phase versus hassleblad

[NikoJorj]

I wouldn't have thought I'd dare to intervene in a MFDB-vs.-MFDB thread... but as this is mostly about dynamic range I'd throw in the 2 (euro) cents of a cheapskate only shooting with a goodol'300D (yeah, the original DRebel ; some results in the link below in my signature).

First, the theoretic definition of sensor DR as difference between saturation and noise floor is theoretically correct... But the difference between theory and practice is greater in practice than in theory, isnt' it?
Where perception kicks in is in the noise structure : with the same measured Signal to Noise Ratio, a perfect gaussian noise on one side will be far more acceptable, photographically speaking, than the same noise only made of banding (that fugly orthogonal pattern in noise, very evident in some cases with my camera).
I'd really like to stress that one : banding noise shall also be measured in a way or another, besides noise. That can also be applied to other structures of noise, as eg the "noise blotching" (color noise splattered on a few pixels) of some P&S sensors (though that is a bit farther from the topic).

Second, I'd say there is yet another factor in real-life DR, especially when dealing with large ranges above 10 stops : lens veiling (sometimes called flare - I'm not talking about the colored artifacts showed with a point light source near or in the field, but about the general contrast reduction caused by light scattering in the lens or sensor chamber).
This parameter can be offset by shooting the same low-contrast target in a dimmer environement with + and - exposure correction, but in a real world image, the whole thing is about to capture shadows and highlights in the same shot, so veiling can also be there.
In a first time, lens veiling could improve the ability to record high DR scenes, as it levels up the shadows (it was a darkroom trick in the form of preexposure if I remember well).
However, with deeper shadows, it doesn't only level up the tonality but decreases contrast, to the point that initial texture is drowned in a grey veil, and so it could also limit DR.
I've encountered this phenomenon while playing with modern raw files (see link below) : sometimes while pulling up the shadows, it only brings an uniform grey before bringing up objectionable noise.

Third, about marketing claims vs. facts : the "playing with raw files" was actually a game on a french forum (link in french with MANY side discussions), where someone (some here may know Thierry Legault?)  took some raw files of various origins and zeroed the N least significant bits, effectively converting them into 14bits to 8bits raws.
Bottom line : no reasonably visible difference past 10 bits, no pixel-peeping significant difference past 12 bits, even with MFDB files (MFDB pages 7-8-9, note that the first test on page 1 had a 2bit bias). There is just plain noise in the last bits, as said elsewhere. However, I suspect now that in some cases the veiling might limit the DR too much.

[fredjeang]

Well, assuming that the current talk about MF sensors being that much superior DRwise is actually true, don't you think that such miracle technology available for 6500 US$ would be game changing?

Wait... it is actually already available in second hand P45 for no more than 9.000 US$. You got to wonder why both sellers and buyers have aligned their prices with high end DSLRs although they remain that superior...  



Well, many of the people in this thread are known for their excellent photographs. They managed to take excellent photographs with slide film also, and don't appear to have changed their style that much... although they gained a huge 5 stops of DR.  

As far as my selfish self is concerned, you will find a few images here. http://www.flickr.com/photos/bernardlanguillier/

Most shot with a DSLR (SLR/n, D2x, D3 and D3x in the past 15 months), some shot with a Mamiya ZD in those days where I was still a believer in some form of MF black magic.  

Cheers,
Bernard

Ok, Bernard.

I know your portfolio. My points of course where not directed to you personaly but in general. When I was talking about the lack of images I was not thinking of directing people to one's wesite to see his beautiful pictures. Let me explain myself:
This topic has been on and on for ages now, and as soon as there is a possibility, the forum (or the forums) is in flame.
There are clearly 2 gangs but in general "dslr-band" are very sensitives with the asumed superiority of MF in terms of IQ and tend to react with more passion.
In both bands, there are knowledgable and experienced photographers and I have no doubt that you are one of them.
But then, both bands contradict themselves constantly and one of the weapon most used is DxO. Others are kind of home-made physical tesis.
As a spectator because I'm not technicaly competent, and in desire for reliable information, these endless contradictions are bringing more confusion, and I'm sure I'm not the only one who feel it.
Many points are interesting, truly I learn a lot of technical aspects but at the end there is no sensation of information but passion.

I thought that Michael's note would have brought more light, but it just reactivated the war and arguments against MF superiority.  
So, imagine you start photography and see this. What would you think? Simply that it is a real mess.
That is where the idea of ilustrating arguments with 100% images is coming.
Jenbenn just put an interesting nuance in his post and it is indeed what I think.

Why did I say that I tend to trust more Michael and some other members here than a lot of dslr posters? Not because the others are not good photographers and I give my respect to everyone. BUT, and it is an important "but", because I know that Michael is using daily both systems and prints a lot from these 2 systems. That is the point Jenbenn has made: daily; and I do think he is right.
So I do not see why Michael would put a note in his web, that it his seen worldwide, confirming the DR superiority of MF if he was not sure, or at least if the real field experience would not have demonstrated such a fact.
But then, this reactivated the flames.

In that context, the idea to make a real test, in real working conditions between 35mm and MF is not that stupid I think. But with downloadable 100% files.
I'm particularly interested to see the 2 systems in different situations with texture, complicated patterns, uncontroled light outdoor and studio shots and see how they performs. Such an intensive test would bring a real therapy to the photography community all over the world and hopefully will end these showers of graphics and DoX testings.

To conclude, I do not understand why there is such a separation: If I could now, I would own both systems and use them according to my needs, desires and according to what they are best at.

Regards,

Fred.

ps: if you read french, let me note that there is a dedicated french website, and they are serious, about high resolution photography here
In any case they talk about 35mm dslr, only usable for them for field pre-reportage. I suspect them to be kind of integrists and purist but there is no way they consider the 35mm at the level of MF. They just ignore complitely these gears. I imagine that if 35mm had reach such a point, they would have put a kind of "alleluia" note.

[Jeremy Payne]

...

...

[michael]

I'm not going to fan the flames any more than needed, but I can't resist making one more comment in support of Fred's.

The world-wide market for medium format backs is approximately 7-8,000 units. That's all manufacturers combined.

These devices (with camera body and a couple of lenses) are priced in the $30–$50,000 range U.S. As Mark pointed out in his essay, that's the cost of a current luxury car.

One more factoid: About 70% of MF systems are bought by working professional photographers, the rest by fine art photographers and wealthy amateurs.

Here's the point. For the pros an MF system is a tool for earning their livelihood. It is purchased with the intention of helping them make money. No person would spend a good chunk of a year's income on a tool that did not offer an advantage. In my experience every pro that I know (and I know quite a few around the world) has done their own exhaustive comparisons and test, not relying on web site chatter, magazine reviews and online tests.

The fine art photographers are in a similar boat. They too make their livelihood (or a good part of it) from exhibiting and selling their prints. If a $1,000 to $6,000 DLSR gave them what they needed, why on earth would spend $30-$40,000 on a back?

As for the wealthy amateurs, I'll just say that it my experience (having taught and worked with several thousand folks like this over the years on my workshops and tours) I have yet to meet one who wasn't a savvy business person who understands the value of a dollar. Like the pros and the fine art guys, they spend that kind of money for an MF back because they believe that it will provide them with a tool that allows them to produce superior quality images. And if you speak to the people at major VARs who sell this gear, you'll learn that the amateurs are often even more demanding as customers than pros.

So, where does that leave us? Are all, or even most of the people that buy medium format equipment fools, willing to spend their hard-earned money chasing a chimera, or simply looking to boost their egos regardless of cost? I don't think so, and it seems to me that anyone that does, simply hasn't taken the time to do a simple side-by-side comparison for themselves.

Frankly, there is no argument, only lack of knowledge and experience, or living in denial.

Michael

[BernardLanguillier]

Ok, Bernard.

To conclude, I do not understand why there is such a separation: If I could now, I would own both systems and use them according to my needs, desires and according to what they are best at.

Regards,

Fred.

ps: if you read french, let me note that there is a dedicated french website, and they are serious, about high resolution photography here
In any case they talk about 35mm dslr, only usable for them for field pre-reportage. I suspect them to be kind of integrists and purist but there is no way they consider the 35mm at the level of MF. They just ignore complitely these gears. I imagine that if 35mm had reach such a point, they would have put a kind of "alleluia" note.

Fred,

OK, fair enough.

I personnally have no problem with the idea that a 40.000 US$ back might have a bit more DR than the camera I use.

This whole thread is a reaction to the totaly unreasonnable assumption that the gap is huge, and to the lack of rigor of the explanations to support these statements. I am reacting more as an engineer than as a DSLR owner who would be afraid to lose whatever fight. For my applications I would not buy a back even if it cost less than my D3x, and I really mean it.

Et oui, je suis francophone et les liens m'interessent, merci.

Cheers,
Bernard

[PierreVandevenne]

There are clearly 2 gangs but in general "dslr-band" are very sensitives with the asumed superiority of MF in terms of IQ and tend to react with more passion.

IQ is a vast subjective topic and can be debated forever. Lenses do play a non-negligible role <G>

But as far as sensors capabilities are concerned, it is actually very simple, convenient because 1 stop essentially equals 1 bit.

Looking at the H3DII 50 for example, it is known that it uses the KAF-50100 sensor. The sensor characteristics are available here

http://www.kodak.com/global/en/business/IS...138/13219/13220

It saturates at 40300e and has a read noise of 12.5e for a 70.2 db dynamic range.

Checking the specs with the very simple formula given here (a manufacturer of scientific CCD products)

http://www.ccd.com/ccd111.html

40300 / 12.5 = 3224 (in other words, basically right between 11-12 bits/stops)
log (3224) = 3.508...
3.508 * 20 = 70.167 db

What is the measured DxOMark?

http://www.dxomark.com/index.php/eng/Image...elblad/H3DII-50

11.5 stops.

A DSLR with 5.5 stops (6 less) would just be able to produce posterized jpegs.

If you want to gain stops, and you reduce read noise to 1 e instead of 12.5 e, you've gained 4 stops. You should still work on well capacity, which can't really be increased unless you move to larger pixels or different processes. Anyway, Hasselblad is stuck with the Kodak sensor and will not improve read noise, so the point is a bit moot.

One factor that plays in favour of MFDBs, if you compare it to a DSLR in a print that has the same physical size, is that some relative binning occurs since the MFDB has more pixels. This is a good thing, but it won't give a 6 stops improvement unless the number of pixels are drastically different.

[imagico]

As discussed elsewhere in this thread, the relevance of their figures for actual photography is doubtful, but the same doubt applies to both the DSLR and MFDB figures, and there has been no explanation proposed as to why the gap between their figures and real world ones would have to be larger for DSLRs than for MFDB, and certainly not 6 stops larger.

I think this greatly summarizes the whole issue.

It would probably be unfair to put too much weight on the claimed 6 stop difference since Mark Dubovoy says he did not do any quantitative comparison.  But Bernard's argument is still valid even if it is about just 5 stops of course.

Greetings,

Christoph

[Graeme Nattress]

A claimed 6 or 7 stop difference is a vast difference - a 42db increase in SNR. It's not something small or trivial. It's not something that should need measuring, but measuring it should produce an utterly un-ambiguous result. It's also a very easy thing for a camera manufacturer to demonstrate with a fairly affordable calibrated backlit chart. Dalsa, a sensor manufacturer and supplier to MF backs lists their 36x48mm 48MP CCD sensor as a SNR of 74db - 12.33 stops in other language. This is very much the dynamic range of a top end image sensor for photographic applications. This either means that Canon and Nikon etc. have a 74db-42db = 32db sensor, which is utter rubbish, or they too have a ~74db sensor in their cameras, which from measuring them on charts I can well believe. Now, as I mentioned above, there's a number of image processing and lens factors that can effect the appearance of the DR, but if we're just talking sensors here, I'm very confident that MF backs don't have a 6 or 7 stop advantage.

Graeme

[fredjeang]

Et oui, je suis francophone et les liens m'interessent, merci.

Cheers,
Bernard

Interesting topic here: http://www.galerie-photo.com/service-tirag...nt-lafolie.html
about a japanese paper for prints: Tengusho.
it seems that this paper has exceptional qualities. I did not know it.
But that would be a topic for another forum room.  

Cheers,

Fred.

[JohnKoerner]

I'm not going to fan the flames any more than needed, but I can't resist making one more comment in support of Fred's.
The world-wide market for medium format backs is approximately 7-8,000 units. That's all manufacturers combined.
These devices (with camera body and a couple of lenses) are priced in the $30–$50,000 range U.S. As Mark pointed out in his essay, that's the cost of a current luxury car.
One more factoid: About 70% of MF systems are bought by working professional photographers, the rest by fine art photographers and wealthy amateurs.
Here's the point. For the pros an MF system is a tool for earning their livelihood. It is purchased with the intention of helping them make money. No person would spend a good chunk of a year's income on a tool that did not offer an advantage. In my experience every pro that I know (and I know quite a few around the world) has done their own exhaustive comparisons and test, not relying on web site chatter, magazine reviews and online tests.
The fine art photographers are in a similar boat. They too make their livelihood (or a good part of it) from exhibiting and selling their prints. If a $1,000 to $6,000 DLSR gave them what they needed, why on earth would spend $30-$40,000 on a back?
As for the wealthy amateurs, I'll just say that it my experience (having taught and worked with several thousand folks like this over the years on my workshops and tours) I have yet to meet one who wasn't a savvy business person who understands the value of a dollar. Like the pros and the fine art guys, they spend that kind of money for an MF back because they believe that it will provide them with a tool that allows them to produce superior quality images. And if you speak to the people at major VARs who sell this gear, you'll learn that the amateurs are often even more demanding as customers than pros.
So, where does that leave us? Are all, or even most of the people that buy medium format equipment fools, willing to spend their hard-earned money chasing a chimera, or simply looking to boost their egos regardless of cost? I don't think so, and it seems to me that anyone that does, simply hasn't taken the time to do a simple side-by-side comparison for themselves.
Frankly, there is no argument, only lack of knowledge and experience, or living in denial.
Michael

All of this makes perfect sense Michael.

However, I still believe there is more than a little amount of hypocricy and double-talk going on in these reviews. I am of the consensus that many of these reviews are made to deliberately "stir controversy" (ala Ken Rockwell), rather than to be 100% factual. And here is what I mean:

I believe it was last year sometime that you yourself wrote an article comparing (of all things) the Canon Powershot G10 to a MF P65 ... where in you yourself stated that (under ideal conditions) the little point-n-shoot Canon was able to take photographs that were so compelling that, at smaller sizes, they "could not be distinguished" from identical photos made by the P65.

You did write such an article, didn't you Michael? I believe several members here will recall this fact. I further recall that this wasn't just your experience; no, you claimed you had 20 (or was it 30?) "top professionals" view several printed images you made with the little point-n-shoot Canon ... and compared them to identical printed images made with a top-end MF back ... and that "none of you could reliably tell the difference" up to a certain size image. And that was looking at them up close.

And now here we are reading another fellow's claim (Dr. Dubovoy's) that states the exact opposite, namely that (forget a P&S) anyone should be able to see the difference between a MF image compared even to the highest-quality DSLR, on small prints, and that this clear and unambiguous difference can be seen from as far as 30 feet away! And here you are, Michael, backing-up Dr. Dubovoy.

Now, I have nowhere near the technical acumen of you distinguised gentlemen, at least not as far as photography goes, but I do recognize inconsistent statements and postures when I read them. So I am wondering, Micheal, how can you say out of one side of your mouth (last year) that neither you, nor 30 "top professionals" could reliably differentiate between P65 images and PowerShot G10 images last year (when both were taken under ideal conditions) ... and yet this year your position is that ANY professional can easily distinguish the image quality of a P65 image and a top-level DSLR image from as much as 30' away?

How can you say both statements and have them both hold water?

Or is the real truth something more like, when Canon butters your bread on a review that even their P&S cameras can favorably-compete with a MF back ... but when the MF companies butter your bread, nothing can compare to them ... even the best DSLRs ... on even the tiniest of images ... and even from 30' away?

I don't mean to be disrespectful, but the above paragraph is the only conclusion that makes sense to me, given the absolute incongruity of the two postures.

Jack

PS: I can absolutely tell the difference between a photo taken with a G10 and a P65 ... and even from most DSLRs ... but not on tiny images and certainly not from 30' away. At some point, a person has to call "bullship" and ask that some accountability to be given for all of these purely subjective (and absolutely self-contradictory) extremist positions.




.

[NikoJorj]

Frankly, there is no argument [...]

For me either, there is no argument that MFDBs costing almost as much as my previous flat do have an edge in IQ.
The debated question is why...

[Graeme Nattress]

For me either, there is no argument that MFDBs costing almost as much as my previous flat do have an edge in IQ.
The debated question is why...

6 or 7 stops is not an edge, it's a modern tank against bows and arrows.

The question is indeed "why?". What measurable parameters of the resulting image are superior, and by what amount.

[deejjjaaaa]

I will admit to little actual first hand knowledge of this, but from what I've been led to believe (and is supported by this article ), the micro lenses are not lenses in the sense of focusing light, but instead redirect light that would normally fall on the wasted area between sites in to a site.  As such I don't believe they have the ability to focus the light on the specific area of the site that is light sensitive.

so why bother to put the microlenses at all if they don't increase the amount of light hitting the light sensitive area... so albeit not ideally they still increase the amount of light reaching the needed place.

[deejjjaaaa]

Frankly, there is no argument

still no raw files, sir...

[michael]

John,

There is nothing dishonest going on; and yes, I stand behind both positions.

How is this possible? Simple. Many things in life are ambiguous, and a tiny change in beginning state can lead to a huge difference in end state.

It was Fitzgerald who wrote, "Intelligence is the ability to hold two opposing ideas in mind at the same time and still retain the ability to function."

In the case of the G10 vs Phase article most of the "trained observers" couldn't see the difference in 11X17" sized prints between the two cameras. But, of note is that there were a few who could. How? Because they didn't look at sharpness alone. They looked at depth of field, tonal accuracy, dynamic range and several other characteristics (including some that are hard to name) that gave the game away. Even pros can be mislead.

There has been more than one person comment that they can tell just by looking at my Home Page shots which have been taken with MF and which with other cameras. In an 800 pixel image this simply shouldn't be the case. But it is easily the case for those with a good eye.

I could go on, but won't. In the end what I've seen over the years (decades) of doing and teaching photography is that nailing down the question of image quality is a slippery fish. Just when you think you've got a grip, it pops out of your hands.

I guess that I'll simply continue my life knowing that photography, as with life itself, is full of contradictions.

Michael

[bjanes]

The question that this raises in my mind (as I see it)  is why are DXO measuring DR by taking the difference between zero signal to noise and full saturation of the sensor? And how does this really correlate to how many F stops above and below middle gray can be recorded with full texture and detail? And further.. why is the difference in DR so great using the two different measurement techniques? And why is is DXO's methodology not relevant in the real world? (as this seems to be what is being implied).

If you shoot to the right, as you should with digital capture, middle gray (18% sensor saturation) is 2.47 stops under saturation, and this would be about the same with any digital camera. The rest of the DR is in tones below mid gray, and it does not make sense to use middle gray as a reference point. Digital cameras with high DR achieve it in the shadow part of the image. For a tabular presentation of these relationships see here. Where you place the noise floor determines the relative contributions of shot noise and read noise to DR. A sensor with large pixels but a poor read noise would look better if you placed the noise floor relatively high. These relationships are demonstrated interactively by this Nikon Microscopy Java tutorial.

The engineering definition of DR (full well/read noise) places the noise floor for DR quite low and the signal:noise here would not be useful for photography. For practical photography, one would place the noise floor higher, but a camera with a good DR at zero signal would also have a good DR 10 stops from saturation. Indeed, the per pixel random noise characteristics of a digital camera can be modeled quite well by taking  only shot noise and read noise into account as demonstrated by Roger Clark. Using these data, you can place the noise floor wherever you want. Pattern noise such as banding has to be dealt with separately if it is significant.

[attachment=20759:NoiseModel.gif]

It seems odd to me that DXO with all their scientific efforts would measure DR in a method that did not take into account the linear nature of the sensor. Its not that I am a DXO pundit - far from it. I just want to fully understand why different measuring methods are being used (as this is clearly the reason for the differences) and which (if any) actually correlates to real world results.

The sensor is linear, but the eye and photography are logarithmic. That is just how things are. But one can easily convert to f/stops using logarithms to the base two (for photography) or log base 10 for conversion to decibels as is done in engineering. What is your problem here? DR can be expressed in linear terms or log terms, but the underlying ratios are the same. For example, for the 1DMII at ISO 100, the full well is 53,000 electrons and the read noise is 16.61 electrons, giving a DR of 53,000/16.61 = 3193:1. In terms of f/stops this is log base 2 (3193) = 11.6 f/stops. If you want a S:N of 10:1 for your shadow cutoff, the DR would be 53,000/21.98 or 11.2 stops. Read noise is still predominant at this exposure. If you wanted noise floor with S:N of 30:1, the DR would be less and shot noise would predominate at this level.

Placing the noise floor relatively high might favor a MFDB since they collect a lot of photons dues to their large sensor size and have relatively poor read noise as compared to current high end DSLRs (at that is what I understand--data are hard to come by). A lot of this confusion could be cleared up if the MFDB owners would upload a raw image of a Stouffer wedge and an image with the lens cap on the camera for an estimate of read noise. Why do they not do this?

[Graeme Nattress]

A lot of this confusion could be cleared up if the MFDB owners would upload a raw image of a Stouffer wedge and an image with the lens cap on the camera for an estimate of read noise.

Absolutely correct. That would be exactly what is needed. However, Stouffer tops out a just over 4 OD, and I'm now using an over 5 OD chart so as to fully ensure I'm not missing anything in the noise floor. So on a Stouffer, I shoot a bracket of exposures to ensure I can see the full range.

Graeme

[BernardLanguillier]

Interesting topic here: http://www.galerie-photo.com/service-tirag...nt-lafolie.html
about a japanese paper for prints: Tengusho.
it seems that this paper has exceptional qualities. I did not know it.
But that would be a topic for another forum room.

Thanks Fred, very interesting indeed. And for sure much more relevant than the assumed DR advantage of these magical backs.  

Cheers,
Bernard

[Mark D Segal]

And now here we are reading another fellow's claim (Dr. Segal's) that states the exact opposite, namely that (forget a P&S) anyone should be able to see the difference between a MF image compared even to the highest-quality DSLR, on small prints, and that this clear and unambiguous difference can be seen from as far as 30 feet away! And here you are, Michael, backing-up Dr. Segal.


Jack

Jack - let me correct this - there are two Mark's active in this discussion - the writer of the article who is Mark Dubovoy with a PhD, and then a member of the discussion audience - me - Mark Segal, who does not sport a PhD. I did not make any statement to the effect that ANYONE should be able to see the difference between MF and high-end DSLR images (but I did say that I could in respect of a bit of my own work), nor did I say anything about seeing such differences from 30 feet away. That was the other Mark. Let us keep our Marks de-confused   .

[BernardLanguillier]

So, where does that leave us? Are all, or even most of the people that buy medium format equipment fools, willing to spend their hard-earned money chasing a chimera, or simply looking to boost their egos regardless of cost? I don't think so, and it seems to me that anyone that does, simply hasn't taken the time to do a simple side-by-side comparison for themselves.

Frankly, there is no argument, only lack of knowledge and experience, or living in denial.

Michael,

Nobody discusses the value of MF nor of MFDB, the only point discussed here is the gap in DR. Taking a bit of distance, I am really wondering what triggered the resurgence of this whole DR discussion in March 2010.

In this context, I am looking forward to trying out the Pentax 645D when it is released, that should enable us to clearly separate the topics of vendors, price and format for the sake of a healthy discussion.

Cheers,
Bernard