Just Published - Mirrorless wars

[ErikKaffehr]

Hi,

This is a good article on the issue: https://www.strollswithmydog.com/how-many-bits-to-fully-encode-my-image/

Best regards
Erik

Is this true?  What about Nikon's trick of scaling the amount of photons/electrons in the A/D process?  I thought it was the number of steps, not the range, that defined the bits per pixel required.

[Rory]

Hi,

This is a good article on the issue: https://www.strollswithmydog.com/how-many-bits-to-fully-encode-my-image/

Best regards
Erik

That is a very good article Erik - thanks.  The part at the end called "In fact it's even less" is what I was referring to.  Also, the article describes how many bits to "fully encode my image".  However, fully encoding and encoding what the human eye can differentiate are two different things.  All I'm saying is that the amount of bits required to encode do not have to correlate to the DR EV amount.  I'm guessing we are saying the same thing in different ways.

[ErikKaffehr]

Hi,

Here is some more information from the same author: https://www.strollswithmydog.com/difference-between-data-and-information/

But, here are two easy explanations:

Let's assume that full well capacity (FWC) is around 64 000 electron charges, let's assume the readout noise is 4 electron charges. In that case we need a sixtin bit number to account for each electron of the full well, but the last two bits will just contain the readout noise.

Now, let's assume the we have FWC = 64000 and let's assume that we expose a mid tone three stops below saturation. The electron count in the well is in that case 64000 / 2^3 -> 64000 / 8 -> 8000

Photon arrival is Poisson distributed, that means that standard of deviation is sqrt(electron count), that sqrt(8000) -> 89. That would mean that 65% of the pixels would have electron counts between 7911 and 8089.

In binary, it would be:
0001111011100111‬
0001111110011001‬

You would see that the last 8 digits change, although the input is exactly the same!

This is a representation of a pretty dark patch on my P45+, with 16 bits claim:


And the same on the Sony A7rII:


Keep in mind that both plots represent a single tone. So that would really be a spike in a single channels. If no photon statistics would be around!

Best regards
Erik

That is a very good article Erik - thanks.  The part at the end called "In fact it's even less" is what I was referring to.  Also, the article describes how many bits to "fully encode my image".  However, fully encoding and encoding what the human eye can differentiate are two different things.  All I'm saying is that the amount of bits required to encode do not have to correlate to the DR EV amount.  I'm guessing we are saying the same thing in different ways.

[siba]

Ahhh. Of course. How silly of me. Now it's all obvious. I thought it was a bit more complicated.

[michaelsh]

...
This is a representation of a pretty dark patch on my P45+, with 16 bits claim:
...

Best regards
Erik

In short then the 'pretty dark patch' would stay a 'pretty dark patch', regardless of 16 bits or not?

[apindrans]

A very interesting assessment of the mirror-less market.
Your earlier review of the Sony A7r got me really excited about the the virtuosity of the camera. However, I shoot with a venerable Canon 40D and have quite a bit of an investment in Canon (L series) and Sigma lenses and that probably puts me into the pro-consumer equipment owner. Consequently the cost of changing to an equivalent Sony system would be quite prohibitive.

Unlike computer software, which has some backward compatibility to older versions, when camera vendors release new technology only new entrants to the market or professional photographers can justify the change.

[Rob C]

A very interesting assessment of the mirror-less market.
Your earlier review of the Sony A7r got me really excited about the the virtuosity of the camera. However, I shoot with a venerable Canon 40D and have quite a bit of an investment in Canon (L series) and Sigma lenses and that probably puts me into the pro-consumer equipment owner. Consequently the cost of changing to an equivalent Sony system would be quite prohibitive.

Unlike computer software, which has some backward compatibility to older versions, when camera vendors release new technology only new entrants to the market or professional photographers can justify the change.

And certainly not all pro photographers, either!

Way back in film days it was understood that the pro market was but the tip of the iceberg; there were always the amateurs who aspired to better cameras as there were pros whose work/tax relationship made new stuff a painless investment, usually in the sense of doubling up bodies just in case and there would sometimes be that case, even in those times of relatively low tech.

There was also a sense that the top cameras were part of the maker's advertising effort, the shiny fairy atop the tree.

That said, I do not recall any great diversity of lenses (as in quality) within a maker's range. You bought whatever focal length, and it was as good as it got; okay, an f2 or an f2.8, but what did it matter, most of the time; f6.3 was about as open as most work was shot at. When Nikon began putting out its E range, which was a cheaper option of the normal offering. I think that was a mistake, sowing both confusion and a sense that standards were being let go... that said, I never heard of people having to go through a range of copies to arrive at a so-called keeper.

The conclusions one is driven to draw are not at all flattering for the companies we know today.

Rob

[adri]

For me, the mirrorless competition (I am not keen on the word "war") will make a difference, as I am about to jump from one brand to whatever other brand that will totally capture my needs.

The megapixel competition is still going on simultaneously with the mirrorless competition. There is talk about more affordable 100MP cameras from Hasselblad and Fuji. It will be interesting to see if Sigma, Tamron, et. al. will offer lenses that that will perform well on 100MP cameras. Or if some of their lenses already do resolve at that level.

Kevin didn't mention the differences between what e.g. Fuji is offering now and the Rangefinder model that has been rumored.

Kevin neither mentioned minor players like Sigma (they apparently will announce a new camera this year; I don't believe that future development/improvement of Foveon technology has reached its end) and perhaps Ricoh/Pentax (are they still in this medium format race?).

I think it's important to be inclusive as much as possible and include the minor players.

This will be an interesting year indeed, and I wonder how much will be revealed before and at Photokina.
 

[Rob C]

For me, the mirrorless competition (I am not keen on the word "war") will make a difference, as I am about to jump from one brand to whatever other brand that will totally capture my needs.

The megapixel competition is still going on simultaneously with the mirrorless competition. There is talk about more affordable 100MP cameras from Hasselblad and Fuji. It will be interesting to see if Sigma, Tamron, et. al. will offer lenses that that will perform well on 100MP cameras. Or if some of their lenses already do resolve at that level.

Kevin didn't mention the differences between what e.g. Fuji is offering now and the Rangefinder model that has been rumored.

Kevin neither mentioned minor players like Sigma (they apparently will announce a new camera this year; I don't believe that future development/improvement of Foveon technology has reached its end) and perhaps Ricoh/Pentax (are they still in this medium format race?).

I think it's important to be inclusive as much as possible and include the minor players.

This will be an interesting year indeed, and I wonder how much will be revealed before and at Photokina.

Don't lose sight of the fact, though, that LuLa doesn't pretend to be a "gear review" site; what it does in that direction is show and discuss stuff that the owners themselves enjoy. Which seems fair enough to me.

Maybe it gets a bit overly sentimental with some brands, but why not if they are the objects of the desires?

(Just yesterday, for the first time, I read the fine print on the T.O.P site: I hadn't realised that a blog was supposed to be run on different psychological lines to a forum: I had fondly imagined that threads were there for the discussing. How boring to realise I was mistaken, that all you're supposed to do is follow the thin yellow/fawn line! I've spent much of my life trying to avoid following officially drawn lines. If I was meant to follow them, then I'd be a tram or, at best, a train.

At least Leicaphilia is actively open to fresh thought, even if, as with T.O.P., it is slow in moving forward because of the need to vet input and the unavoidable delays due to that process.

That's one thing at least where LuLa scores hands down: it allows for instant interaction. In this age, who'd expect less?)

[Guillermo Luijk]

As for the 14 bit vs. 16-bit, my understanding is that outside of massively more DR coming (wouldn't that be awesome compared to more MP), more bits here is mostly marketing than useful but I'm open to a better understanding of all this.

The Nikon D850 at ISO64 has a measured engineering DR (DxOMark) of 13,55EV which can STILL but hardly be encoded in 14-bit RAW files. Should this sensor have just half extra stop of DR and more than 14 bits would become necessary. What does this means? 16-bit linear encodings are just knocking at the door and we'll see them as soon as an enhanced DR sensor appears on the market, not for marketing reasons (as Canon did with the 40D and its fake 14 bits), but as an engineering requirement.

This is the Pentax K5 with Sony Exmor sensor, the first sensor ever really needing 14 bits:


What if this sensor would just have a 12-bit encoding? posterization (the simulation was done by decimating from 14-bit to 12-bit a RAW file prior to debayering):


That is what would happen in the deep shadows of any sensor with insufficient encoding bits.

Regards

[Guillermo Luijk]

Regarding the mirrorless FF war, I wonder what could today be the penetration of Sony in the FF market. In other thread Bernard estimated an almost negligble 1%. With sustained sales from Sony since the introduction of the A7 five years ago I think that figure is strongly pesimistic.

I don't have the needed data, but with some assumptions:
- 50% of FF users bought a new FF body from 2014-2018 and those who did only bought one
- Users are only Canikon or Sony never mixed
- Total market is stable
- Yearly body sales remain constant
...

I cannot reach in any way such a bad figure as 1%. In the following estimation I get 12% of Sony users in the FF market.




Here is the Excel to play with, green cells can be adjusted. What are your thoughts and how could this influence in the final winner of the war?. Kevin seems to see Sony as very well positioned for being the first in having an almost complete mirrorless FF system, but I am skeptical about this seeing current FF market share.

http://www.guillermoluijk.com/misc/mirrorless.xlsx


Regards

[digitaldog]

16-bit linear encodings are just knocking at the door and we'll see them as soon as an enhanced DR sensor appears on the market, not for marketing reasons (as Canon did with the 40D and its fake 14 bits), but as an engineering requirement.

Thanks! Is it 'difficult' or just more expensive to implement this?

[ErikKaffehr]

Hi,

It is not difficult, just slows down things, probably. I would guess that most CMOS sensors use ramp type converters, adding a bit means doubling conversion time. I would think...

But, yes, DR is just below 14 bits now. So, they can make the pixels smaller, which would reduce DR a bit or they can keep the pixel size and have slower conversion.

Best regards
Erik

Thanks! Is it 'difficult' or just more expensive to implement this?

[Luis M. Anibarro]

I just published a short Rant/article about the looming Mirrorless Wars.  It's going to get interesting.  Check it out HERE

great video, thanks for your thoughts

[fdisilvestro]

The Nikon D850 at ISO64 has a measured engineering DR (DxOMark) of 13,55EV which can STILL but hardly be encoded in 14-bit RAW files. Should this sensor have just half extra stop of DR and more than 14 bits would become necessary. What does this means? 16-bit linear encodings are just knocking at the door and we'll see them as soon as an enhanced DR sensor appears on the market, not for marketing reasons (as Canon did with the 40D and its fake 14 bits), but as an engineering requirement.

This is the Pentax K5 with Sony Exmor sensor, the first sensor ever really needing 14 bits:


What if this sensor would just have a 12-bit encoding? posterization (the simulation was done by decimating from 14-bit to 12-bit a RAW file prior to debayering):


That is what would happen in the deep shadows of any sensor with insufficient encoding bits.

Regards

It would be interesting to see what happens if you take the 12 bit file and add 2 random bits to artificially convert it to a 14 bit file and compare it to the original.

[adri]

Don't lose sight of the fact, though, that LuLa doesn't pretend to be a "gear review" site; what it does in that direction is show and discuss stuff that the owners themselves enjoy. Which seems fair enough to me.

Maybe it gets a bit overly sentimental with some brands, but why not if they are the objects of the desires?

Perhaps fair indeed, but not always desirable from us, the readers', point of view (and we are paying for LULA now, after all, which I still think was a bad, but perhaps necessary, idea); we may want to expect more than (now mostly Sony) fan boys' comments.

Objectivity is a strength, not a weakness. Stuff only the owners enjoy and salivate over has the great potential to isolate the readers and to become tiring or boring. I think overall, we expect more. Take e.g. the case of Sigma cameras; some people rave about the Foveon technology; others do not. And on both sides of liking and disliking, truly often not for the right reasons. This is why we need objectivity and not emotionalism.

I am also saying this, as many people have returned to the simplicity of film. There's a reason for that. Too many blogs are filled with endless pixel-peeping and geeky technological and complex discussions. Are we supposed to be engineers too? Have that kind of knowledge? It's totally overwhelming what's happening in so many blogs (e.g. dpreview).

I am tempted myself to resurrect shooting with film.

Some digital cameras encourage this more simpler way of landscape and artistic shooting; no need for 10-20 frames per second, etc. Is a "slow camera" a bad camera? Not if it has stellar IQ.

Also, in this day and age, suddenly cameras are judged on their video capabilities. I personally never shoot video and would rather have a high end camera without video, and save some money. It would be nice if higher end cameras were offered as just stills cameras only for a reduced, yet affordable, price (I'm talking well below $10k).

"Objects of desire": The Fuji GFX 50S (or R to come?) fits in that category for me. It's not a Sony. Lol.

Right now, I'm waiting for more announcements before I add another system to my existing arsenal. You get my drift: I'm not a crowd follower. When everyone turns right, I want to see what's happening on the left side. :-)

I believe I'm certainly not alone in this.

[Rob C]

Perhaps fair indeed, but not always desirable from us, the readers', point of view (and we are paying for LULA now, after all, which I still think was a bad, but perhaps necessary, idea); we may want to expect more than (now mostly Sony) fan boys' comments.

Objectivity is a strength, not a weakness. Stuff only the owners enjoy and salivate over has the great potential to isolate the readers and to become tiring or boring. I think overall, we expect more. Take e.g. the case of Sigma cameras; some people rave about the Foveon technology; others do not. And on both sides of liking and disliking, truly often not for the right reasons. This is why we need objectivity and not emotionalism.

I am also saying this, as many people have returned to the simplicity of film. There's a reason for that. Too many blogs are filled with endless pixel-peeping and geeky technological and complex discussions. Are we supposed to be engineers too? Have that kind of knowledge? It's totally overwhelming what's happening in so many blogs (e.g. dpreview).

I am tempted myself to resurrect shooting with film.

Some digital cameras encourage this more simpler way of landscape and artistic shooting; no need for 10-20 frames per second, etc. Is a "slow camera" a bad camera? Not if it has stellar IQ.

Also, in this day and age, suddenly cameras are judged on their video capabilities. I personally never shoot video and would rather have a high end camera without video, and save some money. It would be nice if higher end cameras were offered as just stills cameras only for a reduced, yet affordable, price (I'm talking well below $10k).

"Objects of desire": The Fuji GFX 50S (or R to come?) fits in that category for me. It's not a Sony. Lol.

Right now, I'm waiting for more announcements before I add another system to my existing arsenal. You get my drift: I'm not a crowd follower. When everyone turns right, I want to see what's happening on the left side. :-)

I believe I'm certainly not alone in this.

There's no doubt about that!

However, as you obviously know, no site can  give you everything, because sites are made up of people who, for better of for worse, are driven by personality and, thus, emotions that include likes and dislikes. No site can be all things to all men (or women) and let's face it, LuLa has a pretty broad demographic of users that doesn't appear to be in any particular hurry to go somewhere else. Where to, becomes the first question for anyone thinking of jumping ship.

Film. Film is not simplicity. I spent a career working solely with film. The only thing that appears to be simple with film, in retrospect, is that you didn't get to pixel-peep and read learned discussions on the Internet that, perhaps fortunately (?) wasn't around for most of us. All you needed to do was learn some basic exposure and development routines - and stick to them like a zealot - and all was usually well. Simple on the face of it, and no more difficult than is using a camera today if you choose to set it to as near manual as you can (my way) and pretend you are using your old film bodies. But, in both cases, film and digital, simplicity is surface: below that, all sorts of very complex stuff is going down, with all kinds of possibilities of failure just biding their time to make you feel an idiot.

Like you, I never shoot video. Neither do I shoot sequences at eye-defeating speed or at all, come to think of it. And yes, I think it would be good if there was choice between a pared-down, high quality camera and one complete with the church steeple. A slight problem could be in deciding what's vital and what's not well enough to suit the ideals of a reasonable set of buyers.

Part of your argument is that you don't need whatever goodies than IQ for landscape and deliberate shooting; I don't do landscape at all, and I'm also a pretty slow worker now - different needs/opportunities - but neither do I have interest in shooting the skies at night, and so I guess a lot of the features that kind of photographer needs mean zero for me. What I'm trying to suggest is that there isn't really such a thing as the average photographer. How can a camera maker decide on a way to spread the goodies across bodies without sabotaging his own sales?

If I were to be granted my digital wishes, they would be:

1. a digital 135 format camera that permitted me to use my old Metz should-pack flash at around the 1/1000th second speed;

2. a digital 500 Series Hasselblad that used a full-frame sensor.

Both would have to be within my limited price-ceiling, like yours, and well below 10k anythings!

Would I use them much? Probably no more than I use what I already have, is the unfortunate truth.

Rob

[Kevin Raber]

Adri . . .  I have explained in a video all about why Sony gets attention.  Soon, that will change.  We have a good piece coming up with Olympus.  Which by the way I own and we covered a lot on this site.  I have a video and article almost ready to go about using the Fuji X-H1.  I own every Fuji camera and a ton of lens as well as Panasonic GH cameras.  And, throw in some Phase One technical camera gear.  Soon, we'll be covering both Nikons and Canons venture into mirrorless.  I hope they come out with innovations that re fun.  It will be fun to see if they innovate at the rate that Sony does.  And, I think you'll find a lot of talk about Sony on many other sites too as Sony is the only company really pushing cameras out to talk about.  That could change soon.

As far as film goes, been there and done that.  I was one of the early digital pioneers and have never looked back.  Yes, I miss film and the success that I had as a photographer.  Both Michael and I tried to get back to film.  That lasted a few weeks.  Many old-timers here may remember his venture into a Rollei twin lens set up. That lasted about three weeks.  I am talking to someone about doing a collaborative piece together shooting with film and printing in a darkroom.  We may do that later this year.

The real fun with more mirrorless options will be here soon.  Let's see what develops.

[ErikKaffehr]

Hi Kevin,

I tried film a few times and came back with a bleeded nose each time.

Kodak was like a T-ford, it served it's purpose well. But we don't see a lot of them doing any useful work...

Best regards
Erik

Adri . . .  I have explained in a video all about why Sony gets attention.  Soon, that will change.  We have a good piece coming up with Olympus.  Which by the way I own and we covered a lot on this site.  I have a video and article almost ready to go about using the Fuji X-H1.  I own every Fuji camera and a ton of lens as well as Panasonic GH cameras.  And, throw in some Phase One technical camera gear.  Soon, we'll be covering both Nikons and Canons venture into mirrorless.  I hope they come out with innovations that re fun.  It will be fun to see if they innovate at the rate that Sony does.  And, I think you'll find a lot of talk about Sony on many other sites too as Sony is the only company really pushing cameras out to talk about.  That could change soon.

As far as film goes, been there and done that.  I was one of the early digital pioneers and have never looked back.  Yes, I miss film and the success that I had as a photographer.  Both Michael and I tried to get back to film.  That lasted a few weeks.  Many old-timers here may remember his venture into a Rollei twin lens set up. That lasted about three weeks.  I am talking to someone about doing a collaborative piece together shooting with film and printing in a darkroom.  We may do that later this year.

The real fun with more mirrorless options will be here soon.  Let's see what develops.

[Rob C]

Hi Kevin,

I tried film a few times and came back with a bleeded nose each time.

Kodak was like a T-ford, it served it's purpose well. But we don't see a lot of them doing any useful work...

Best regards
Erik

Erik, because something may not suit the great unwashed - or even some of the über classes, just to retain a democratic persepective -  does not imply that it is useless. Far from it.

With the usual caveat of "money no object", I would enjoy going back to my old 'blad 500 system and producing transparencies and black/white negatives again. I no longer print for several reasons already developed here too often; that said, those cameras, for me, would still be valuable today if only for the sheer pleasure of using them. They represent a pinnacle of format - and tactile - perfection. Nothing else that I owned ever gave me that degree of satisfaction and sense of photographic control.

Rob