Capture Integration 1st Takes on IQ3 100 Trichromatic and XF Feature Update #4

[Steve Hendrix]

Initial articles on the announcements from September 12 by Phase One:

Phase One IQ3 100 Trichromatic by Dave Gallagher:

"I have anticipated hearing some extreme reactions to today’s announcements from Phase One. On the surface it might be hard to understand why subtle changes to a filter pattern on a sensor can really matter. And to these questions let me pose my own.... "

https://captureintegration.com/top-ten-the-significance-of-todays-announcements/


Phase One XF feature Update #4 by Brad Kaye:

"Once again, through this newest firmware offering Phase One has vastly improved the capabilities of the XF with new features and refinements. Feature Update #4 works with either the HAP-1 or HAP-2 autofocus sensor, and ... "

https://captureintegration.com/top-10-feature-update-4-hap-2-update/


More to come - due to the remnants of Hurricane Irma coming through Atlanta, we're slowly getting up to speed with additional testing and thoughts.


Steve Hendrix/CI

 

[ejpeiker]

Thanks for the articles - note that the focus trim targets are already downloadable and the ZIP comes with ones for different paper sizes:
https://www.phaseone.com/en/SupportMain/Manuals/Manuals_XF.aspx

For some reason it's on the Manuals page and not the Downloads page.

 

[Steve Hendrix]

Thanks for the articles - note that the focus trim targets are already downloadable and the ZIP comes with ones for different paper sizes:
https://www.phaseone.com/en/SupportMain/Manuals/Manuals_XF.aspx

For some reason it's on the Manuals page and not the Downloads page.

Thanks EJ.

I'm still checking to see if these can be ordered separately. It's a convenience to print, but still nice to have the option to order one.


Steve Hendrix/CI

 

[DrakeJ]

Initial articles on the announcements from September 12 by Phase One:

Phase One IQ3 100 Trichromatic by Dave Gallagher:

"I have anticipated hearing some extreme reactions to today’s announcements from Phase One. On the surface it might be hard to understand why subtle changes to a filter pattern on a sensor can really matter. And to these questions let me pose my own.... "

https://captureintegration.com/top-ten-the-significance-of-todays-announcements/


Phase One XF feature Update #4 by Brad Kaye:

"Once again, through this newest firmware offering Phase One has vastly improved the capabilities of the XF with new features and refinements. Feature Update #4 works with either the HAP-1 or HAP-2 autofocus sensor, and ... "

https://captureintegration.com/top-10-feature-update-4-hap-2-update/


More to come - due to the remnants of Hurricane Irma coming through Atlanta, we're slowly getting up to speed with additional testing and thoughts.


Steve Hendrix/CI

Hey Steve!

In your article you write:
"Based on how far you turn right or left or tilt up or down in the first capture, the camera will at –the moment of focus– dial in a predicted offset value that keeps the subject in focus when you recompose to the same location."

I don't really understand this premise, it has to make a prediction based on previous recomposition all the time? This would render a pretty severe percentage drop in hit rates during a portrait session unless you always do the same movement...?

 

[Steve Hendrix]

Hey Steve!

In your article you write:
"Based on how far you turn right or left or tilt up or down in the first capture, the camera will at –the moment of focus– dial in a predicted offset value that keeps the subject in focus when you recompose to the same location."

I don't really understand this premise, it has to make a prediction based on previous recomposition all the time? This would render a pretty severe percentage drop in hit rates during a portrait session unless you always do the same movement...?

What Brad has not yet detailed in that post is the way that the Phase One AFr works - which could be quite different than how Hasselblad's True Focus works - (as alluded to by JDuncan in a post on another thread).

Essentially, the AFr mode is employing predictive intelligence based on learned behavior from how the user moves. Your first shot, you may hear a "meh" audible feedback from the XF. What it is saying, that one is maybe not so on target. But the second and successive shots will learn how you're moving and adjust and dial in based on that learned intelligence.

It sounds improbable, but so far in our testing, seems to work extremely well. So, for a given set of shots, you'd drop a first shot to let the camera know the general way you're moving. When you would change your movement or angle dramatically to another set of shots, the same thing, you'd inform the AFr what you're doing (roughly) with the first shot, and it will get the hang if it quite quickly.


Steve Hendrix/CI

 

[DrakeJ]

What Brad has not yet detailed in that post is the way that the Phase One AFr works - which could be quite different than how Hasselblad's True Focus works - (as alluded to by JDuncan in a post on another thread).

Essentially, the AFr mode is employing predictive intelligence based on learned behavior from how the user moves. Your first shot, you may hear a "meh" audible feedback from the XF. What it is saying, that one is maybe not so on target. But the second and successive shots will learn how you're moving and adjust and dial in based on that learned intelligence.

It sounds improbable, but so far in our testing, seems to work extremely well. So, for a given set of shots, you'd drop a first shot to let the camera know the general way you're moving. When you would change your movement or angle dramatically to another set of shots, the same thing, you'd inform the AFr what you're doing (roughly) with the first shot, and it will get the hang if it quite quickly.


Steve Hendrix/CI

Sounds like a really really awkward way of getting around Hasselblad's patent. Why give a "meh" result the first time, that shot could be the one...

Edit: Sorry for sounding like a debbie downer. I'm still going to give it a go and I hope it'll work fantastic =)

 

[DougDolde]

I can see a difference in the samples Dave posted with the slider but it's so subtle and hard to tell which is correct (or better) from the subject used.

 

[tcdeveau]

Sounds like a really really awkward way of getting around Hasselblad's patent. Why give a "meh" result the first time, that shot could be the one...

Edit: Sorry for sounding like a debbie downer. I'm still going to give it a go and I hope it'll work fantastic =)

Not that it really matters for the new Phase features, but FWIW I don't think Hasselblad has a patent on True Focus. I didn't do an exhaustive search, but last I checked they had a patent application for the technology that was abandoned a few years ago, meaning the application never matured into an issued patent. I posted about it more in detail somewhere in another thread a few months ago (I draft and prosecute patents for a living). The tech could be buried in another one of their patents somewhere, but the search I did in the US and internationally didn't give me that indication. Just because manufacturer A has a product with a feature that manufacturer B's product lack doesn't mean that manufacturer A has a patent on that feature. Someone else could still have a patent on the tech though, and Hasselblad could merely be licensing it from someone else.

Nice job Phase with the new Trichromatic and XF feature update

 

[Steve Hendrix]

Sounds like a really really awkward way of getting around Hasselblad's patent. Why give a "meh" result the first time, that shot could be the one...

Edit: Sorry for sounding like a debbie downer. I'm still going to give it a go and I hope it'll work fantastic =)

I understand, but at this point I don't see it as a way to get around a patent. I see it as potentially a better method that may prove to be more flexible in how it adjusts to re-compose movements. The proof is in the pudding, as they say. It's also nice that Phase One users weren't required to pay for a camera upgrade, as Hasselblad users did, to gain the feature benefit.


Steve Hendrix/CI

 

[ErikKaffehr]

Hi Steve,

My take is that the present information is full of hyperbols and no content. Very clearly, making the CFA design more orthogonal, that is having less overlap may give better color separation at some colors, like a Velvia look. A overlap free CFA would not be able to reproduce significant part of the color spectrum. So, what P1 describes is a very bad CFA design.

They also offer a an image to download, a totally unusable image.

So info is probably misleading... It can be a great design, but there is no relevant information, no relevant images. Just a lot of marketing hyperbols.

I am also a bit opposed to claims about highlight recovery in MFD.

• For one thing, it contradicts imaging science
• I also made some fairly detailed comparison between cameras available to me and found no visible or measurable difference in highlight reproduction between 2006 generation CCD and 2015 generation CMOS.
• Imaging science says no difference and there is no difference, in higlights.
• In shadows, the 2015 generation CMOS wins.

Higlights:

Shadows:

Regarding colour fringing, any colour fringing would probably result from weak IR or UV suppression combined with poor correction of the secondary spectrum outside the visual range. Lloyd Chambers has illustrated this for Leica S-lenses, but I have seen it demonstrated many times. Both CCD and CMOS are monochromatic devices, with the difference being in the readout method. The photodiode is not different between the different devices.

It is feasible that CCD designs for MFD would have less crosstalk than some CMOS designs for DSLRs. In that sense the old Kodak based CCDs were said to be very good.

Please don't misunderstand me, the Thricromatic may be great, it is just that the information coming from P1 is much like Pyongyang information or alternate facts and that would be a consideration if you are selling a 45k$US system.


Best regards
Erik

 

[ejpeiker]

My take after looking at everything available so far is that for what I use the XF for, landscapes, there is likely little to be gained from the refined color array filter since colors are generally always modified to some extent anyway, at least not enough of a change to justify the likely cost of an upgrade from the current IQ3100. On the other hand, a base ISO of 32 is VERY appealing, but then I have to ask myself if 1/3 of a stop is really worth whatever the cost is too unless it's low single digit thousands of dollars for the upgrade which I doubt. IF I were ordering a new IQ I would certainly get this new version but replacing an existing one - probably not very cost effective.

The FW release on the other hand, I am putting to use already!

 

[ErikKaffehr]

Hi EJ,

A low base ISO would be consistent with less permissive filters in the CFA. Or, that could be like enhanced full well capacity.

In general, photographic devices use the linear part of the photon transfer curve and clip when leaving the linear part. It has been noted that the Nikon D810 seems to have a higher FWC (Full Well Capacity) than other sensors of similar generation, thus yielding less shoot noise and a lower measured base ISO. There has been some speculation that this may be enabled by linearisation of the pixel signal before writing to raw.

Whatever trick Nikon has used to reduce base ISO, Phase One could also do. There are three ways to do it. Increasing FWC, making CFA denser or reducing the pixel aperture. Increasing FWC is the most difficult path.

Best regards
Erik

Hi Steve,

My take is that the present information is full of hyperbols and no content. Very clearly, making the CFA design more orthogonal, that is having less overlap may give better color separation at some colors, like a Velvia look. A overlap free CFA would not be able to reproduce significant part of the color spectrum. So, what P1 describes is a very bad CFA design.

They also offer a an image to download, a totally unusable image.

So info is probably misleading... It can be a great design, but there is no relevant information, no relevant images. Just a lot of marketing hyperbols.

I am also a bit opposed to claims about highlight recovery in MFD.

• For one thing, it contradicts imaging science
• I also made some fairly detailed comparison between cameras available to me and found no visible or measurable difference in highlight reproduction between 2006 generation CCD and 2015 generation CMOS.
• Imaging science says no difference and there is no difference, in higlights.
• In shadows, the 2015 generation CMOS wins.

Higlights:

Shadows:

Regarding colour fringing, any colour fringing would probably result from weak IR or UV suppression combined by poor correction of the secondary spectrum outside the visual range. Lloyd Chambers has illustrated this for Leica S-lenses, but I have seen it demonstrated many times. Both CCD and CMOS are monochromatic devices, with the difference being in the readout method. The photodiode is not different between the different devices.

It is feasible that CCD designs for MFD would have less crosstalk than some CMOS designs for DSLRs. In that sense the old Kodak based CCDs were said to be very good.

Please don't misunderstand me, the Thricromatic may be great, it is just that the information coming from P1 is much like Pyongyang information or alternate facts and that would be a consideration if you are selling a 45k$US system.


Best regards
Erik

 

[ejpeiker]

Yeah I understand the reasons why it would be a lower ISO - my guess is just that the filter array, due to not allowing as much color overlap, just lets a little less light through to the pixel wells. For me it just allows a slower shutter speed but it's probably not that big of a deal since I use ND's a lot anyway and the change isn't enough to not do that.

 

[DougDolde]

So ISO 35 is something new? My IQ180 has it

 

[ejpeiker]

So ISO 35 is something new? My IQ180 has it

The CMOS backs are ISO 50 for the 100mp and ISO 100 for the 50mp

 

[Steve Hendrix]

Hi Steve,

My take is that the present information is full of hyperbols and no content. Very clearly, making the CFA design more orthogonal, that is having less overlap may give better color separation at some colors, like a Velvia look. A overlap free CFA would not be able to reproduce significant part of the color spectrum. So, what P1 describes is a very bad CFA design.

They also offer a an image to download, a totally unusable image.

So info is probably misleading... It can be a great design, but there is no relevant information, no relevant images. Just a lot of marketing hyperbols.

I am also a bit opposed to claims about highlight recovery in MFD.

• For one thing, it contradicts imaging science
• I also made some fairly detailed comparison between cameras available to me and found no visible or measurable difference in highlight reproduction between 2006 generation CCD and 2015 generation CMOS.
• Imaging science says no difference and there is no difference, in higlights.
• In shadows, the 2015 generation CMOS wins.

Higlights:

Shadows:

Regarding colour fringing, any colour fringing would probably result from weak IR or UV suppression combined with poor correction of the secondary spectrum outside the visual range. Lloyd Chambers has illustrated this for Leica S-lenses, but I have seen it demonstrated many times. Both CCD and CMOS are monochromatic devices, with the difference being in the readout method. The photodiode is not different between the different devices.

It is feasible that CCD designs for MFD would have less crosstalk than some CMOS designs for DSLRs. In that sense the old Kodak based CCDs were said to be very good.

Please don't misunderstand me, the Thricromatic may be great, it is just that the information coming from P1 is much like Pyongyang information or alternate facts and that would be a consideration if you are selling a 45k$US system.


Best regards
Erik

Hi Erick - I greatly respect the advanced minds on the forum, the cool thing about GetDPI is the varied backgrounds of expertise from so many contributors. I am not a scientist nor an engineer, but I do take a scientific approach from the standpoint of conducting the experiment to prove the theory. And my feeling so far from our time spent with the IQ3 100 Trichromatic is that I'm seeing the color improvement and I am not - so far - seeing colors that are un-represented vs previous models. Perhaps what Phase One is describing is one part of the equation.

At any rate, I'm not going to defend the information coming from Phase One. While we love the products Phase One creates, we acknowledge the messaging, but prefer to put the product to the test in the real world and arrive at our own conclusions to advise our clients. (and we've always done this from Day One).

I agree that highlight recovery, and/or the treatment of highlights (highlight recovery may be an inaccurate way to describe it) has not been the most dramatic aspect of image quality improvements over the years. But we'll - again - conduct our real world experiments and hope that they don't frustrate too many of the prevailing theories (though it is certainly a possibility).


Steve Hendrix/CI

 

[Steve Hendrix]

The CMOS backs are ISO 50 for the 100mp and ISO 100 for the 50mp

Excepting the IQ3 100 Trichromatic, which has a base of ISO 35.


Steve Hendrix/CI

 

[Paul2660]

I can't speak to the trichromatic 100 and highlight
recovery but the base IQ100 is amazing in regards to this. It has taken time to realize this but it has totally changed the way I shoot with this back.

It's unlike any other CMOS chip I have used.

Paul Caldwell

 

[ejpeiker]

Excepting the IQ3 100 Trichromatic, which has a base of ISO 35.


Steve Hendrix/CI

That was the premise of the post, the person I was responding to asked why we think ISO 35 is something new....

 

[Landscapelover]

I want real-world comparison of the landscape pictures. Fall color will be a perfect.
As EJ mentioned, it will be a very hard sell to the previous owners of the IQ3100 unless the color difference is significant and the upgrade price is very attractive.
I hope this is the hint of the upcoming 150mp DB, which will be much more exciting