Digital Sensor "Centerfold" correction

[wentbackward]

I've experienced this problem in the past, in particular with the HR40 on an IQ160. In fact I'm on my 3rd back, however the problems were much worse than shown here and were all visible lines showing, not just the centre one.

 

[Pemihan]

So the back you're having now doesn't show this?

Peter

I've experienced this problem in the past, in particular with the HR40 on an IQ160. In fact I'm on my 3rd back, however the problems were much worse than shown here and were all visible lines showing, not just the centre one.

 

[Pemihan]

Here is what I'm seeing.. Cambo WRS, DM33 (Leaf Aptus-II 7) SK47XL. Just a tiny bit of shift.
The first image has had no LCC applied, the second has, using Leaf Capture and converting to B&W in Lightroom with Nik Silver Efex Pro.
Is this what to expect? Is everybody who do a lot of B&W on a tech cam seeing this?

 

[Wayne Fox]

I think the greyscale conversion on mid to lighter tones can accentuate the sensor seams. Also if you push too hard in processing C1 it can cause it to be more noticeable.

I had this problem last March, and did send my IQ180 in for recalibrating, and it did help ... at least I haven't noticed it since.

 

[Pemihan]

Thanks Wayne, I'm really starting to think my back is in serious need of a recalibration...

Have anybody else experienced something like what I'm seeing?

Peter

 

[GrahamWelland]

Not as severe but I'm in the same boat - mainly because I'm doing more B&W vs colour I think and therefore more sensitive to this.

 

[wentbackward]

So the back you're having now doesn't show this?

Peter

Peter, Yes, my current back seems fine.

 

[GrahamWelland]

Doug,

One of the issues I have with the KB article listed is that it covers the edge fall off and fine line introduction associated with the wides and light angles but how does that apply to the centerfold concerns? I know in my own case that I'm running in to the issue with a 150mm SK XL lens with no shifts whatsoever which wouldn't really fall in to this use case whatsoever.

Also, it would be better if that tech note differentiated between LCC correction with & without the use of a centre filter too. This makes a significant difference in my experience at least with the SK 35 XL with the IQ160. Shifts need to be limited certainly if you want to avoid extreme colour issues (I never go beyond 10mm for example).

 

[Pemihan]

Graham, did you ever get a solution to the problem?

Peter

 

[etrump]

I think the greyscale conversion on mid to lighter tones can accentuate the sensor seams. Also if you push too hard in processing C1 it can cause it to be more noticeable.

I had this problem last March, and did send my IQ180 in for recalibrating, and it did help ... at least I haven't noticed it since.

I understand the calibration can be done remotely with an LCC exposure.

 

[stngoldberg]

Several years ago, Woody posted about his problem with "centerfold' on his HD60, and I discussed the issue with him I think in a PM. He correctly assured me that I wouldn't experience the "centerfold" with a tech camera and an HD50. I subsequently bought an Arca Swiss RM3DI and have been centerfold free.
Woody was the first person, I think to discuss this problem, and thanks to him I escaped it.
Stanley

 

[etrump]

My IQ180 is coming back from Phase this week after being calibrated. It got to the point where even without adjusting the image the lines were visible, especially in an LCC. A slight saturation increase would exaggerate the effect.

Seemed like it did increase with the age of the back.

 

[ondebanks]

In the past, when we were discussing the possibility of a square (eg 48mmx48mm) CCD, it was often considered that the cost would be prohibitive. I think at that time, few of us realized that the CCD was actually being "stitched" together?

For this problem, the root cause is the stitching. What the software/firmware can do is just to "hide the line". Because there is no physical CCD area overlapping for this kind of "stitching", the tolerance must be very high.

It's not a problem of just the thin join line where the sensor segments abut - you wouldn't notice that. It's a problem of entire halves/quadrants/sixths of the sensor being read out through different channels, with their own amplifier characteristics.

Ray

 

[ondebanks]

Woody,
It's not really just an IQ issue. Any camera can exhibit this if abused sufficiently.

If by "camera" you mean digital back or sensor, that is not always the case.

As I explained here, some Kodak-sensored backs will never show this problem, because there is only one readout channel for the entire chip.

Dalsa sensors are more likely to show it because they use multi-channel readout architecture in all of their CCDs. That makes them faster to read out, but the penalty is what you see here.

I hope that with recalibration, it can be cured to everyone's satisfaction.

Ray

 

[Kolor-Pikker]

If by "camera" you mean digital back or sensor, that is not always the case.

As I explained here, some Kodak-sensored backs will never show this problem, because there is only one readout channel for the entire chip.

Dalsa sensors are more likely to show it because they use multi-channel readout architecture in all of their CCDs. That makes them faster to read out, but the penalty is what you see here.

I hope that with recalibration, it can be cured to everyone's satisfaction.

Ray

It also depends on other factors like how the readout channels are laid out, and how fast the data can be read, for example DSLRs have 8~12 readout channels that read data off the sensor in rows, and you never see this kind of anomaly.

If medium format sensors went over to using a more tried and true method such as this one, this wouldn't be a problem. I'm honestly surprised that many MF sensors are read in quadrants, though it could be a limitation of CCDs.

After reading "The story of the IQ260" on LuLa, I get the impression that Phase One pushed the limits of CCD sensor design with the IQ260, but it seems to me as effective as developing a new CRT monitor using 2013 technology. Potentially very good in some aspects, but inherently flawed.

What I'm leading to here is that I often see the argument that new sensor tech is difficult to develop in a high-margin/low-volume market, but take a look at what RED is doing in the world of cinema. They started in roughly 2006, funded out of the owners pocket (admittedly a deep one), and though thick and thin, and people calling them scam artists for having the audacity to declare that they will sell a $30,000 or so camera that shoots 4K Raw video, what do we have now 6 years later? A 6000x3000 pixel, 30x15mm sensor that has a solid 16-17 stops of dynamic range and can shoot at 100 frames per second, developed, manufactured and assembled from scratch 100% in the USA. And they want to go for bigger sensors.

Unlike RED, Phase One and others are long-established players in imaging technology, which makes me wonder why one doesn't attempt to break the flow and develop a genuinely out of this world sensor or camera design. The new RED costs $30-50k depending on configuration, which is similar to digital back prices, and you could argue that the number of film makers that can afford one is similar to the number of photographers that can buy a digital back - that is, if a thousand units sell worldwide, it can be considered a runaway success

Sorry for the tangent.

 

[etrump]

My IQ180 back just returned from calibration at Phase and it now shows absolutely no centerfold effect even when cranking full saturation or clarity sliders. Prior to returning it, almost any adjustment would exaggerate the problem.

 

[Pemihan]

Ed, are you using Schneiders or Rodies?

Peter

 

[etrump]

Ed, are you using Schneiders or Rodies?

Peter

Both

 

[Pemihan]

I got my DM33 replaced with a new Aptus II 7, which is much better, but I still see the centerfold, especially when shifting, but some times also straight on.
Contemplating sending it in for calibration..

 

[AreBee]

Can the Leica S2 and S suffer from the centrefold effect, given that they also use a CCD sensor?