP. (I didn't know anything at all about cameras prior to starting to read up over the last week, so please - don't think I'm trolling if I get something wrong.)

Some cameras use 3 CCD sensors to record an image - one sensor at N resolution for each color red, green, and blue. RED uses a single CMOS sensor or resolution N to record an image along with a bayer filter to break the pixels into colors at a ratio of 1 Red, 1 Blue, and 2 Green. The image recorded in the camera is exported to the media (CF card, HDD, SDD) in a RAW format, meaning ISO, ASA, and other adjustments normally done in camera can be altered in post without degredation.

Whereas most cameras use 3 CCD sesnors at N x N resolution to produce a single frame of resolution N x N, RED uses just the one CMOS at N x N resolution to produce a single frame of resolution N x N. The fact that it does not have three sensors does not imply that it produces a final resolution of 1/3 its sensor resolution. Therefore, a RED camera with a sensor rated to record at 2K 2:1 actually outputs a file that has 2K (actually 2048px) in the horizontal and 1K (1024 px) in the vertical. A 3K 2:1 outputs a file that is 3K (3072 px) wide and (1536 px) tall.

RED cameras also use a rolling shutter, also used by many film cameras. This type of shutter can lend itself to some problems - skew, partial exposure, and the "Jellocam", which can be largely avoided or corrected for in the use of a well built mechanism taking these elements into account or in firmware.

 

[Stuart English]

RED camera v's 3-CCD

RED camera v's 3-CCD

You would get less artifacts from 3 chips verses 1 chip.

If all things were equal between cameras you might anticipate that to be so.

Some "proof" may exist in pro-sumer cameras of this "fact", but RED and 3-CCD 1080p designs are not at all equal.

Comparing a still from a DSLR v's a still extracted from a traditional video camera is a good proxy for the quality difference you will observe.

This is nothing to do with "new math", and anyone who may be suggesting otherwise to you are either misinformed or may have an alternative agenda.

If you want to make a comparison, try a Scarlet 2/3" (3K RAW) camera v's a top end 3-CCD camera such as a Sony F900 (1080p)

 

[Chuck Z]

Though it's a bit off topic, I'd at least like to make sure that my understanding of things is correct and I think it might help the OP. (I didn't know anything at all about cameras prior to starting to read up over the last week, so please - don't think I'm trolling if I get something wrong.)

Some cameras use 3 CCD sensors to record an image - one sensor at N resolution for each color red, green, and blue. RED uses a single CMOS sensor or resolution N to record an image along with a bayer filter to break the pixels into colors at a ratio of 1 Red, 1 Blue, and 2 Green. The image recorded in the camera is exported to the media (CF card, HDD, SDD) in a RAW format, meaning ISO, ASA, and other adjustments normally done in camera can be altered in post without degredation.

Whereas most cameras use 3 CCD sesnors at N x N resolution to produce a single frame of resolution N x N, RED uses just the one CMOS at N x N resolution to produce a single frame of resolution N x N. The fact that it does not have three sensors does not imply that it produces a final resolution of 1/3 its sensor resolution. Therefore, a RED camera with a sensor rated to record at 2K 2:1 actually outputs a file that has 2K (actually 2048px) in the horizontal and 1K (1024 px) in the vertical. A 3K 2:1 outputs a file that is 3K (3072 px) wide and (1536 px) tall.

RED cameras also use a rolling shutter, also used by many film cameras. This type of shutter can lend itself to some problems - skew, partial exposure, and the "Jellocam", which can be largely avoided or corrected for in the use of a well built mechanism taking these elements into account or in firmware.

You nailed it.

 

[David Mullen ASC]

Difference between 1080 X 1920 3 chip camera and Epic 5K 1 chip:
Divide 5K by 3 and you get a 1667 pix Epic camera.
The 3 chip delivers 1920 pix, a higher resolution.

Is my math wrong,

Well, yes -- even if you simply count which photosites are filtered, half of the Bayer sensor has green-filtered photosites, and one-quarter red and blue each. So if you measure the pixel resolution in the green channel only (which is often the most important channel for providing definition), that would be more than 1920 across, it would be 2500 if you start out with 5K.

But debayering algorithms are a bit more complicated than simply extracting half the pixels for green and a quarter for red and blue, then uprezzing each channel. For each filtered photosite, there can be some intelligent guesses as to the information for the other two colors.

So as a general rule, you could say that the final measurable resolution of the debayered image is roughly 3/4 of the original. However, there are lots of other factors that come into play that affect actual resolution and perception of sharpness. I certainly agree that some 1080P images from 3-CCD cameras can be quite sharp-looking. But they can also suffer from aliasing artifacts and some prism-block artifacts. Plus you are limited I think to f/1.6 or so as a maximum lens aperture for a prism-block camera due some sort of halation or flaring / veiling problem (I believe.) And you're limited to 2/3" sensor sizes.

I see it less as one approach being better than another, just they had different looks, just as 3-strip Technicolor looked different than Eastmancolor.

 

[Lee Saxon]

The 4k Red one, a 1 chip CMOS with active recording area of 4096 x 2304 has 9,437,184 photo sites. This delivers an image with 9,437,184 pixels. Now since it is a bayer pattern it's accuracy lands somewhere around a 3.2k image which would have somewhere around 7,361,003 pixels, or 78% effective.

Why does it lose so much resolution? My DSLR is 12.84 million total pixels and 12.4 million effective!

 

[Jeff Kilgroe]

Why does it lose so much resolution? My DSLR is 12.84 million total pixels and 12.4 million effective!

And just what DSLR is this? Have you actually shot a resolution chart with it? Keep in mind that "effective" pixels are not true resolution. The RED One currently has 9.43M effective pixels when shooting 4K 16:9 and it records at that full 4096x2304 9.43 megapixel resolution. But nominal measured resolution is closer to about 3.3K.

Same with DSLR's. No digital system out there gives equivalent measured resolution to their effective sensor resolution. If they did, the aliasing would be terrible. Many, perhaps even most, RGB CMOS sensors use a Bayer pattern and are subject to the same sort of resolution degradation of about 18-20% over effective pixel resolution that we see with the RED One.

 

[Lee Saxon]

And just what DSLR is this? Have you actually shot a resolution chart with it? Keep in mind that "effective" pixels are not true resolution. The RED One currently has 9.43M effective pixels when shooting 4K 16:9 and it records at that full 4096x2304 9.43 megapixel resolution.But nominal measured resolution is closer to about 3.3K.

Nikon D2x. I was referring to 12.4 mp recorded image. DSLR spec sheets tend to call that "effective pixels" or "effective resolution" so I thought he was telling me that the R1 recorded a 7.36 mp image.

The part about it recording a 9.44 mp / 4K image that's actually 7.36 mp / 3.3K resolution just 'wooshed' so far over my head I couldn't even see it.

Got a link to a Wikipedia article or something?

 

[Casey Green]

RED looks better.

 

[Erik Bien]

The part about it recording a 9.44 mp / 4K image that's actually 7.36 mp / 3.3K resolution just 'wooshed' so far over my head I couldn't even see it.

Got a link to a Wikipedia article or something?

Get a cup of coffee and watch these videos from Panavision if you really want a glimpse of how deep this particular rabbit hole goes (don't say you weren't warned!).

 

[Dan Hudgins]

Aliasing

Aliasing

Using a 4K to 6K sensor over a 1920 to 2K sensor has more to do with reducing Bayer and Aliasing artifacts than with "resolution". In a Digital Movie camera you need some kind of low pass filter, optical like a OLPF or digital/software to reduce aliasing which shows up on edges when things or the camera moves.

Using a 4K Bayer sensor has some edge over three 1920 sensors since the OLPF and software used can blur the aliasing better with the smaller pixels. In fact if you make the pixels small enough you can get sharper images after you downsize since the OLPF does not need to be so heavy. (Although you need some softness for any digital image even RGB to reduce aliasing.)

In the end you will downsize the 4K images to 1920 or 2K for projection, but you can end up with better results than if you start at that resolution because of the interaction between the lens and the sensor pixel size.

 

[Patrick Tresch]

(don't say you weren't warned!).

Thanks but it's old stuff we allready had many threads about this "presentations"...

See you.

Pat

 

[A. Järvinen]

The 4k Red one, a 1 chip CMOS with active recording area of 4096 x 2304 has 9,437,184 photo sites. This delivers an image with 9,437,184 pixels. Now since it is a bayer pattern it's accuracy lands somewhere around a 3.2k image which would have somewhere around 7,361,003 pixels, or 78% effective.

Doesn't bayer pattern affect also vertical resolution? If bayer pattern reduces resolution by ratio of 3.2*1024 / 4096 = 0.8 then 'effective' resolution would be (3.2*1024) * (2304*0.8) = 6039798 pixels, or 64%. Or am I missing something here?

Best regards,
A. Järvinen