Analysis of RAW samples at Fred Miranda show weak DR
- Thread starter Adelino
- Start date
Re: DRpocalypse 2017
Presuming enough sensor folks have dissected things that the measurement methodology is correct, this boils down to whether we think the RAW file was from a production camera vs. Canon possibly having left out a bedrock ground floor user expectation -- an on-chip ADC setup -- from the 6D2.
My money is still on an on-chip ADC setup in the 6D2, so I am still leaning toward this being pre-production output.
- A
BillB said:
Presuming enough sensor folks have dissected things that the measurement methodology is correct, this boils down to whether we think the RAW file was from a production camera vs. Canon possibly having left out a bedrock ground floor user expectation -- an on-chip ADC setup -- from the 6D2.
My money is still on an on-chip ADC setup in the 6D2, so I am still leaning toward this being pre-production output.
- A
Upvote
0
Other zany ideas from a non-sensor person:
Please forgive the mania with this. My afternoon coffee just kicked in and I want to eliminate other nutty ideas off of our virtual fishbone diagram right now. ;D
- A
- Any chance this is some new compressed RAW format we're seeing the output of? (Surely we'd able to glean format/compression from the file itself, wouldn't we?)
- Can these RAW files be doctored? Would we know if they were?
- This last one is a flying lark of an idea -- any chance the Auto Lighting Optimizer might have mucked up the RAW file? I always thought that affected JPGs only, but I spotted this from Canon Europe:
"If you make use of Canon's Digital Photo Professional workflow software, ALO can be applied to RAW images taken on compatible cameras during the post-processing workflow."
...but that sounds more like an after-the-fact DPP algorithm than something that gets cooked into the RAW file. (Just thought I'd throw it out there.)
Please forgive the mania with this. My afternoon coffee just kicked in and I want to eliminate other nutty ideas off of our virtual fishbone diagram right now. ;D
- A
Upvote
0
Re: Bill Claff Chimes In
Assuming that these calculations are indeed true, here is my optimistic scenario (that may not be technically feasible). I'm going to say that this sensor is a reject with failures in the ADC part of the silicon. The sensor itself passed, so they tossed it in a demo unit with off die ADC and saved their limited good chips coming off the line for the needed inventory to satisfy all us pre-order folks. ;D
Khalai said:
Assuming that these calculations are indeed true, here is my optimistic scenario (that may not be technically feasible). I'm going to say that this sensor is a reject with failures in the ADC part of the silicon. The sensor itself passed, so they tossed it in a demo unit with off die ADC and saved their limited good chips coming off the line for the needed inventory to satisfy all us pre-order folks. ;D
Upvote
0
Re: DRpocalypse 2017
GIGO - garbage in, garbage out
I'm sure the calculations are correct, but the variables put into them may not be. In the end, the sensor in production units will be the sensor in production units. Insisting one way or the other, before we have the actual production units, is just meaningless blather.
So I'm sure we'll get a couple of dozen more pages. What's release day, again?
BillB said:
GIGO - garbage in, garbage out
I'm sure the calculations are correct, but the variables put into them may not be. In the end, the sensor in production units will be the sensor in production units. Insisting one way or the other, before we have the actual production units, is just meaningless blather.
So I'm sure we'll get a couple of dozen more pages. What's release day, again?
Upvote
0
Re: DRpocalypse 2017
I say, wait for the release.....
it would be MORE work for Canon to not use the same design modules on the 6D2 as their other recent releases. They are not going to incur more work and expense to make a product worse.....ahsanford said:
I say, wait for the release.....
Upvote
0
It's possible that Canon will put out some additional promo literature prior to release. Could have some more sensor data in it -- like a cut- away showing sensor architecture. That would at least put to bed question like on or off sensor ADC.
Upvote
0
Re: DRpocalypse 2017
Nah.
Based on my limited understanding of CMOS manufacturing, production yields will always be an issue with FF sensors.
On-sensor ADCs increase complexity, which in turn results in lower yields.
So, the likely reason why the 6DII senor doesn't have on-chip ADCs is to maximize yields.
The 6DII will probably sell for $1600 in two years, so having low manufacturing costs is quite important for this camera.
Production yields are less of an issue for crop sensors.
This explains the weird balance of power, where the smaller/cheaper crop sensors are more advanced than their 6DII cousin.
This is really cost management, though, not 'crippling' or anything like that.
Also note that the 1DXII and 5DIV sensors are very new designs, so yields are likely still improving.
But by 2021, when the 6DII will be due for an update, the 5DIV sensor yields should be quite high.
At that time, the 6DIII will finally get (a variation of) the 5DIV sensor with on-chip ADCs.
Pure speculation and fun, of course - and a brand new 6DIII rumor to spice things up
.
Don Haines said:
Nah.
Based on my limited understanding of CMOS manufacturing, production yields will always be an issue with FF sensors.
On-sensor ADCs increase complexity, which in turn results in lower yields.
So, the likely reason why the 6DII senor doesn't have on-chip ADCs is to maximize yields.
The 6DII will probably sell for $1600 in two years, so having low manufacturing costs is quite important for this camera.
Production yields are less of an issue for crop sensors.
This explains the weird balance of power, where the smaller/cheaper crop sensors are more advanced than their 6DII cousin.
This is really cost management, though, not 'crippling' or anything like that.
Also note that the 1DXII and 5DIV sensors are very new designs, so yields are likely still improving.
But by 2021, when the 6DII will be due for an update, the 5DIV sensor yields should be quite high.
At that time, the 6DIII will finally get (a variation of) the 5DIV sensor with on-chip ADCs.
Pure speculation and fun, of course - and a brand new 6DIII rumor to spice things up
.
Upvote
0
Re: DRpocalypse 2017
In the Canon page for the 6D2 they state, "the EOS 6D Mark II’s sensor captures images of 6240 x 4160 pixels with a pixel size of 5.67 µm square for outstanding detail and a superb signal-to-noise ratio, resulting in great images. "
Something funny is going on. I still have not cancelled my preorder.
Brian
ahsanford said:
In the Canon page for the 6D2 they state, "the EOS 6D Mark II’s sensor captures images of 6240 x 4160 pixels with a pixel size of 5.67 µm square for outstanding detail and a superb signal-to-noise ratio, resulting in great images. "
Something funny is going on. I still have not cancelled my preorder.
Brian
Upvote
0
Re: DRpocalypse 2017
None of the words Canon uses can be considered to be factual, rather just the result of observations being made and thus opinions. The burger place just down the corners makes burgers and they advertise them as being amongst the best in the world, don't ya know. And the taco place next to it uses an authentic Mexican recipe. Neither of those statements provide any solid ground about what the end product is, just as Canon's statement says nothing absolute. From their perspective, it's not wrong but similarly neither does Canon state that "6D Mark II delivers 13 stops of DR." Nor do they even say "class leading."
I think this is the most likely answer to all of the questions and head scratching and in essence, Canon's BSI/ADC on-chip process for 35mm sensors isn't as well refined as (say) Sony's.
Canon have said "DR about the same as the 5D4", which leaves room for interpretation. At ISO >= 3200, that statement is true. Similarly at Bill's plots of ISO 12,800 and higher, the 6D Mark II has slightly better DR than the 6D.
hbr said:
None of the words Canon uses can be considered to be factual, rather just the result of observations being made and thus opinions. The burger place just down the corners makes burgers and they advertise them as being amongst the best in the world, don't ya know. And the taco place next to it uses an authentic Mexican recipe. Neither of those statements provide any solid ground about what the end product is, just as Canon's statement says nothing absolute. From their perspective, it's not wrong but similarly neither does Canon state that "6D Mark II delivers 13 stops of DR." Nor do they even say "class leading."
x-vision said:
I think this is the most likely answer to all of the questions and head scratching and in essence, Canon's BSI/ADC on-chip process for 35mm sensors isn't as well refined as (say) Sony's.
Canon have said "DR about the same as the 5D4", which leaves room for interpretation. At ISO >= 3200, that statement is true. Similarly at Bill's plots of ISO 12,800 and higher, the 6D Mark II has slightly better DR than the 6D.
Upvote
0
Re: DRpocalypse 2017
;D
neuroanatomist said:hbr said:
Seems you've missed the demonstrations posted here and especially on DPR, which unequivocally show that 'great images' require at least 12.8 stops of DR and >4 stops of shadow lifting capability. Clearly, Canon is lying.
;D
Upvote
0
Similar threads
