Possible Canon EOS 5D Mark IV Spec Talk [CR2]

[Lee Jay]

ISO 100-204,800

Who can tell me why one should use an ISO of 204,800???
These specs does not seem very logic compared what you can get now.

In 2004, I heard the same thing about ISO 3200.
In 2011, I heard the same thing about ISO 12,800.
In 2013, I head the same thing about ISO 25,600.

I've been in plenty of extreme low-light conditions where I could see easily but I couldn't take pictures, even with an f/1.4 prime at ISO 12,800.

I did a little test one time to see what ISO it would take to get a picture about like I could see with my eyes after a full dark adaptation, at the same equivalent shutter speed (about 1/10th). My eyes can do black and white at an f-stop of about f/3.5 and an ISO of about 2 million when compared with a digital camera.

 

[DR. High ISO]

• 18mp Full Frame CMOS
• ISO 100-204,800
• 61 AF Points (all crosstype)
• 12fps
• Dual CFast
• 4K Video Capture

[Whichever model would be this..] Dear Santa..

 

[Lee Jay]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

Someday someone needs to explain to me why this myth persists after a decade of things going the other way despite similar basic sensor performance (QE).

 

[lintoni]

ISO 100-204,800

Who can tell me why one should use an ISO of 204,800???
These specs does not seem very logic compared what you can get now.

In 2004, I heard the same thing about ISO 3200.
In 2011, I heard the same thing about ISO 12,800.
In 2013, I head the same thing about ISO 25,600.

I've been in plenty of extreme low-light conditions where I could see easily but I couldn't take pictures, even with an f/1.4 prime at ISO 12,800.

I did a little test one time to see what ISO it would take to get a picture about like I could see with my eyes after a full dark adaptation, at the same equivalent shutter speed (about 1/10th). My eyes can do black and white at an f-stop of about f/3.5 and an ISO of about 2 million when compared with a digital camera.

The higher (usable) ISO the better for widefield astrophotography.

 

[pedro]

If Canon really come out with an 18 MP low light version of their next 5D I could be in the market for one in a few years. Till then, my 5D3 rocks...

 

[vscd]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

Someday someone needs to explain to me why this myth persists after a decade of things going the other way despite similar basic sensor performance (QE).

It's no myth, it's physics. Did you ever ask yourself why the ISO-Monster A7S has lesser Pixel (12MP) and the 5DS with 53MP looses ISO against the old Model (ISO6400)?

 

[Don Haines]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

Someday someone needs to explain to me why this myth persists after a decade of things going the other way despite similar basic sensor performance (QE).

Many things affect high ISO performance.... we can not fixate on just one yet many of us do.

Ultimately, it comes down to light...

A higher quantum efficiency of the sensor will lead to more light being converted into electrons, and will give you increased high ISO performance.

Lower read noise/higher SNR will give you a cleaner signal and will result in improved high ISO performance.

BSI sensors have less circuitry blocking the incoming light, which gives you more usable light, and results in increased high ISO performance.

Finer lithography blocks less light, and like above, gives better high ISO performance.

A cooler sensor (temperature cooler, not trendy cooler ) has less thermal noise and gives better high ISO performance.

And yes, pixel size does matter (sort of).... a smaller pixel will collect less light than a larger pixel and that pixel will have lower high ISO performance than a larger pixel.... but you have to remember that a picture is not made up out of one pixel.... it is the results of ALL the pixels that give you the picture. For example, think of two sensors built with the exact same technology and of the exact same size. One is a 10Mpixel sensor and the other is a 40Mpixel sensor. The pixels on the 10Mpixel sensor, being four times larger, will collect 4 times the light as the pixels from the 40Mpixel sensor and the ISO performance will be 2 stops better. However, we can not forget that the 40Mpixel sensor has 4 pixels covering the same area as each of the 10Mpixel pixels and that if you bin those 4 pixels together in post production you end up with very similar performance to the 10Mpixel sensor. Yes, you will loose a slightly greater amount of area to lithography, but you gain a bit from increased sampling density. Practically, the end result will be the same..... So in the end, larger pixels give you better PER PIXEL ISO performance but with normalizing the resulting images, very little difference in PER PICTURE ISO performance.

And yes, size matters. A FF sensor will capture more light than a crop sensor and for the reasons above, better ISO performance.

 

[binky7]

18mp Full Frame CMOS
ISO 100-204,800
61 AF Points (all crosstype)
12fps
Dual CFast
4K Video Capture

These specs suggest this model would be positioned to compete with the Sony A7s.

Let's start from the bottom of this list and work our way up:

- 4K video capture would be at least equivalent to the A7s or better if it can do it internally.

- The Dual CFast card capability suggests that they're willing to take the new card standard pains/grumbles from us to get a fast-enough card standard that can do 4K in-body.

- 12fps... If you consider the state of technology Canon has currently attained, their dual DIGIC 6 chips used in the new 5Ds & 5DsR bodies have enough internal data bandwidth to push 50mp x 5fps = ~250,000,000 bps. Now let's say we drop the same dual DIGIC 6 chips and connect a low-light sensitive sensor of say 18mp x 12fps= ~216,000,000 bps. That's well within the realm of possibility for current 2015 technology.

- 61 point AF is already a known update that we've seen on the 5Ds & 5DsR. No mystery there.

- The ISO 100-204,800 is a telling sign of this low-light competitor body from Canon, and is probably an H2 setting, which would put it at a likely native ISO of 100-51,200. For those who don't see how impressive noise-free high ISO capabilities are, just take a look at this Sony A7s demo here http://www.youtube.com/watch?v=7RyiS-mrp1c

- Finally, at least from Canon's marketing perspective, Canon's new low-light body with 18mp > Sony's 12mp A7s.
And calling it a 5D X would also mean that there's still room for the 5D IV for the general market looking for the successor to the Mark III.

 

[DR. High ISO]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

(I would say) ... by increasing image noise and decreasing DR.
(Never trust ISO numbers, camera makers always cheating a little bit with ISO-s.. )

 

[Lee Jay]

However, we can not forget that the 40Mpixel sensor has 4 pixels covering the same area as each of the 10Mpixel pixels and that if you bin those 4 pixels together in post production you end up with very similar performance to the 10Mpixel sensor.

I think this this the main reason the myth persists.

 

[Shoot3r]

: This system could be to just test paired Digic6+ or 7s. It could be for only testing the CFast card system. May be both to confirm a given standard before moving to a (new)24mp sensor. I certainly hope that the folks at Canon are smart enough to test new tech as it comes available and not wait until everything for a new system is acquired. That would be a debug/eval nightmare. Car companies do the same thing. If they have a new idea for a break system they test it on an existing platform rather than wait to create all the parts for a new model or line. R&D occurs at many levels.

 

[fusionz]

Makes sense - the 5ds and 5ds R= studio, 5d mIV = video / low-light/ semi-pro sports. Probably a good thing they leaked this as the SONY A7s is making waves in the video community and the 5d iii looks very tired compared to the new mirrorless options. 18MP if it has crazy iso performance will be useful to many and won't cause their current lens offerings to be outdated. All they need now is some real dynamic range and they would really be successful. I guess we can always hope. They better hurry up too, I would assume an A7s II @ 24 or 36MP is not far behind....

 

[DR. High ISO]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

Someday someone needs to explain to me why this myth persists after a decade of things going the other way despite similar basic sensor performance (QE).

Interesting article:
http://www.clarkvision.com/articles/does.pixel.size.matter/#sensorconstant

 

[vscd]

However, we can not forget that the 40Mpixel sensor has 4 pixels covering the same area as each of the 10Mpixel pixels and that if you bin those 4 pixels together in post production you end up with very similar performance to the 10Mpixel sensor.

I think this this the main reason the myth persists.

You assume the area of both sensors are the same. This may be true in theory, but practically there are microlenses above the pixels/sensors. Please look at the schemes of such an structure. 4 small lenses have not the same area as one lense one the whole area of 4 small pixels. You didn't answer my question anyway. I guess, on purpose.

 

[Light_Pilgrim]

If this happens, I am off Canon. I was a very happy camper so far for 6 years.
I currently own a 5D MKIII, but it looks like there is just no upgrade.
5Ds is not an upgrade and it looks like 5D MKIV is not an upgrade either.
Where is the equivalent of the D810?

 

[Lee Jay]

However, we can not forget that the 40Mpixel sensor has 4 pixels covering the same area as each of the 10Mpixel pixels and that if you bin those 4 pixels together in post production you end up with very similar performance to the 10Mpixel sensor.

I think this this the main reason the myth persists.

You assume the area of both sensors are the same.

Of course - I'm not comparing full-frame to crop here.

This may be true in theory, but practically there are microlenses above the pixels/sensors. Please look at the schemes of such an structure. 4 small lenses have not the same area as one lense one the whole area of 4 small pixels. You didn't answer my question anyway. I guess, on purpose.

The microlenses are actually the great equalizer. They take away a lot of the fill-factor difference cause by constant lithography size and variable pixel pitch.

 

[Light_Pilgrim]

If this is true, then I think Canon will hand over the still photo segment to Nikon while settling for MF wannabe and Cinema line cameras.

I have to say I am massively disappointed that there is no camera for ambient light landscape and people/street photographers. 5D MKIII is a good one, but it will not be replaced.

D810 is the what 5D MKIV should be....but it looks like it will never be the case.

 

[pixel8foto]

And why is a new card format an issue? I use the same cards in the same cameras, every day. Shoot, take card out to edit, put card back in, shoot again. If you're spending £2-£3k on a camera, another £100 odd for a superior format card to go with it seems perfectly reasonable.

They're crazy expensive right now.

A Lexar 64GB Cfast 2.0 card is $352.43.
A Lexar 64GB UDMA 7 CF card is $72.95.
A Lexar 64GB UHS-II SD card is $54.18.

Sorry, that doesn't seem "perfectly reasonable" to me.

For new tech? Every development starts off more expensive than the technology it supersedes. That's not a good reason to stand still.

 

[Lee Jay]

Having higher pixel densities just reduces ISO performance.

No it doesn't.

Yes it does.

Someday someone needs to explain to me why this myth persists after a decade of things going the other way despite similar basic sensor performance (QE).

It's no myth, it's physics. Did you ever ask yourself why the ISO-Monster A7S has lesser Pixel (12MP) and the 5DS with 53MP looses ISO against the old Model (ISO6400)?

The A7S has a better sensor, not just larger pixels.

The 5DS has the same pixel size as the 7DII, which is limited to ISO 16,000 bases, 51,200 extended. The fact that the 5DS is limited the way it is, despite the larger sensor (which gives it an inherent 1 1/3 stop advantage), proves that this is just another Canon artificial crippling exercise.

 

[pixel8foto]

Don't get people dismissing even the suggestion of a camera on the basis that its rumoured spec doesn't meet an arbitrary pixel count of their wanting. 18MP on a 1DX is plenty for many uses and delivers files with plenty of capacity to crop and print large. You want more from Canon? There's a 50MP model with different crop options already slated.

And why is a new card format an issue? I use the same cards in the same cameras, every day. Shoot, take card out to edit, put card back in, shoot again. If you're spending £2-£3k on a camera, another £100 odd for a superior format card to go with it seems perfectly reasonable.

There already IS a 1Dx, which is likely where this sensor is from. They don't need 2

I have no idea where the sensor from this hypothetical camera is from, if such a thing even exists. Do you?

Assuming a 1DXii or not, it's perfectly plausible that existing 1DX tech could be employed in a lower-level camera in the future. That seems like standard progression. I'd love to see some of the 1DX's functionality in a 5Dx form-factor DSLR, were that possible. Or maybe it'll be similar MP but with superior performance?

There already is a £5K 1DX *today*. Next year there may be a £3K camera with similar spec.