Phase One IQ3 100mp. Shipping now.

[torger]

The whole appeal of the back as a universal solution is pretty much at stake here.

I see it as a threat to the whole movement-centric view camera shooting style on the wide end.

I think this will happen: in the XF system the CMOS will become so popular that few new CCDs are sold, meaning that Phase One will not make CCD backs for the IQ4 series. That is these CMOS sensors is what tech cam industry must adapt to.

In the tech cam camp the 100MP and CMOS live view will be so sexy that many will choose to ignore the cast issues and thus the CMOS will sell well to tech cams too. Over time people will discover issues and "solve" them by shifting less, and start doing perspective corrections in post and such things. That is the tech cams become more like P1's A system, just a very sharp center frame snapper.

To get from here either a new generation tech lens friendly CMOS need to appear, which unfortunately I think is unlikely to happen in near time, or Rodenstock needs to make a set of new wide angle lenses which are stronger retrofocus. Schneider as we know has already left. There's also the Cambo Actar initative which we may see more of but this time for medium format size, that is that a tech cam manufacturer source in their own lenses just to get a working system.

The most likely scenario I think is a new set of wides, but how long will it take? It can be a few years, a period when tech cams may become known for their many issues on the wide end and lose popularity.

If I were Alpa, Cambo, Linhof, Arca-Swiss or Silvestri now I would be pretty disappointed because it will hurt their business that the highest end back will have quite serious issues with the wide angle lenses for their systems.

If I had an 60/80MP Dalsa now with my tech cam I would be really really really careful before upgrading to this. Note that I was also quite negative about the IQ250 wide performance and we could see that many still thought it was good enough for their needs... so we'll see how this one lands over time. As a large format shooting style fan I'm now certainly a bit worried about the situation though.

[Paul2660]

The 32HR can be pushed to almost 15mm +/- on the CCD backs, so would be nice to see more extreme samples.

Is it me or is there some strange vertical blotching in the sky areas of both 32HR images?  The edges are not quite well defined to be called banding, but there is somethings strange going on there.

Hi Peter and Jsiva,

The blotchy look you are seeing in the extreme shift, seems to have been carried over from the 50MP chip.  When I tested the 50MP on extreme shifts, with the 28mm and 40mm and 32mm (loaner), I saw this same issue.  It only manifested in solids, like this blue sky.  Made worse in B &W.

Opened a case with my dealer and Phase One on the issue back in March 2015 and nothing ever did come from it, however Phase did admit to seeing the effect and agreed it was not my lens or something in post I did to cause it.

I found that Topaz denoise can help smooth this out, as it has an actual de-banding selection. 

Interesting to see this problem again.  Note, this is not to be confused with microlens ripple which makes a much more solid stripping and is most times removed by the LCC.  In my work the LCC did not effect this blotchy banding effect.

Edit:   The 100MP does seem to handle the color cast issues on shift much better, if the smaller shot from the Alpa site is before the LCC was processed.  With the small 50MP chip all of the left side would have been visibly red, and you don't see that in the Alpa shot.  So good news there. 

Paul C

[Doug Peterson]

Patience please everyone. A broad range of good tests, preferably with raws shared rather than JPGs, are called for before conclusions are made.

Our first units arrive tomorrow morning. We have many tests to do, with tech cameras being just one (though one of the more important).

[torger]

Patience please everyone.

But I have no patience! Aaaaaaarrgghhhhhhh *running around in circles*

[Kevin Raber]

Chris Barrett and I will be shooting tomorrow with the ALPA system and numerous lenses.  I just did some tests shots with the 23mm Schneider and they are pretty damn nice.  It's a problem putting RAWs up on the Luminous-Landscape site because of the size.  Our download file system has a max of 100MB and the RAWs exceed that.  Our IT guys are looking into seeing if it is possible.  Chris and I will report back plus we'll have a more in depth report in the weeks to come.  If you can make it to Indianapolis and I'm around you're welcome to give it a try with me.

[torger]

A raw LCC shot of a shifted lens is telling if analyzed. The bayer array can be seen as groups of 2x2 pixels of one red, one blue and two green. In a shot with "normal" color casts green1 and green2 in each of those 2x2 groups should not differ much as they should locally see the same color. If the values start differing say more than 5% in the shifted areas this means that there's crosstalk, one green gets leaks from mostly red pixels and the other mostly from blue. This can't be cancelled out by a normal LCC algorithm and that's why you see the brown dead color in the shifted area of the SK43. The effect on the 32HR is much smaller but I think it's there unless there's some strange flare or post-processing effect.

You can still make images when this happens but you can get some hard-to-predict effects on your hands.

Also note that this sensor just like the IQ250 sensor probably have offset microlenses which assume that the center of the lens is in the center of the sensor. This means that it works best for center frames, when it's shifted large amounts the offset microlenses can get the reverse effect and hurt the performance rather than improving it. This means that when you make shift tests you should not just test say 0mm and then 20mm and interpolate the results inbetween as 20mm could be drastically worse than 15mm in the same imaged area. Better to test in say 5mm intervals.

Another good "sanity check" test is to make one shot with the back in normal orientation, and then make the exact same shot with the back upside down in a scene with stable light. If the LCC algorithm is unable after white balancing to present the same uniform color of both these shots then there's some substantial crosstalk going on. More saturated colors make it easier to detect.

If you have really severe crosstalk you can see "mazing" at pixel peep, but C1's demosaicer is really robust against this issue so then you really need huge amounts (like 30+ percent if I remember correctly)

The "danger" with these effects are that they can be quite subtle and if you don't know what to look for you may miss them only to discover it later on when you've already bought the stuff.

I'd also look for wavy microlens ripple effects, there was some of that on the IQ250 which was hard for the LCC algorithm to cancel out.

Also don't forget that Rodenstock lenses are more retrofocus the wider they are. That is the 32mm is more retrofocus than the 40mm. They've had some design target of maximum light angle and to meet that they've used increasingly complex lens designs (or smaller image circles) on shorter focal lengths. I don't know the exact differences but the point is that if you get great results of a 23mm center frame you can still have issues with a 10mm shifted 40mm. The popular SK60XL can also show problems with larger shifts as it's symmetric.

[Jeffery Salter]

@Jeffery, in the past, using the SK43 and the IQ260, it wasn't uncommon for me to shift up to 15mm.  With the SK35, I could get away with maybe 12.  So the 32mm HR may be a workable solution if it can go to 12 or 13mm.  The problem with architecture, though, was there was never a wider solution that was really workable and I always had to carry the 17 TS-E and a dslr for those occasional superwide shots.

Thank you Chris. I appreciate you sharing this info. When I get my tech cam my first lens will be an HR 40. (Mainly for landscape with people and without)  It’s great to learn something new.  Not sure when I will apply the knowledge, but it’s good to have it in my mental tool kit.

The 32HR can be pushed to almost 15mm +/- on the CCD backs, so would be nice to see more extreme samples.

Thanks Jsiva.

[Theodoros]

The most likely scenario I think is a new set of wides, but how long will it take? It can be a few years, a period when tech cams may become known for their many issues on the wide end and lose popularity.

If I were Alpa, Cambo, Linhof, Arca-Swiss or Silvestri now I would be pretty disappointed because it will hurt their business that the highest end back will have quite serious issues with the wide angle lenses for their systems.

It may... but it may also work the other way around (as it does to some extend now...). Meaning that the "clever" guys will buy the old lenses for cheap, add even cheaper the old 33, 39 & 50mp CCD sensor backs... and invest the difference on the cameras just producing jaw dropping images with "the good old fashion way" of focusing...

Heck... I bet the FUJI GX-680 will be the best camera & lenses to use with this back... but again, one has to still find a solution for a "true" WA then.... Wait a minute... where is that old 6x8 film back? 

[jsiva]

Hi Peter and Jsiva,

The blotchy look you are seeing in the extreme shift, seems to have been carried over from the 50MP chip.  When I tested the 50MP on extreme shifts, with the 28mm and 40mm and 32mm (loaner), I saw this same issue.  It only manifested in solids, like this blue sky.  Made worse in B &W.

Opened a case with my dealer and Phase One on the issue back in March 2015 and nothing ever did come from it, however Phase did admit to seeing the effect and agreed it was not my lens or something in post I did to cause it.

I found that Topaz denoise can help smooth this out, as it has an actual de-banding selection. 

Interesting to see this problem again.  Note, this is not to be confused with microlens ripple which makes a much more solid stripping and is most times removed by the LCC.  In my work the LCC did not effect this blotchy banding effect.

Edit:   The 100MP does seem to handle the color cast issues on shift much better, if the smaller shot from the Alpa site is before the LCC was processed.  With the small 50MP chip all of the left side would have been visibly red, and you don't see that in the Alpa shot.  So good news there. 

Paul C

Thanks Paul.  I think you are right regarding similarities with the 50MP chip. 

Yes, the smaller inset image appears to be quite good, as in no visible magenta/red.  Flip side though is noise and still strong light loss in the shifted part of the image post LCC.

[Doug Peterson]

Chris Barrett and I will be shooting tomorrow with the ALPA system and numerous lenses.  I just did some tests shots with the 23mm Schneider and they are pretty damn nice.  It's a problem putting RAWs up on the Luminous-Landscape site because of the size.  Our download file system has a max of 100MB and the RAWs exceed that.  Our IT guys are looking into seeing if it is possible.  Chris and I will report back plus we'll have a more in depth report in the weeks to come.  If you can make it to Indianapolis and I'm around you're welcome to give it a try with me.

DT would be glad to host the raws on our website Kevin. Hit me up by email.

[JV]

Hi Doug,

Would you ever use this back for weddings?

Thanks, Joris.

[JoeKitchen]

When looking at the 32mm sample from Alpa, it gave me the same feeling that I would get when looking at a P45+ file with my SK 35mm shifted with no CF.  It is not pretty, but using a CF cleans up the file very nicely. 

Considering the Rodie 32mm has 3 stops of light fall off (compared to the 3.5 stops with the SK 35mm), I have to assume (and hope) that image was captured without a CF.  This could bode very well if this is the case; the file could be very clean with the CF used. 

This back could work very nice with a tech camera if so. 

I would say that with a lens as wide as the 32mm on a full frame MFDB, I would only shift 6/8 mm for most interiors and many exteriors.  Yes, there would be exceptions, like in lower Manhattan where the streets are very thin and the interiors very small, but overall it would be produce an acceptable quality, probably better that any 35mm solution.

Anyway, the Rodie HR 40mm and Rodie 55mm is where I am most interested.  Those are my sweet spots for interiors. 

I am really curious to see how the resolving power (or lack thereof) of the Rodie 55mm holds up with such small pixels.  With my P45+, it is very sharp until about 15/18 mm of shift. 

[tjv]

I am really curious to see how the resolving power (or lack thereof) of the Rodie 55mm holds up with such small pixels.  With my P45+, it is very sharp until about 15/18 mm of shift.

With my Credo 60, the Rodie 55mm is pretty average when shifted beyond 10mm so I expect with this back–particularly if these early test samples are anything to go by–it'd very quickly show it's age off centre, both with regards to sharpness dropoff, but also luminance falloff. I wish Rodenstock would design a new 55mm lens upto task of the modern, 80+ mpx sensors.

[Ken R]

I think the 40mm HR-W is a stunning lens. It behaves well and it is not too big, it is easy to filter (no need for a CF) and at least with my IQ160 it allows a large range of movements. I use it basically up to the 15mm limit (L-R) on my RM3Di. If I have some rise/fall dialed in along with the 15mm shift I might hit the limit of the image circle in the very corners but the image quality is superb almost up to the very sharp edge of the disc. The 70mm HR-W allows even more movements and it requires even less extreme L.C.C.'s. Very very mild ones in fact.

My bet is that these two lenses are gonna work beautifully with the IQ3 100MP back. Looking forward to the test results. 

[Joe Towner]

Our download file system has a max of 100MB and the RAWs exceed that.  Our IT guys are looking into seeing if it is possible.

This is a great excuse for Dropbox links - let them deal with the heavy transfers.

[torger]

I would be surprised if Alpa has shot the 32HR without CF, but of course we don't know. If they have noise will clean up somewhat, but the non-uniformity color issues will not. I've learnt though that uniform color is to many a lot less important than corner pixel-peep sharpness so it will be very exciting to see how much issues people are willing to accept.

[torger]

I forgot to mention one more sanity check test that's really useful --- sensors may perform very differently in portrait vs landscape orientation, depending on how wiring is oriented on the chip. That is you can have lots of crosstalk in one orientation and much less in the other.

So when testing shift performance you should test in both portrait and landscape orientations.

[BernardLanguillier]

Team,

I have been playing with the city raw sample image in C1 Pro 9.02. I'd be curious to hear what settings you guys are using for the IQ3 100 files?

Personally, here is what I did:
- ICC profile: IQ3 100mp Outdoor daylight
- Curve: High contrast
- WB: 7750, Tint: 24
- exposure: +0.3
- Contrast: 15
- Brightness: 20
- Saturation: 5
- Hilight recovery: 60
- Shadow: 30
- Clarity: 30
- Structure: 5
- Sharpening: amount 400, Radius 0.3, Threshold: 0.8

I did also apply verticals correction and a few layers to selectively darken the sky.

It looks much better to me than the initially provided settings. What did you guys do if I may ask?

Thanks.

Cheers,
Bernard

[jduncan]

Hi,

Hasselblad used to make multishoot, but according to Doug their lenses are not up to the task, so why even try?

Another way to see it that Hasselblad had a 200 MP product for a long time and that actually worked well under benign conditions.

Best regards
Erik

Hasselblad still does multi-shoot, and multi-shoot does not stress the lens any more than a single shoot.
So if the lenses are good for a 50mpixel capture, they will be for the 200mpixel capture of the H5D-200c.

You insist that the product is not for sale when it is, based on the  50mpixel cmos sensor.
http://www.bhphotovideo.com/c/product/1078381-REG/hasselblad_3013708_h5d_200c_ms_digital_camera.html
The new machine performs far better than the old one.
Phase one has a very good  camera, and as always fantastic  software and backs. There is no longer a need to a need for this kind of stuff.

Best regards, 

[ErikKaffehr]

Hi,

You take make posting the wrong way. Doug said that the Hasselblad lenses were not good enough for the new sensor. So I made the point that they worked fine with multishot, which yields 200 M. But you are dead wrong that if a lens is good enough for 50 MP it will also be good enough for 200 MP multishot. The other way around is quite true.

Just consider this, you cannot resolve something by multishoot that is not projected by the lens.

Best regards
Erik

Hasselblad still does multi-shoot, and multi-shoot does not stress the lens any more than a single shoot.
So if the lenses are good for a 50mpixel capture, they will be for the 200mpixel capture of the H5D-200c.

You insist that the product is not for sale when it is, based on the  50mpixel cmos sensor.
http://www.bhphotovideo.com/c/product/1078381-REG/hasselblad_3013708_h5d_200c_ms_digital_camera.html
The new machine performs far better than the old one.
Phase one has a very good  camera, and as always fantastic  software and backs. There is no longer a need to a need for this kind of stuff.

Best regards,