Canon 5D Mk4 announced

[Phil Indeblanc]

That would be a nice touch...even nicer if it was removable wireless :-)

[uaiomex]

Same here. Still keeping my 6D. No true incentives for me.

I've been waiting for years now to see an articulated LCD screen in a pro Canon camera. Seems like it's never gonna happen 

[kers]

this might be of interest

http://www.rawdigger.com/howtouse/Canon-dual-pixel-technology

one stop extra in the highlights

[shadowblade]

this might be of interest

http://www.rawdigger.com/howtouse/Canon-dual-pixel-technology

one stop extra in the highlights

Of course, that doesn't tell us what the base DR of the sensor is...

Either way, with increased DR compared to the 5D3 due to the on-sensor ADC, as well as a further stop from dual-pixel technology, it looks promising.

I'm hoping for a 50+ MP 5Ds2 with similar technology soon. Yes, it would mean a very short product cycle for the 5Ds/5Dsr. But, so what? The 5Ds was the last camera released without the improved sensor. It's also the camera which would benefit, more than any other, from the technology, due to its intended use. The 5D4 is positioned against the D750's successor as a wedding/general-purpose camera, and looks like it will do pretty well at that (although we haven't seen what Nikon has in mind yet). The 5Ds/5Dsr are positioned directly against the D810 and A7r2, and falls flat in one of the key areas of importance for many users of such cameras. The new sensor technology can rectify this.

[scyth]

The 5D4 is positioned against the D750's successor as a wedding/general-purpose camera

https://www.dpreview.com/products/canon/slrs/canon_eos5dmkiv = "The EOS 5D Mark IV will go on sale in early September with an MSRP of $3499."

https://www.dpreview.com/articles/9542152490/nikon-announces-d810-full-frame-dslr = "The D810 will be available in late July for $3299.95"

https://www.dpreview.com/articles/7222282087/nikon-d750-fx-dslr-announced = "The Nikon D750 will be available in September body-only for $2299.95"


do you think you will buy the successor of D750 or the successor of D8*** if you are going to buy Nikon and spend the same money ?

[Bart_van_der_Wolf]

Of course, that doesn't tell us what the base DR of the sensor is...

Hi,

Assuming that Rawdigger interprets the Raw file correctly, and based on a single camera, the Base DR (@ISO 50 and ISO 100 are the same) of the sensor is about 12.64 stops. Add something like 1 stop if the Dual Raw pixels could be exploited by software during Raw conversion.

In more detail, for single pixel mode, after Black point subtraction for the Saturation level, and read noise based on the masked pixels:
      Sat / sigma    log2     Dynamic range
R :   15872 / 2.09   -->    DR_R  =  12.89
G :   15872 / 2.49   -->    DR_G  =  12.64
B :   15872 / 2.87   -->    DR_B  =  12.43
G2:  15872 / 2.44   -->    DR_G2 = 12.67

Either way, with increased DR compared to the 5D3 due to the on-sensor ADC, as well as a further stop from dual-pixel technology, it looks promising.

Yes, a very attractive package.

I'm hoping for a 50+ MP 5Ds2 with similar technology soon. Yes, it would mean a very short product cycle for the 5Ds/5Dsr. But, so what?

It may not be in Canon's interest to rush things. They have a typical far away Japanese planning horizon, and tend to use steady incremental upgrades to their models. In research they are much farther ahead.

Cheers,
Bart

[Bart_van_der_Wolf]

And this might be of interest

Dual pixel raw can make interesting Stereoscopic image

The technique is simple, just use Canon's DPP to produce two images that have Left and Right Bokeh shift applied, and use another application to turn the image pair into a Stereoscopic format. StereoPhoto Maker will allow to produce such stereo images (e.g. Anaglyph) from image pairs.

Since the stereo-base is very small (typically less than human vision), close-up subjects work best.

Cheers,
Bart

[BernardLanguillier]

No detailed DR measure yet? Usually they come right after camera availability?

Cheers,
Bernard

[bcooter]

I've been waiting for years now to see an articulated LCD screen in a pro Canon camera. Seems like it's never gonna happen 

Canon can be frustrating at times with some of the things they leave in or take out.  Given that I have great luck with all of their cameras.  I've owned every 1ds and x camera, a bunch of 1d and only the 5d2, as the 5d3 produced a file I found harder to color and was a little too smooth for my tastes.

I guess it depends on what you shoot and what you are use to and lately Canon get's little love on forums and blogs, but they've always worked for me.

I'm not dissing Sony as they are quite innovative, though I have a difficult time with Sony color.  Maybe it's me, or the fact I'm just not that use to Sonys but my a7sII seems to have a mind of it's own.   Granted I shoot it mostly for motion footage and it's an 8 bit camera with a small bit rate in motion settings, but we can shoot one scene and the color is great, the next scene is like a totally different camera. 

But other's rave about it's low light capabilities so maybe it's me.

BC

[BernardLanguillier]

No detailed DR measure yet? Usually they come right after camera availability?

I'll answer my own question, the DxO measures of the 5Div have been published.

As expected they are significantly improved compared to previous Canon bodies and will serve all-round shooters very well.

A bit less impressive for tripod shooters where the 5D mkiv only equals the performance of an 8 years old D3x, but still good enough that most of the negative talk about Canon DR should come to a stop.

It is great to see them solve this issue!!!

Cheers,
Bernard

[Bart_van_der_Wolf]

I'll answer my own question, the DxO measures of the 5Div have been published.

Hi Bernard,

As has been my experience before, the DxOMark figures are close to the basic values everybody can detect by using RawDigger. They of course have a more extensive series of measurements, but the basics are usually consistent with what users can determine. I expected (Reply #86) an average DR of 12.64 @ base ISO , DxoMark reports a 'Screen' DR of 12.6 EV . So much for the DxOMark doubters.

As expected they are significantly improved compared to previous Canon bodies and will serve all-round shooters very well.

A bit less impressive for tripod shooters where the 5D mkiv only equals the performance of an 8 years old D3x, but still good enough that most of the negative talk about Canon DR should come to a stop.

It's a welcome improvement, but complainers will always complain.

Cheers,
Bart

[shadowblade]

Is that 12.6 stops with or without the extra stop from dual pixel RAW?

Just as importantly, is there any hint of pattern noise in the shadows? After all, that is what limited the 5D2's useable DR to less than 8 stops (well below the technical DR limit) and continued to impact on the 5D3, to a lesser degree.

[Bart_van_der_Wolf]

Is that 12.6 stops with or without the extra stop from dual pixel RAW?

Without.  AFAIK, there is currently no Raw converter that can exploit the additional stop, so that bonus is something for the future.

Just as importantly, is there any hint of pattern noise in the shadows? After all, that is what limited the 5D2's useable DR to less than 8 stops (well below the technical DR limit) and continued to impact on the 5D3, to a lesser degree.

The new sensor design apparently got rid of (most of) the pattern noise. But time and testing of multiple samples will tell.

Cheers,
Bart

[Jack Hogan]

AFAIK, there is currently no Raw converter that can exploit the additional stop, so that bonus is something for the future.

Will there ever be a bonus?  Half the pixel, half the FWC.  Where would the additional stop of headroom come from if the half pixel is already maxed out?

[Bart_van_der_Wolf]

Will there ever be a bonus?  Half the pixel, half the FWC.  Where would the additional stop of headroom come from if the half pixel is already maxed out?

Hi Jack,

Half the pixel, half the FWC, but also half the photosite aperture / exposure.

The mathematical approach would be to reduce the post-processing exposure of the combine channel, and add the half pixel exposure. More elegant schemes can be made, based on clipping or highlight roll-off

Cheers,
Bart

[Jack Hogan]

Half the pixel, half the FWC, but also half the photosite aperture / exposure.

I still don't get it Bart: half the area, half the FWC. Wouldn't the full and half pixel saturate at the same exposure?  With the same gain the count would be half in the saturated half pixel.

Jack

[Bart_van_der_Wolf]

I still don't get it Bart: half the area, half the FWC. Wouldn't the full and half pixel saturate at the same exposure?  With the same gain the count would be half in the saturated half pixel.

Hi Jack,

It seems like the full pixel is stored after summing the 2 subpixels, so if half a pixel is ETTR then the full pixel is at ETTR +1 EV (in the focus plane**). One obviously attempts to ETTR the full pixel, in which case the half pixel offers 1 additional stop of highlight recovery at ETTR -1 EV.

** Due to the phase shift at OOF areas, the registration would be have to be re-established for each pixel, which is complex and fraught with opportunity for errors, so that's probably why Canon have not implemented it.

It's an interesting find by Iliah Borg, but currently more a theoretical benefit than a practical one.

Cheers,
Bart

[Ray]

I'm still surprised that a full-frame format such as the 5D4 appears to have one full stop less DR at base ISO than the cropped-format Nikon D7200.

At higher ISOs the 5D4 has better DR of course, as one would expect, but not better to the same degree that the D7200 is better at base ISO. For example, between ISO1600 and 6400, the 5D4 is around 1/2 to 2/3rds of one EV better than the D7200. Only above ISO 12,800 is the 5D4 DR better by one full stop and more.

Since I never use ISOs above 6400 because of unacceptable noise in general, I don't see much advantage here with regard to 5D4's DR capability.

[BernardLanguillier]

We can only hope that the D900 will leverage the same Toshiba sensor tech as the D7200. That would put us firmly over 15 stop DR and would be a real 16bits sensor, probably the first ever, although the Sony 100mp MF part may be there already.

Cheers,
Bernard

[Ray]

I'd find a full-frame successor to the D810, with D7200 pixels, irresistible, even at a higher price than the current D810.

The pixel count of such a camera would be 54mp, and its  DR noticeably better than the current D810, as you mention, Bernard.  4k video with the facility to extract single frames in RAW, would  be an attractive bonus. I'd be willing to pay extra for that also.

When one compares the 5D4 with the D7200 at the pixel level, on DXOMark graphs, the DR advantages of the 5D4 at high ISO are even less, and the DR advantages of the D7200 at base ISO are even greater than one full stop.

Cheers!