IQ 180 DxOMark ranking

[marcmccalmont]

I just noticed that DxOMark has tested and ranked the IQ180 overall score 91
I have found a high correlation to DxOMark and what I see FWIW
Marc

[Bart_van_der_Wolf]

I just noticed that DxOMark has tested and ranked the IQ180 overall score 91
I have found a high correlation to DxOMark and what I see FWIW

Hi Mark,

Thanks for signalling this. Indeed, I also have the personal experience (from in-depth testing) that the DxOMark tests correlate quite well with reality, assuming one interprets the tests properly.

One thing that struck me, besides the great test scores, is the difference between Manufacturer's indicated ISO, and actually measured ISO. At an ISO 100 setting, they measure an actual ISO 29 sensitivity, and that difference is persistent at other ISOs (except for the Plus version at higher ISO).

That means that there is a huge amount of highlight protection built in, but that of course comes at the expense of Dynamic Range at the shadow end. Putting it differently, there is a lot (1.79 stops) of ETTR headroom which, when utilized, boosts the DR by some 2 additional stops to a very nice level (despite the small sensels).

Makes one wonder why there is such a huge difference? Do the highlights clip nasty, or what could be the real reason?

Anyway, it's confirmed by these tests that this is one fine piece of equipment ...

Cheers,
Bart

[BernardLanguillier]

One thing that struck me, besides the great test scores, is the difference between Manufacturer's indicated ISO, and actually measured ISO. At an ISO 100 setting, they measure an actual ISO 29 sensitivity, and that difference is persistent at other ISOs (except for the Plus version at higher ISO).

That means that there is a huge amount of highlight protection built in, but that of course comes at the expense of Dynamic Range at the shadow end. Putting it differently, there is a lot (1.79 stops) of ETTR headroom which, when utilized, boosts the DR by some 2 additional stops to a very nice level (despite the small sensels).

Makes one wonder why there is such a huge difference? Do the highlights clip nasty, or what could be the real reason?

I believe that the reason is simply that most photographers prefer the security provided by the apparent ability to recover highlights, even if it at the cost of not knowing exactly where ETTR starts.

It could be argued that an IQ180 has sufficient DR that optimising exposure for clean shadows is not that relevant anymore.

Cheers,
Bernard

[marcmccalmont]

I have a Pentax K5 and an IQ180 and it appears to me the IQ180 has about .5 stop more DR than the K5 when I ETTR the same scene? I dont quite understand why Phase protects the highlights so much? I've been using ISO 32 and ETTR as much as I can. I don't see any "Nasty" highlight clipping just the normal "cyan in the sky" when the blue channel clips.
Marc

[marcmccalmont]

Bernard
Your up late!
Marc

[marcmccalmont]

A little off topic but since Bernard chimmed in......
He is of the opinion that a highend DSLR is not far off of a MFDB, and I have to agree (I think SOTA sensors are now the sony CMOS not CCD's) with the exception that the real reason to purchase a MFDB is to use technical camera lenses that are just scarry good when used with a DB that can resolve like the IQ180. I get goose bumps with my Rodenstock 70HR and the IQ180!
Marc

[Bart_van_der_Wolf]

I have a Pentax K5 and an IQ180 and it appears to me the IQ180 has about .5 stop more DR than the K5 when I ETTR the same scene? I dont quite understand why Phase protects the highlights so much? I've been using ISO 32 and ETTR as much as I can. I don't see any "Nasty" highlight clipping just the normal "cyan in the sky" when the blue channel clips.

Hi Marc,

That's why it puzzles me that such a high headroom is apparently preferred by Phase One. I can't believe they think that professional photographers can't expose correctly for the situation at hand, there 'must' be another reason.

Cheers,
Bart

[BernardLanguillier]

That's why it puzzles me that such a high headroom is apparently preferred by Phase One. I can't believe they think that professional photographers can't expose correctly for the situation at hand, there 'must' be another reason.

There could be another reason, but this discussion has taken place many times here, and none of our Phaseone related forum friends have ever proposed any detailed explanation on this topic.

My guess is that DR in the digital world is a subject that is still a little misunderstood by many photographers, including some of those able to afford a high end back. There is still a widespread belief that highlight behavior is related to DR while we know that only shadows matter for linear sensors. Most photographers are unwilling to change the way they expose to account for the different characteristics of digital. ETTR is only known by a tuny fraction of photographers, 95% of people out these still expose using essentially a grey card.

For someone still thinking with a film mindset, the digital they hate is the one generating ugly and non recoverable blown highlights transitions. In this context, the most concrete materialization of the supposed superiority of MFDB for end users buying them, besides the resolution, is the apparent ability to recover highlights. Considering that resolution is mostly not needed by many of those folks... this highlight recovery story might be the number one factor contributing to the satisfaction of back buyers.

Let's be realistic here, negatives were always less stressful to use than slides. Since even the less digital aware back owner will be able to apply a steep S curve in Photoshop, the behavior that is most reassuring for many photographers is a negative like safe rendering where even large over-exposure has not measurable consequence in terms of non recoverable data. A more slide look will be easy to re-create in post.

The thing is that the raw converters have to be tuned to produce the correct rendering considering the important under exposure resulting from calling ISO100 what is in essence ISO30.

I believe that Phaseone is a lot more realistic about the actual liking of photographers than say Nikon or Canon.

Cheers,
Bernard

[Bart_van_der_Wolf]

[...]
The thing is that the raw converters have to be tuned to produce the correct rendering considering the important under exposure resulting from calling ISO100 what is in essence ISO30.

Hi Benard,

Well, the default "film curve" tonescale in C1 does suggest that I overexpose my ETTR shots on the 1Ds3, by 1/3rd to 2/3rd stops. Of course on the linear tonescale they do not clip at all (I'm not a fool), so there is indeed some built-in assumption that photographers will underexpose to protect highlights. However, underexposure by 1.79 stops (which formally cannot be even called exposing to an ISO rating), is more like proclaiming that photographers are morons.

There's got to be a better reason, I hope.

Cheers,
Bart

[BernardLanguillier]

However, underexposure by 1.79 stops (which formally cannot be even called exposing to an ISO rating), is more like proclaiming that photographers are morons.

Again, there could be another reason, but even if there is none, I don't see this as a sign that Phaseone is looking down at the skills of their customers.

It would simply mean that they have designed a product which is a good match to the way many people still appear to be shooting nowadays, meaning use a light meter for incident lighting, which will essentially give you a grey card reading of the scene you are trying to shoot.

Cheers,
Bernard

[BernardLanguillier]

Bernard
Your up late!
Marc

So were you obviously.

Cheers,
Bernard

[bjanes]

One thing that struck me, besides the great test scores, is the difference between Manufacturer's indicated ISO, and actually measured ISO. At an ISO 100 setting, they measure an actual ISO 29 sensitivity, and that difference is persistent at other ISOs (except for the Plus version at higher ISO).

That means that there is a huge amount of highlight protection built in, but that of course comes at the expense of Dynamic Range at the shadow end. Putting it differently, there is a lot (1.79 stops) of ETTR headroom which, when utilized, boosts the DR by some 2 additional stops to a very nice level (despite the small sensels).

Makes one wonder why there is such a huge difference? Do the highlights clip nasty, or what could be the real reason?

In their preface to the review, DXO does state that the new Phase One is "King of all sensors", but they then note "It is still interesting to note that theoretically, the Phase One IQ180 could score much better with such a huge sensor surface, if the pixel quality would be closer to the quality of the best APS-C (or full frame) sensor’s pixel."

This implies that they still have not got the read noise down and that the sensor would benefit from ETTR, and it is puzzling that they would allow nearly 2 stops of headroom. I presume that they apply a tone curve to brighten up the underexposed image on the LCD and normalize the histogram, making ETTR more cumbersome.

Regards,

Bill

[BernardLanguillier]

In their preface to the review, DXO does state that the new Phase One is "King of all sensors", but they then note "It is still interesting to note that theoretically, the Phase One IQ180 could score much better with such a huge sensor surface, if the pixel quality would be closer to the quality of the best APS-C (or full frame) sensor’s pixel."

Looking at the pace of improvement of MF sensors vs DSLRs ones, it doesn't take much to understand that, unless a miraculous breakthrough occurs in the Dalsa research labs, the only advantage of MF sensors will soon be pixel count.

Now, there will still be many excellent reasons to favor MF in terms of lenses, different UI, different DoF, more pro looking camera, thethering speed, different support structure, satisfaction of knowing you are shooting with the best, differentiation with less loaded photographers,...

Besides a 36Mp DSLR will be more demanding in terms of technique than a 39 MP MFDB since a given amount of camera shake will affect more pixels.

Cheers,
Bernard

[ErikKaffehr]

Hi,

Regarding the shake it depends a bit how you shake. Is your favorite kind of shake lateral or angular?

I'm not that much in favor of DxO-mark putting everything in a single figure of merit, but it seems to me that doubling sensor size makes for about ten units of DxO. My guess is that if/when we see the new 36 MP full size sensor rumored from Sony the crown on the IQ180 may become a little loose. I guess, because the A77 sensor comes in at 78 and it looses a bit in DxO-mark due to the translucent mirror. So I guess that it will be at around 78 + 10 (for full frame). Nikon would use the same sensor but tweak it a bit more?

Best regards
Erik

Looking at the pace of improvement of MF sensors vs DSLRs ones, it doesn't take much to understand that, unless a miraculous breakthrough occurs in the Dalsa research labs, the only advantage of MF sensors will soon be pixel count.

Now, there will still be many excellent reasons to favor MF in terms of lenses, different UI, different DoF, more pro looking camera, thethering speed, different support structure, satisfaction of knowing you are shooting with the best, differentiation with less loaded photographers,...

Besides a 36Mp DSLR will be more demanding in terms of technique than a 39 MP MFDB since a given amount of camera shake will affect more pixels.

Cheers,
Bernard

[marcmccalmont]

I hoping for a Sony CMOS sensor in the IQ181
DxO ranking 100, real liveview and Movies!
Marc
oh yea 98% trade in value with a IQ180!!!

[torger]

The problem I have with Dxomark is that they are lousy to explain their testing methods, so it is hard to see what they exactly measure.

Someone who knows how they measure? Is it per pixel, a specific percentage of an image area, or a specific square mm area on the sensor? Is there an advantage of large sensor area in the measurements or not? And is it an advantage to have large pixels instead of small, if they measure per pixel it will, but if they measure a specific area it will be a fair comparison.

The fairest would probably be to measure say a fixed percentage of the sensor, preferably the whole area but that will probably be impractical (you want several test patches in an image to speed up testing), but if each test patch is at least a few percent of the area it would cover enough pixels to take away the pixel factor. Say if each test patch is fixed to 5% of the total sensor area, pixel size would not give any unfair advantage, but you will get an advantage of having a larger sensor area (collecting more light - less photon shot noise) which is fair.

My guess is that they measure on test patches covering a few percent of the sensor area, but that the size of them is not precisely tuned.

[marcmccalmont]

It is encouraging that all top of the line cameras now are very very good!
IQ180 WRS 40mm HR
Marc

[bjanes]

The problem I have with Dxomark is that they are lousy to explain their testing methods, so it is hard to see what they exactly measure.

Someone who knows how they measure? Is it per pixel, a specific percentage of an image area, or a specific square mm area on the sensor? Is there an advantage of large sensor area in the measurements or not? And is it an advantage to have large pixels instead of small, if they measure per pixel it will, but if they measure a specific area it will be a fair comparison.

The fairest would probably be to measure say a fixed percentage of the sensor, preferably the whole area but that will probably be impractical (you want several test patches in an image to speed up testing), but if each test patch is at least a few percent of the area it would cover enough pixels to take away the pixel factor. Say if each test patch is fixed to 5% of the total sensor area, pixel size would not give any unfair advantage, but you will get an advantage of having a larger sensor area (collecting more light - less photon shot noise) which is fair.

My guess is that they measure on test patches covering a few percent of the sensor area, but that the size of them is not precisely tuned.

Perhaps you have not studied their test methods as presented on the web site in sufficient detail. Their test target is described here. As you can see, they measure a small area of the sensor for each reading. The standard deviations represent per pixel performance. They then normalize for resolution as shown here. These data are presented on their graphs as Screen and Print modes.

Regards,

Bill

[marcmccalmont]

"As can be seen, high-resolution sensors will gain more SNR, DR, TR and CS when reduced to a lower reference resolution. For DxOMark Sensor Overall Score and Metrics, we chose a reference resolution equal to 8 Megapixels"
Marc

[torger]

Thanks for links and info, I've missed some of that. I shall read more about the tests...