rrccad

Lives in Canada Canada
Joined on Jul 5, 2012

Comments

Total: 3081, showing: 1 – 20
« First‹ Previous12345Next ›Last »
In reply to:

falconeyes: I am sorry to disappoint, but the math is flawed.

> This means that, in essence, the camera is automatically capturing two shots, bracketed by about 1EV.

This is a misconception. The A, and A+B signals aren’t statistically independent. There is ZERO gain in DR over the A+B signal alone. Think of the two Raws A, B. The added Raw A+B does indeed have added DR over both A and B. But that’s the standard Raw to start with. There at best could be a minimum effect from reducing the read error if A and B are read additionally to A+B.

The entire thing is NOT like bracketing. It is zero extra info in for zero extra info out. The people who wrote the software unfortunately didn’t look at statistical independence.

Also, there is no reason to believe that A doesn’t clip where A+B does. Because A and B are formed from photosites of half size and half FWC wrt to a full A+B photosite aka pixel.

"so it's very well that there must be overflow in A and B because without it you could saturate A or B prematurely because of phase difference."

"Not how that works; this is just a simple matter of botched scaling"

Right.

and you're still a wannabe engineer in actuality. Canon has been very concerned about overflow on one side of the pixel causing color shift as it would.

clipping of a channel can still occur and average does not stop that, and that's not why they have chosen one value of the entire summation, and another for the individual half pixels.

nor does average "fix" anything not to mention to be actually implemented on the sensor itself since it's analog summed.

site references, oh please here's one.. the dozens of canon patent applications on the very subject. but you know better. of course.

Link | Posted on Feb 13, 2018 at 17:10 UTC
In reply to:

falconeyes: I am sorry to disappoint, but the math is flawed.

> This means that, in essence, the camera is automatically capturing two shots, bracketed by about 1EV.

This is a misconception. The A, and A+B signals aren’t statistically independent. There is ZERO gain in DR over the A+B signal alone. Think of the two Raws A, B. The added Raw A+B does indeed have added DR over both A and B. But that’s the standard Raw to start with. There at best could be a minimum effect from reducing the read error if A and B are read additionally to A+B.

The entire thing is NOT like bracketing. It is zero extra info in for zero extra info out. The people who wrote the software unfortunately didn’t look at statistical independence.

Also, there is no reason to believe that A doesn’t clip where A+B does. Because A and B are formed from photosites of half size and half FWC wrt to a full A+B photosite aka pixel.

to add .. you're still not getting why this clips before the sum of FWC for A and B, and I'm tired of explaining it to wannabe engineers that think they know better than Canon.

Link | Posted on Feb 13, 2018 at 15:10 UTC
In reply to:

Gmon750: Yep.. all the dual-pixel, armchair engineers are out in full-force right now on DPR.

LOL .. Right?

it's like a feeding freenzy for the sensor fabrication and dual pixel engineers out there that know more. Clearly Canon should just cast a wide net through dpreview and hire all these engineers, it's a goldmine in here.

Link | Posted on Feb 13, 2018 at 15:09 UTC
In reply to:

falconeyes: I am sorry to disappoint, but the math is flawed.

> This means that, in essence, the camera is automatically capturing two shots, bracketed by about 1EV.

This is a misconception. The A, and A+B signals aren’t statistically independent. There is ZERO gain in DR over the A+B signal alone. Think of the two Raws A, B. The added Raw A+B does indeed have added DR over both A and B. But that’s the standard Raw to start with. There at best could be a minimum effect from reducing the read error if A and B are read additionally to A+B.

The entire thing is NOT like bracketing. It is zero extra info in for zero extra info out. The people who wrote the software unfortunately didn’t look at statistical independence.

Also, there is no reason to believe that A doesn’t clip where A+B does. Because A and B are formed from photosites of half size and half FWC wrt to a full A+B photosite aka pixel.

oh yes, I'm sure canon's engineers know less about it than you do .. lol.

good grief dpreview is getting batty.

Link | Posted on Feb 13, 2018 at 14:54 UTC
In reply to:

falconeyes: I am sorry to disappoint, but the math is flawed.

> This means that, in essence, the camera is automatically capturing two shots, bracketed by about 1EV.

This is a misconception. The A, and A+B signals aren’t statistically independent. There is ZERO gain in DR over the A+B signal alone. Think of the two Raws A, B. The added Raw A+B does indeed have added DR over both A and B. But that’s the standard Raw to start with. There at best could be a minimum effect from reducing the read error if A and B are read additionally to A+B.

The entire thing is NOT like bracketing. It is zero extra info in for zero extra info out. The people who wrote the software unfortunately didn’t look at statistical independence.

Also, there is no reason to believe that A doesn’t clip where A+B does. Because A and B are formed from photosites of half size and half FWC wrt to a full A+B photosite aka pixel.

Nope you are not getting it. It's kind of funny you think you know more about how DPAF works and how you think Canon botched it when they have done countless patent applications on the DPAF sensors and related technologies. Average would actually be kind of stupid and still not get around the problem anyways.

Link | Posted on Feb 13, 2018 at 05:54 UTC
In reply to:

falconeyes: I am sorry to disappoint, but the math is flawed.

> This means that, in essence, the camera is automatically capturing two shots, bracketed by about 1EV.

This is a misconception. The A, and A+B signals aren’t statistically independent. There is ZERO gain in DR over the A+B signal alone. Think of the two Raws A, B. The added Raw A+B does indeed have added DR over both A and B. But that’s the standard Raw to start with. There at best could be a minimum effect from reducing the read error if A and B are read additionally to A+B.

The entire thing is NOT like bracketing. It is zero extra info in for zero extra info out. The people who wrote the software unfortunately didn’t look at statistical independence.

Also, there is no reason to believe that A doesn’t clip where A+B does. Because A and B are formed from photosites of half size and half FWC wrt to a full A+B photosite aka pixel.

Except you're not taking into account the sole reason DPAF sensors exist which is to record phase difference between the split pixels, so it's very well that there must be overflow in A and B because without it you could saturate A or B prematurely because of phase difference.

Link | Posted on Feb 13, 2018 at 04:28 UTC
In reply to:

Sacher Khoudari: Doesn't make any sense to me. If the signal from A+B is saturated (full well capacity reached), then each A and B will be most likely saturated too. You won't gain any DR here. Nor on the lower end, because by doubling the number of pixels (A+B -> A and B), each pixel will have about 41% more noise.

@Entropy512 "- I'm quite familiar with what their basic architecture is. It's pretty similar to how newer Sonys do it, except Sony steals blue pixels for the "dual" pixel. Now as far as specifics, such as your assertion that the subframe in a DPRAW is taken at a different time (as you have asserted) - please prove those specifics of the architecture."

I made no such assertion, perhaps reading comprehension is in order. the entire sensor is read in two ways, one for AF which each pixel is split and read independently, and other when summed.

secondly .. LOL .. DPAF isn't nothing like sony's implementations.

" Iliah has come up with some pretty solid evidence that A+B truncates well before A or B alone truncate. "

It should - So what? for reasons already explained. if you can't follow along, don't claim you know something :)

Link | Posted on Feb 13, 2018 at 04:24 UTC
In reply to:

Sacher Khoudari: Doesn't make any sense to me. If the signal from A+B is saturated (full well capacity reached), then each A and B will be most likely saturated too. You won't gain any DR here. Nor on the lower end, because by doubling the number of pixels (A+B -> A and B), each pixel will have about 41% more noise.

@Illiah, that's a valid clipping case though. A or B hitting FWC when the sum would not necessarily clip if it was a singular pixel is not a valid case.

Link | Posted on Feb 12, 2018 at 23:50 UTC
In reply to:

Sacher Khoudari: Doesn't make any sense to me. If the signal from A+B is saturated (full well capacity reached), then each A and B will be most likely saturated too. You won't gain any DR here. Nor on the lower end, because by doubling the number of pixels (A+B -> A and B), each pixel will have about 41% more noise.

@entropy -- perhaps you should do some light reading on what DPAF sensors are to start off with.

Link | Posted on Feb 12, 2018 at 23:50 UTC
In reply to:

Sacher Khoudari: Doesn't make any sense to me. If the signal from A+B is saturated (full well capacity reached), then each A and B will be most likely saturated too. You won't gain any DR here. Nor on the lower end, because by doubling the number of pixels (A+B -> A and B), each pixel will have about 41% more noise.

pixels are summed before ADC. they are read separately for AF calculations.

the total value stored in each sub pixel at the point of clipping the overall pixel is less than the FWC capacity of the sensor overall pixel itself.

it has to be because the primary reason for dual pixel is to provide phase difference between A and B sub pixels.

Clipping A or B while the overall value of A+B would not be clipped is a bad bad thing to have happen. A+B may also still clip but that's another instance. the main goal is to protect a+b from clipping too soon.

This entire subthread is ignoring phase difference and that A <> B, and if A or B is clipped A+B is a false value.

there's going to be no magical camera with extra DR because they aren't using it all in the 5D Mark IV, nor is canon doing anything "dumb" .

Link | Posted on Feb 12, 2018 at 23:38 UTC
In reply to:

Entropy512: For this to actually work to enhance DR would require Canon to be doing something dumb, such as clipping the A+B (which is a computed value) by a lower maximum value than 2x the maximum possible value for A.

Or another way of thinking about it: The sensels used to capture the A image will have half the full well capacity of the "computed" sensels A+B image generated by adding two pixel values together.

@entropy512, no actually the dumb mistake is your assumptions.

Link | Posted on Feb 12, 2018 at 22:35 UTC
In reply to:

HS Wells: finally camera that has bluetooth ? no more USB cable connect to PC to transfer files ! Come on it's 2018. No more usb cable please.

Finally a camera that has Bluetooth? Where have you been?

Link | Posted on Feb 7, 2018 at 11:38 UTC

This really wouldn't work. you'd just be told to unlock the photos / camera or risk going to jail.

Even a country like Canada, you can go to jail if you don't unlock your phone for customs agents at the border. think it'd be different for encrypted cameras?

Link | Posted on Feb 6, 2018 at 19:16 UTC as 72nd comment | 6 replies
In reply to:

Imager of: 15fps 4k video recoding fpr 5 minutes. Wow, that’s stunning! 🤔

yes because reading past the title is hard.
"the EOS M6 camera features Combination IS, which uses the camera’s Digital IS with a compatible lens’s Optical IS to help lessen the chance of blur due to camera shake. "

Link | Posted on Jan 31, 2018 at 16:46 UTC
In reply to:

Imager of: 15fps 4k video recoding fpr 5 minutes. Wow, that’s stunning! 🤔

@sibuzaru
canon claimed it was electronic stabilization. technically correct and it actually works.

this was also technically correct, but dpreview listed / really stated it wrongly, whether or not you can say that 4k at 15fps actually works is a different story.

Link | Posted on Jan 31, 2018 at 14:09 UTC
In reply to:

eno2: This looks like a very nice entry LV camera and surprisingly very well priced.

To bad Fuji doesn't want to give us a top of the line APS-C camera with Bayer sensor array. :(

". Considering there are so many pros dumping D800s and Sony A7Rs for the Fuji XT2,"

surreeee there iss

Link | Posted on Jan 31, 2018 at 14:07 UTC

the patent application also does a stepless aperture which is probably more important for EF lenses because IMO, I'm not sure if the lenses themselves support that.

Link | Posted on Jan 27, 2018 at 14:48 UTC as 15th comment | 2 replies
In reply to:

rrccad: One thing to note, this is very americanized. there's alot of people on this planet that will never be able to afford an iPhoneX or related top end smartphone because they are popping in at $1000 or more.

oh i hear you cosina, i'm the same way - however after travelling especially around central and south america; these top end phones are even more pricey down here, and it's alot harder to justify a phone that is around 1 month or more of your wage.

Link | Posted on Jan 22, 2018 at 17:00 UTC
In reply to:

rrccad: One thing to note, this is very americanized. there's alot of people on this planet that will never be able to afford an iPhoneX or related top end smartphone because they are popping in at $1000 or more.

can't read?

Link | Posted on Jan 22, 2018 at 16:38 UTC
In reply to:

Photato: Where is the iPhone superior to the M100?
1. Wider Color Gamut P3 vs sRGB
2. 4K video vs 1080P
3. Panorama function vs Nothing
4. Time Lapse uses all the iPhone sensor, M100 uses a mushi 1080P video.
5. Faster setup of custom WB
6. Live Photos are 12MP worth, M100 Live Photos are 720P video.
7. Video can be controlled and monitored from other iPhone or iPad, M100 doesn't has that functionality.
8. Longer battery life.
9. No need of separate charger, great for travel.
10. Bigger, brighter and higher resolution screen.
11. Next generation more efficient codecs fro Image and Video such HEIC and HEVC.
12. Built-in GPS
The list is longer but gonna leave it up to here.

I'm comparing apples to apples (no pun intended) which is; it's 1300 without any mobile phone lock in for it.

perhaps you should do the same. the fact remains, it's alot more expensive than a M100, it's meant to do things differently, so what? it's a tool. get over yourself and your perpetual whining. good grief.

I can easily blurr a EF-M lens because I know how to use a camera.

the same list could be made with ANY manufacturer's lowest end camera.

Link | Posted on Jan 22, 2018 at 16:37 UTC
Total: 3081, showing: 1 – 20
« First‹ Previous12345Next ›Last »