November 21, 2014, 09:54:58 PM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - jrista

Pages: 1 ... 76 77 [78] 79 80 ... 309
1156
EOS Bodies / Re: Can Canon deliver a FF sensor that is class leading?
« on: June 04, 2014, 03:22:01 PM »
...
Who says?  FWIW I don't believe that the flat spot has anything to do with the sensor at all -- it is due to the noise being dominated by that of the ADC which follows the sensor in the Canon designs.  If you really dig into things you will find that the Canon sensors are actually the class leaders in terms of dynamic range, they just foul it up in terms of their system architecture.  The answer to the question "can Canon deliver a class leading sensor" is yes... they already do.  Take a look at the data from Sensorgen for the 5DIII for example:  Min read noise 2.4e-, FWC=67531 for a DR of 14.7 stops.

And Nikon's D4 would whip everything from Canon quite nicely then.

Only at lower ISO, and only marginally. In every other respect, the Canon 1D X trounces the D4. In the most important aspects, frame rate and AF system (arguably the critical traits of such cameras), the 1D X has the technological and performance lead.

Quote
If they were to put an ADC on-chip similar to what Sony does, they would have equivalent performance maybe even better.

What Canon needs is a "class leading" system architecture and ADC.  This is a nit, I know but it is important because Canon DOES have class leading sensors and has had them for quite some time.

If you mean "system architecture" to mean "sensor architecture", then yes because the ADC is an integral part of the sensor.

Completely incorrect. In Canon "systems", the ADC is OFF-DIE. I don't know how many times I've written that when answering you...but that means it is NOT an integral part of the sensor. Canon's ADC units are in their off-sensor die digital signal processors...the DIGIC chips. That, as it turns out, is actually how MOST CMOS image sensors are designed. Sony Exmor is currently somewhat unique in that it has a column-parallel ADC integrated right into the sensor. Canon has patents for that kind of technology...they just haven't employed it in a commercial product yet.

So for the time being, David is correct...it's "system architecture", not "sensor architecture".

1157
EOS Bodies / Re: Can Canon deliver a FF sensor that is class leading?
« on: June 04, 2014, 07:28:57 AM »
So, while I understand what your trying to say...I think it's a misnomer. Dynamic range is itself defining what's usable...so saying that you can only use part of what's usable... Redundant and inaccurate. And, incorrect. ;P

Ok, thanks for explaining, I'm really not much of a tech geek even after all this time on CR :-p ... so for my education and to avoid further confusion: With which word(s) would you label the dynamic range starting from the shadows when no banding is visible anymore? Because that's what I called "usable" as fpn is what makes your shot "unusable".

That is dynamic range. The definition, in mathematical terms, is 20 * log(MaxSignal/NoiseRMS), to get dynamic range in decibels. The NoiseRMS is the Root Mean Square of the noise...basically if you sampled the actual noise in each pixel, and computed the RMS from that, that would be your noise floor. Dynamic range describes the usable potential signal range from the noise floor to the maximum signal (saturation point, in the case of ISO, that is also the Full Well Capacity, FWC.)

Maximum signal strength defines the range from "zero to"...if you had no noise, you would have the full range, say from 0.000...1e- to 76606e- (in the case of the 6D). If you could use the full potential signal range, the formula above becomes 20 * log(MaxSignal/0)...but you can't do that, it's divide by zero, which is undefined (undefined because the value becomes infinite...but infinity is also technically undefined). So, if you had no noise, you would have infinite dynamic range. Because noise exists, dynamic range must therefor be something less than infinity. You can keep lowering the noise floor. In astrophotography, we use supercooled CCD cameras that have extremely low dark current (while a DSLR may have as much as 20e-/s/px, a cooled CCD can have as little as 0.002e-/s/px), extremely high Q.E. (77% with Sony's new ICX CCDs is not uncommon, and some of the older Kodak KAF sensors (now owned by TrueSenseImaging, TSI) had as much as 90% Q.E.), and very low read noise at optimal gain. Further, we use more extreme noise reduction techniques, such as bias and dark calibration, which eliminates fixed sensor pattern noise as well as hot pixel noise, leaving us with what is effectively a pure gaussian read noise and photon shot noise signal. We then average together dozens of individual light frames to reduce noise even more (reduction is SQRT(subCount), so if we stack 100 subs, we get 1/10th the noise). In astrophotography, dynamic range of a final integration can be 25-30 stops or more (and we do "stretching", basically shadow lifting, that is so extreme it would make the D800 cry at how utterly sucky it is! :P). But it's still not infinite DR, because there is still noise. There is always noise. :P

1158
EOS Bodies / Re: Can Canon deliver a FF sensor that is class leading?
« on: June 03, 2014, 06:16:27 PM »
almost the full theoretical dr is actually usable
This is not true. The 6D does NOT have the "full theoretical DR", not even close.

Talking of correct quoting :-) ... but thanks for the explanation anyway, always great to read your in-depth posts. You did mis-understand me in this case though: the "usable dynamic range" was related to pattern noise, not to the theoretical limits of a Canon sensor if you glue it inside a Nikon camera body.

I think you may be misunderstanding what dynamic range is. Dynamic range is defining the "usable" range of signal. I think it becomes a bit redundant to say "usable dynamic range". The entire dynamic range is usable, because it's the range from the noise floor to the maximum signal...dynamic range implicitly excludes the "unusable" part of the signal. In an "ideal" system, one without noise, dynamic range would (theoretically) be infinite (The formula: 20 log(FWC/NoiseRMS) fails if NoiseRMS == 0...that's effectively a vertical and instantaneous asymptotic explosion to infinity. :P). That would mean that you have an infinite range of USABLE signal.

So, while I understand what your trying to say...I think it's a misnomer. Dynamic range is itself defining what's usable...so saying that you can only use part of what's usable... Redundant and inaccurate. And, incorrect. ;P

however the ADC bit depth limits it to 14, and downstream electronics (namely the ADC) introduce so much noise that it flattens the curve, rather than leaving it in it's linear state.

If you're interested in how the Magic Lantern people to improve the adc chain, see their forum (I posted the link above)


I've read some of their stuff. I haven't read it all. From what I have read, a lot of what they do benefits from Canon's downstream secondary amplifier.

It doesn't matter if the 6D ISO 100 noise doesn't have banding...the problem is that it still has a ton of noise.

Noise doesn't worry me that much as the 5d3/1dx/6d manage to make it look like film nose, so that elevates it way above my 60d (and as far as I see 5d2) concerning "destructive noise". My eye is not a technical instrument to measure snr.

I totally agree that the NATURE of noise can make it acceptable or not. Horizontal and vertical bands are horrid and completely unnnatural. The 6D is indeed a massive improvement there. The noise is quite aesthetically pleasing and random. Canon has certainly made strides here.

1159
EOS Bodies / Re: Can Canon deliver a FF sensor that is class leading?
« on: June 03, 2014, 03:00:51 PM »
And if Canon can do that, can it finally deliver a FF sensor that is also class leading? By class leading, I'm referring to noise control and DR

Canon already has a class-leading sensor, it's in the 6d: nearly zero (esp. vertical) banding (better than 1dx @iso100) means almost the full theoretical dr is actually usable. It has good dr @base iso (boosted +1/3ev by ML and =15ev with dual_iso) and superior dynamic range at high iso: http://sensorgen.info/CanonEOS_6D.html

This is not true. The 6D does NOT have the "full theoretical DR", not even close. The 6D has the same problem as every other Canon camera: A flattened DR curve at low ISO. Canon's DR tops out at around 12 stops of DR. Their sensors are more than capable of more than that, according to Roger Clark the sensor itself is actually probably capable of a little over 15 stops of DR natively (in analog space, before digitization), however the ADC bit depth limits it to 14, and downstream electronics (namely the ADC) introduce so much noise that it flattens the curve, rather than leaving it in it's linear state. It doesn't matter if the 6D ISO 100 noise doesn't have banding...the problem is that it still has a ton of noise. It has 26.8e- worth of read noise, which while less than the 1DX's 38.2e-, it also has a lower FWC, so it's dynamic range is roughly the same.

The problem isn't the sensor. Canon's sensors are very good. Canon's problem is their high frequency off-die ADC units housed in the DIGIC chips. They are just plain noisy.

Roger Clark has evaluated a lot of Canon sensors. His work finds the lowest noise level in the sensor itself, which would be intrinsic sensor noise, devoid of actual read noise. That's dark current noise in the sensor, along with whatever noise the per-pixel amplifiers might introduce. In Canon sensors, that noise level is around 2e-. In the case of the 6D, the sensor's analog dynamic range would be 20log(76606/2) dB, or 91.65 dB, which in terms of stops is 15.27. In the case of the 1D X, which also has ~2e- intrinsic sensor noise, the maximum possible dynamic range would be 93.1 dB or 15.52 stops. These levels aren't realizable due to the amount of read noise at ISO 100. If Canon can get their read noise under control, and get their ISO 100 noise levels down to 3-4e-, their dynamic range would be ~14-14.5 stops. Throw in a little bit of quantization error and PRNU and a 14-bit ADC, and Canon's DR jumps up to the level of Sony Exmors. Throw in a 16-bit ADC, and Canon should be able to achieve 14.5 stops of DR pretty easily. If they can lower their read noise levels even more, they could achieve well more than 15 stops of DR.

Their SENSORS are capable. The rest of their electronics are not. Canon's biggest problem is their approach of offloading the ADC into the DIGIC chip, and running them at very high frequency. Plus, their use of a downstream secondary amplifier doesn't help, but that only kicks in at higher ISO as far as I am aware.

1160
Third Party Manufacturers / Re: This Is How You Sell a Used Lens !!
« on: June 03, 2014, 02:52:36 PM »
Now DAT  guy is a SALESMAN! BOOM! EPIC! (And where DO you get Spaghetti-Ohs?)

1161
EOS Bodies / Re: New Full Frame Camera in Testing? [CR1]
« on: June 02, 2014, 11:31:43 PM »
The most anal people I know about image colour are flower photographers and ceramicists, ever photograph a red flower and it not look anything like the flower did? Try deep blue, purple, and mauve flowers, they are a very difficult to get accurate and you have to use a camera profile specifically for the light you shot in.

A lot of trouble with flowers is even more that people seem to stick to sRGB which makes many flowers impossible to show correctly. A wide gamut monitor will give you a much better chance (of course it's true that the WB and profiles and all can still mess with things).

True, sRGB is too endemic. It's really time we started moving towards larger gamuts. Even AdobeRGB isn't quite good enough, as most of the gain with AdobeRGB is in the greens. The deep reds and blues and violets, where a lot of flower color resides, don't really change much with AdobeRGB. ProPhotoRGB may not be the best either, as its extent is even beyond that of human perception, but it's still got the ability to map almost every color at the richest saturation the human eye can discern.

Sadly, even 10-bit screens with 14- and 16-bit 3D LUTs are still not quite good enough at showing reds. I have these Peonies that are just about to burst into color. I've tried photographing them in years past, and I've never been able to get the reds and pinks to come out right...they clip and there is little tonality. Bleh. It's such a pain. My roses have a similar problem, however most of those have a deeper red that actually does fall into gamut for AdobeRGB.

No question, though...rich saturated color, particularly in the non-greens, can be a real problem.

1162
Oh, no. Jrista what happened?? About a year ago, you had finally gotten down with the concept of normalization, but now you are back to your old game of normalization doesn't make sense again. :(


Well, it's no surprise that you buy into DXO's bull. There are two values on DXO's site for DR. One is a measure, as in something actually MEASURED from a REAL RAW file. The other is an EXTRAPOLATION. It isn't even a real extrapolation, it is just a number spit out by a simple mathematical formula...they don't actually even do what they say they are doing.

And since the simple formula is simply it actually gives worse results if anything compared to fancy techniques, not better.

Quote
The first of these is Screen DR. Screen DR is the ONLY actual "measure" of dynamic range that DXO does. It is the SINGLE and SOLE value for DR that is actually based on the actual RAW data. In the case of the D800....do you know what Screen DR is? (My guess is not.)

The other of these is Print DR. Print DR is supposedly the dynamic range "taken" from a downsampled image. The image size is an 8x12 "print", or do DXO's charts say. As it actually happens to be, and this is even according to DXO themselves...Print DR is not a measure at all. It isn't a measurement taken from an actually downsampled image. You know what it is? It is an extremely simple MATHEMATICAL EXTRAPOLATION based on...what? Oh, yup...the only actual TRUE MEASURE of dynamic range that DXO has: Screen DR. Print DR is simply the formula DR+ log2 SQRT(N/N0). DR is ScreenDR. N is the actual image size, N0 is the supposed downsampled size. The formula is rigged to guarantee that "Print DR" is higher than Screen DR...not even equal to, always higher. And, as it so happens, potentially 100% unrelated to reality, since it is not actually measured.

Oh brother. It is not rigged! Why are you back to calling normalization rigged again???? Do you realize that 90% of modern tech and science wouldn't work out if what you say was true?

Quote
DXO doesn't even have the GUTS to ACTUALLY downsample real images and actually measure the dynamic range from those downsampled images. They just run a mathematical forumla against Screen DR and ASSUME that the dynamic range of an image, IF they had downsampled it, wouold be the same as what that mathematical value says it should be.

Print DR is about as bogus as "camera measurement 'science'" can possibly get. It's a joke. It's a lie. It's bullshit. The D800 does not have 14.4 stops of DR, as DXO's Print DR would indicate. The Screen DR measure of the D800? Oh, yeah...it's LESS than 14 stops, as one would expect with a 14-bit output. It's 13.2 stops, over ONE FULL STOP less than Print DR. The D600? Says Print DR 14.2, but Screen DR is 13.4. D610? Print DR 14.36, but Screen DR 13.55. D5300? Print DR 13.8, but Screen DR 13. A7? Print DR 14, but Screen DR 13.2. A7s? Print DR 14 but Screen DR 13. NOT ONE SINGLE SENSOR with 14-bit ADC output has EVER actually MEASURED more than 14 stops of dynamic range. That's because it's impossible for a 14-bit ADC to put put enough information to allow for more than 14 stops of dynamic range. There simply isn't enough room in the bit space to contain enough information to allow for more than 14 stops..not even 0.1 more stops. Every stop is a doubling. Just as every bit is a doubling. Bits and stops, in this context, are interchangeable terms. In the first bit you have two values. With the second bit, your "dynamic range" of number space doubles...you now have FOUR values. Third bit, eight values. Fourth bit, sixteen values. Fifth bit, thirty two values. To begin using numeric space beyond what the 14th bit allows, which would be necessary to start using up some of the 15th stop of dynamic range, you need at least 15 bits of information. It's theoretically, technologically, and logically impossible for any camera that uses a 14-bit ADC to have more than 14 stops of dynamic range.

no, no, no and no

comparing noise at different energy scales as if the scales were the same is what would be totally bogus!

Quote
Here is another fact about dynamic range. Dynamic range, as most photographers think about it these days, is the number of stops of editing latitude you have. While it also has connotations to the amount of noise in an image, the biggest thing that photographers think about when it comes to dynamic range is: How many stops can I lift this image? We get editing latitude by editing RAW images. RAW. Not downsampled TIFFs or JPEGs or any other format. RAW images. How do we edit RAW images? Well...as RAW images. There IS NO DOWNSAMPLING when we edit a RAW image. Even if there was...who says that we are all going to downsample our images to an 8x12" print size (3600x2400 pixels, or 8.6mp)? We edit RAW images at full size. It's the only possible way to edit a RAW image...otherwise, it simply wouldn't be RAW, it would be the output of downsampling a RAW to a smaller file size...which probably means TIFF. Have you ever tried to push the exposure of a TIFF image around the same way you push a RAW file around? You don't even get remotely close to the kind of shadow lifting or highlight recovery capabilities editing a TIFF as you do a RAW. Not even remotely close. And the editing latitude of JPEG? HAH! Don't even make me say it.

Therefor, the ONLY valid measure of dynamic range is the DIRECT measure, the measure from a RAW file itself, at original size, in the exact same form that photographers are going to be editing themselves. Screen DR is the sole valid measure of dynamic range from DXO. Print DR is 100% bogus, misleading, fake.

Please go the library and check out a book on normalization and mathematics.



Quote
Their sensors may be good, but how Sony themselves are using their sensors is crap.

sometimes they do do some annoying things, that is true at least

I don't disagree with you. But your missing my point. I tried to be clear about what I'm referring to. Noise is one thing. And noise in an image doesn't just come from read noise, it's the photon shot noise in the signal as well. And YES, downsampling normalizes results. I'm not debating that.

I'm specifically debating the notion that you actually have 14.4 stops worth of EDITING LATITUDE with a D800, or 14.2 with a D800, etc. Because that's what everyone things about. That's what everyone is referring to when they bring up the DR difference. It's not a bad thing, and there is no question that Sony Exmor has more DR than a Canon sensor. The problem I have is the missleading notion that DXO's Print DR "results" have created in the community.

We don't push the exposure of downsampled TIFF files around, so it makes no sense to refer to 14.4 stops of DR in the context of, say, discussing the benefit of DR when working with landscapes. It really doesn't make any sense to refer to 14.4 stops of 8mp image DR when discussing actual photographic editing in ANY context EXCEPT when directly comparing cameras, and then, only in a very neat and tidy context...such as when your actually on the DXO web site. In all other contexts, the only legitimate measure is that taken directly from the RAW...from the actual image we actually work with out in the actual world. In that context...the D800 has 13.2 stops of DR.

Does that make sense? As far as I'm concerned: Comparison Shmarison! :P I care about actual real-world editing latitude. Mathematically extrapolated imaginary downsampled fake "measurements" don't tell me jack about what I am ACTUALLY going to be able to do FOR REAL. Screen DR? It tells me exactly what I want to know. It tells every photographer what they want to know: How much can I lift my landscape photos? Print DR is lying...it tells you you could lift more than you actually can, because you don't edit RAW images downsampled, JPEGs don't even remotely cut it, and TIFF images, because they are RGB triples rather than RAW indepentent digital signal values, you can't lift the shadows nor compress the highlights the same way...not without significant artifacts after a push or pull of a couple stops. (i.e. you may be able to lift shadows by two, maybe three stops without artifacts with a TIFF at "14.4" stops DR, but you could easily lift a Nikon D800 RAW by six stops at 13.2 stops DR.)

So I don't disagree. I agree. I am just working within a different context. The context I believe most photographers approach the subject of DR (based on the things they reference when they approach it.) To you and me, dynamic range means signal cleanliness across the entire band. To most everyone else, it means: How much can I lift without banding in the shadows? :P

1163
EOS Bodies - For Stills / Re: The Answer to Everyone's Complaints
« on: June 02, 2014, 07:11:09 PM »
Been thinking about this for awhile and decided I'd act on it.

If you are unhappy with Canon's perceived failure to offer (choose any of the following)  high megapixel body, mirrorless full frame camera, more dynamic range, shadow detail) I propose the following:

Send me a check or Pay Pal me $2,000 as a deposit, telling me what feature you want improved.

I will deposit it in a special savings account and will not spend the money, except under the conditions I will list below.

After you and 50,000 of your like-minded friends have sent me these checks, I will forward the money to Canon as a deposit for them to use in developing the camera feature you want. It is no guarantee they will comply, but I suspect the fact that 50,000 people have demonstrated a willingness to place a $2,000 down payment on the camera of their choice, will indeed get Canon's attention.

Now, there is one condition I am placing on this offer. Anyone who starts or participates in a thread whining about how Canon isn't paying attention to them or is about to go bankrupt because their personal needs are not being attended to, I will deduct 10% of your deposit as an trolling tax.

Basically, you must put your money where your mouth is. This has the added advantage of demonstrating whether or not there is a demand for the features you find so significant.

I am willing to add accounts for other features, in case I'm missing any. By the way, it is 50,000 x $2,000 for each feature. As some of these features are contradictory, you cannot ask for more than one for each deposit.

If, after two years, any account has not reached the threshhold of 50,000 participants, I will refund the money minus a small handling fee.

The money will be deposited in The Bank of Nigeria: "We guarantee your money will still be there when you go looking for it!"

1164
Ah, you guys just don't like a good debate! :P Sometimes debate is healthy...

Although I'll grant it's less debate and more "Let's beat that last little bit of brain matter and bone of the dead horse over there...AGAIN" on these forums...but hey, that isn't my fault. :o Some people are just too thick to let the facts soak into their skulls.  I'm just responding to the call:



 ;D

1165
...
The first of these is Screen DR. Screen DR is the ONLY actual "measure" of dynamic range that DXO does. It is the SINGLE and SOLE value for DR that is actually based on the actual RAW data. In the case of the D800....do you know what Screen DR is? (My guess is not.)

Curious then that someone else has slurped up DxO's data and says that the DR for the D800 is 14.0.

Quote
That's because it's impossible for a 14-bit ADC to put put enough information to allow for more than 14 stops of dynamic range. There simply isn't enough room in the bit space to contain enough information to allow for more than 14 stops..not even 0.1 more stops. Every stop is a doubling. Just as every bit is a doubling.

Assuming the relationship is linear.

Quote
It doesn't matter what Sony does in their BONZ X chip. The sensor output is 14-bit RAW. The only thing their BIONZ chip can do is...the same thing YOU do.

Unless they've come up with a new algorithm to map sensor output.

Or maybe they've decided that DR is like ISO and can be used for marketing joy. At this point it's all speculation because nobody has one of their cameras to test.

It doesn't matter what they do in the middle. The FILE you WORK WITH is 14-bit. All that BIONZ does is do some post processing. That doesn't change the dynamic range, it only changes the contrast within the original dynamic range. That's it. There is no magic here, nothing special. A 14-bit file has enough numeric space for 14 stops of dynamic range. Anything you do to the data, such as apply a non-linear curve, simply COMPRESSES the information contained within those original 14 stops to SOMETHING LESS!! There is no ex nihilo here.

It's no different than say upconverting from sRGB to AdobeRGB. There is no value to doing that, since you already lost the original color fidelity when you converted to sRGB in the first place. You cannot restore colors in the AdobeRGB space once they are lost...they are gone forever. It's the same thing with bit depth and stops of dynamic range.

The only way Sony can literally achieve 15.3 stops of dynamic range is if their sensor ADC units are 16 bit, the processing pipeline is 16-bit, AND the output RAW file is 16-bit. And I mean a FULL 16-bit, not some half-witted 13+3-bit lossy compressed RAW file, as that would just decimate the full tonal range. I mean a full, uncompressed (or at least lossLESSly compressed) 16-bit RAW file. There is no other way to achieve more than 15 stops of dynamic range...you have to have the numeric space to represent those stops.

1166
Quote
Sony's OWN SITE says the Sensor output is 14-bit. The sensor is an Exmor. Exmor uses CP-ADC ON-DIE. The last output of the sensor is FROM the ADC.

Therefor...the A7s IS 14-BIT!
...

Yup. As you say, Sony show they've got a 14bit sensor output delivering 15.3 stops of DR. Interesting.

I can see you and neuro and a whole host of other web folks getting ready to point out how it is impossible and when DxO measure it and say they've got 15.3 stops of DR, DxO will be telling lies and faking it and then someone will go out and do real world tests that also corroborates with it and you'll all still be saying that it is a lie.

I want to see what they've got to show before I denounce them. After all, they're already getting more than 14 stops of DR (DxO says 14.8?) with the Nikon D800 so why should another .5 stop of DR be unreasonable? Would be interesting to see what a real world test of the DR of those cameras turned out, kind of aligned with what the Zacutto(?) folks did with various cameras for video.

Obviously it isn't a linear conversion (even now) but I'm kind of curious to know what's going on.

A curve of some kind is pretty obvious, whether it is a gamma curve or something else ...

Well, it's no surprise that you buy into DXO's bull. There are two values on DXO's site for DR. One is a measure, as in something actually MEASURED from a REAL RAW file. The other is an EXTRAPOLATION. It isn't even a real extrapolation, it is just a number spit out by a simple mathematical formula...they don't actually even do what they say they are doing.

The first of these is Screen DR. Screen DR is the ONLY actual "measure" of dynamic range that DXO does. It is the SINGLE and SOLE value for DR that is actually based on the actual RAW data. In the case of the D800....do you know what Screen DR is? (My guess is not.)

The other of these is Print DR. Print DR is supposedly the dynamic range "taken" from a downsampled image. The image size is an 8x12 "print", or do DXO's charts say. As it actually happens to be, and this is even according to DXO themselves...Print DR is not a measure at all. It isn't a measurement taken from an actually downsampled image. You know what it is? It is an extremely simple MATHEMATICAL EXTRAPOLATION based on...what? Oh, yup...the only actual TRUE MEASURE of dynamic range that DXO has: Screen DR. Print DR is simply the formula DR+ log2 SQRT(N/N0). DR is ScreenDR. N is the actual image size, N0 is the supposed downsampled size. The formula is rigged to guarantee that "Print DR" is higher than Screen DR...not even equal to, always higher. And, as it so happens, potentially 100% unrelated to reality, since it is not actually measured.

DXO doesn't even have the GUTS to ACTUALLY downsample real images and actually measure the dynamic range from those downsampled images. They just run a mathematical forumla against Screen DR and ASSUME that the dynamic range of an image, IF they had downsampled it, wouold be the same as what that mathematical value says it should be.

Print DR is about as bogus as "camera measurement 'science'" can possibly get. It's a joke. It's a lie. It's bullshit. The D800 does not have 14.4 stops of DR, as DXO's Print DR would indicate. The Screen DR measure of the D800? Oh, yeah...it's LESS than 14 stops, as one would expect with a 14-bit output. It's 13.2 stops, over ONE FULL STOP less than Print DR. The D600? Says Print DR 14.2, but Screen DR is 13.4. D610? Print DR 14.36, but Screen DR 13.55. D5300? Print DR 13.8, but Screen DR 13. A7? Print DR 14, but Screen DR 13.2. A7s? Print DR 14 but Screen DR 13. NOT ONE SINGLE SENSOR with 14-bit ADC output has EVER actually MEASURED more than 14 stops of dynamic range. That's because it's impossible for a 14-bit ADC to put put enough information to allow for more than 14 stops of dynamic range. There simply isn't enough room in the bit space to contain enough information to allow for more than 14 stops..not even 0.1 more stops. Every stop is a doubling. Just as every bit is a doubling. Bits and stops, in this context, are interchangeable terms. In the first bit you have two values. With the second bit, your "dynamic range" of number space doubles...you now have FOUR values. Third bit, eight values. Fourth bit, sixteen values. Fifth bit, thirty two values. To begin using numeric space beyond what the 14th bit allows, which would be necessary to start using up some of the 15th stop of dynamic range, you need at least 15 bits of information. It's theoretically, technologically, and logically impossible for any camera that uses a 14-bit ADC to have more than 14 stops of dynamic range.

Here is another fact about dynamic range. Dynamic range, as most photographers think about it these days, is the number of stops of editing latitude you have. While it also has connotations to the amount of noise in an image, the biggest thing that photographers think about when it comes to dynamic range is: How many stops can I lift this image? We get editing latitude by editing RAW images. RAW. Not downsampled TIFFs or JPEGs or any other format. RAW images. How do we edit RAW images? Well...as RAW images. There IS NO DOWNSAMPLING when we edit a RAW image. Even if there was...who says that we are all going to downsample our images to an 8x12" print size (3600x2400 pixels, or 8.6mp)? We edit RAW images at full size. It's the only possible way to edit a RAW image...otherwise, it simply wouldn't be RAW, it would be the output of downsampling a RAW to a smaller file size...which probably means TIFF. Have you ever tried to push the exposure of a TIFF image around the same way you push a RAW file around? You don't even get remotely close to the kind of shadow lifting or highlight recovery capabilities editing a TIFF as you do a RAW. Not even remotely close. And the editing latitude of JPEG? HAH! Don't even make me say it.

Therefor, the ONLY valid measure of dynamic range is the DIRECT measure, the measure from a RAW file itself, at original size, in the exact same form that photographers are going to be editing themselves. Screen DR is the sole valid measure of dynamic range from DXO. Print DR is 100% bogus, misleading, fake.

It doesn't matter what Sony does in their BONZ X chip. The sensor output is 14-bit RAW. The only thing their BIONZ chip can do is...the same thing YOU do. They can lift shadows, and compress highlights. They can shift exposure around and reduce noise by applying detail-softening noise reduction algorithms. But then, well then you don't actually have a RAW file anymore. You have a camera-modified file. With Sony's propensity for using a lossy compression algorithm in their RAWs, you don't even get full 14-bit precision data per pixel, and that fact has shown in many cases when people go to edit their Sony RAWs in post. The compression artifacts can be extreme. I find it simply pathetic that Sony, with all this horsepower under their thumb, would completely undermine it all by storing their RAW images in a lossy compressed format. It completely invalidates the power of their sensors, and speaks to the fact that Sony is probably just as schizophrenic internally as Nikon is. That will lead to inconsistent products and product lines, poor product cohesion, lackluster design for OTHER aspects of their cameras beyond the sensor, etc. Were already seeing many of these problems with Sony cameras. Their sensors may be good, but how Sony themselves are using their sensors is crap.

1167
My comments are based on my (imperfect) memory of science podcasts and other science journalism I've encountered in the last few years.  If you have contradictory info I'd love to see a reference.

Regarding eye-witness accounts...the reason they are unreliable is people are unobservant. There are some individuals who are exceptionally observant, and can recall a scene, such as a crime, in extensive detail.

My understanding is that new research has shown this to be wrong.  There are a few "savant" types who have very precise/correct memory function, but for "neurotypical" (i.e. "normal") people, this is not so.

I'm not saying everyone can be a savant. I'm saying everyone can learn how to WORK their memory to improve it. I did it...I used to have the same old poor memory that everyone had, I forgot stuff all the time, couldn't remember accurately. By thinking about, exercising, and processing sensory input more actively, I can intentionally bring up other memories that I want associated to the new ones I'm creating. Purposely recalling memories in certain ways and reviewing after creating them has helped me strengthen those memories, improving my ability to accurately recall the original event, be it sight, sound, smell, touch, taste or all of the above.

Whatever current research shows, memory is NOT simply some passive process we have absolutely no control over. It's also an active process that we CAN control, and we can improve our memory if we choose to...either only of specific events of importance, or we can train ourselves to process input in a certain way such that most input is more adequately remembered and strongly associated.

Quote
I believe the brain only fills in information if it isn't readily accessible. I do believe that for the most part, when we see something, the entirety of what we see is recorded.

Again, my understanding is that recent research shows that the adage "seeing is believing" has it backwards: it should be "believing is seeing."  The brain does not record raw image info at all, but constructs a reality that incorporates visual data with existing beliefs and expectations.  It's that highly-processed "reality" that's recorded.  As an example, back in 2004 there was that video tape of a purported Ivory-Billed Woodpecker.  Subsequent analysis showed that it was almost certainly the rather common pileated woodpecker.  The "eyewitnesses," however, recall seeing detail that would clearly distinguish it as an IBW.  Even if it was a pileated, those witness may truthfully and genuinely believe they saw those distinguishing characteristics.

I don't think any of that contradicts the notion that our brains store much or most of everything that goes into them. I don't deny that our beliefs and desires can color HOW we remember...as they could control what we recall. Remember, memory is often about association. If the guy watching the woodpecker was vividly remembering an IBW at the time (would have been an amazing find, for sure! I really hope they aren't extinct, but... :'(), that wouldn't necessarily change the new memories being created, but it could overpower the new memories with the associations to old memories of IBW. Upon recall...you aren't just recalling the new memories, but things associated with it as well. What you finally "remember" could certainly be colored by your desired, causing someone to misremember. Good memory is not necessarily good recall, and it certainly doesn't overpower an individual's desires for something to be true. All that gets into a level of complexity about our our brains work that goes well beyond any courses on the subject I've ever taken.

BTW, I am  not talking about savants who have perfect memory. Eidetic memories or whatever you want to call them, that's a different thing than what I'm talking about. Eidetic memories are automatic, it's more how those individuals brains work, maybe a higher and more cohesive level of processing than normal individuals. That doesn't change the fact that you CAN actively work with your memory to improve it, considerably. I'm not as good at it these days as I used to...severe chronic insomnia have stolen a lot of abilities like that from me, but when I was younger, I used to have an exceptional memory. I remembered small details about everything because I was always working and reviewing the information going in. Before I took that class, my memory was pretty average, after and still largely since, it's been better than average to truly excellent.

That has to do with memory creation itself, though...it doesn't mean my memories can't be colored by prior experiences for desires. I think it lessens the chance of improper recall, but it's still possible to overpower a new memory with associations to old ones, and over time, what is recalled may not be 100% accurate (again, not talking about eidetic memories here, still just normal memory.) There have been cases of obsessive-compulsive individuals having particularly exceptional memory, on the level of supposed eidetics and in some respects better. For the very very rare individual, memories become their obsession, and because it's an obsession, every memory is fully explored, strengthened and associated to a degree well beyond normal. Recall is very fast, and the details can be very vivid. It isn't just image-based either, all sensory input can be remembered this way (sounds, smells, etc.) With such strong associations and synaptic cleft strengthening, such an individuals memories are effectively permanent as well. The difference would be the obsessive-compulsive chooses what memories to obsesse over...so their recall isn't necessarily as complete as an eidetic (who's memory for imagery is more automatic.)

1168
Let's turn this around.  Can you provide a reference showing the a7S has a 16-bit ADC?

Nope. The specs for the a7s only quote the bit depth for the image files, not the ADC.

Quote
Since they are claiming >15 stops of DR

Note that at present the claim for 15.3 stops of DR comes from a 3rd party ... even I'm dubious on that. I'll wait and see what Sony says and more importantly, what DxO can measure.

Quote
it's certainly in their best interest to not make much of the fact that they're using a 14-bit ADC which cannot deliver the actual DR they claim, meaning they're merely cooking the RAW file to include fabricated data.

Or maybe they're using a spreading function (i.e applying a curve to the sensor feed) rather than doing a linear conversion?

Did you completely miss the post where I linked directly to Sony's site that SHOWS the sensor output (which CONTAINS the ADC) is 14-bit? How convenient...that you read my first post, and just magically didn't happen to see my second post. Sony's OWN SITE says the Sensor output is 14-bit. The sensor is an Exmor. Exmor uses CP-ADC ON-DIE. The last output of the sensor is FROM the ADC.

Therefor...the A7s IS 14-BIT! I love it how you DEMAND I PROVE things to you, then simply ignore the FACTS when I smack you upside the face with them.

There is absolutely ZERO question about it. The facts are the facts. The A7s is still "just" an Exmor, and Exmor's use 14-bit ADC. Here, I'll smack you upside the face with them again:

From the horses mouth: http://discover.store.sony.com/sony-technology-services-apps-NFC/tech_imaging.html#BIONZ

Quote
16-bit image processing and 14-bit RAW output
16-bit image processing and 14-bit RAW output help preserve maximum detail and produce images of the highest quality with rich tonal gradations. The 14-bit RAW (Sony ARW) format ensures optimal quality for later image adjustment (via Image Data Converter or other software).



1169
If Sony had come out with a sensor with on-die 16-bit ADCs, that would have been far, far bigger news than the fact that it can do ISO 409k. No one really cares about ISO 409k. The noise levels at that ISO are a simple matter of physics when it comes to stills.

When it comes to the A7s video performance, their DSP, BIONZ X, is the bigger news, since it's doing a significant amount of processing on the RAW signal to reduce noise at ultra high ISO settings. The BIONZ X image processor does 16-bit IMAGE PROCESSING, however the sensor output is 14-bit, and the output of the image processing is ALSO 14-bit. There is a page on Sony's site somewhere that describes this, soon as I find it I'll link it.

The BIONZ X processor is the same basic thing as Canon's DIGIC and Nikon's EXPEED. It's the in-camera DSP. Canon's DIGIC 6 has a lot of similar capabilities to Sony's BIONZ X. They both do advanced noise reduction for very clean high ISO JPEG and video output. They both do high quality detail enhancement as well. I don't believe Canon's DIGIC 6 does 16-bit processing, it's still 14-bit as far as I know. The use of 16-bit processing can help maintain precision throughout the processing pipeline, however since the sensor output is 14-bit, you can never actually increase the quality of the information you start with. That would be like saying that when you upscale an image in photoshop, you "extracted" more detail out of the original image. No, you don't extract detail when you upscale...you FABRICATE more information when you upscale.

Same deal with BIONZ X...during processing, having a higher bit depth reduces the impact of errors (especially if any of that processing is floating point), however it cannot create more out of something you didn't have to start with. That is evident by the fact that Sony is still outputting a 14-bit RAW image, instead of a 16-bit RAW image, from their BIONZ X processor.

UPDATE:

From the horses mouth: http://discover.store.sony.com/sony-technology-services-apps-NFC/tech_imaging.html#BIONZ

Quote
16-bit image processing and 14-bit RAW output
16-bit image processing and 14-bit RAW output help preserve maximum detail and produce images of the highest quality with rich tonal gradations. The 14-bit RAW (Sony ARW) format ensures optimal quality for later image adjustment (via Image Data Converter or other software).



Higher precision processing, but still 14-bit RAW. The fact that the raw sensor output is 14-bit means that the dynamic range of the system cannot exceed 14 bits. The use of 16-bits during processing increases the working space, so when Sony generates a JPEG or video, it can lift shadows and compress highlights with more precision and less error. I suspect their "15.3 stops of dynamic range" is really referring to the useful working space within the 16-bit processing space of BIONZ X. Simple fact of the matter, though, is that when it comes to RAW...it's RAW. Your dynamic range is limited by the bit depth of the ADC. Since the ADC is still 14-bit, and ADC occurs on the CMOS image sensor PRIOR to processing by BIONZ X, then any processing Sony does in-camera can do no more, really, than what you could do with Lightroom yourself.

1170
...

I'm still waiting for your reply to me asking for a reference (you know, a URL) to something that supports your claim of the Sony a7s only having a 14bit ADC ...

B&H Photo's product page, under specifications:

http://www.bhphotovideo.com/c/product/1044728-REG/sony_ilce7s_b_alpha_a7s_mirrorless_digital.html

It took me about 3 seconds to find that. I just searched for "Sony A7s Bit Depth", and that was one of the first five links (the rest, for some reason, were all about the A77...)

It states:


Pages: 1 ... 76 77 [78] 79 80 ... 309