Okay folks . Welcome to the Getdpi debate team
Here in one corner weighing in at SDHC 32gb we have Yup and in the other corner we have weighing in at CF 32gb we have Nope.
Let the debate continue 3 minutes per round for 3 rounds winner gets a free DVD disk. LOL
You can provoke some of the problems shown with and discussed in the context of the IQ180 with the P40.
Schneider 28mm, wide open (5.6), shifted from 0 to 25 mm, all on P40+.
The Schneider 43mm fares a bit better (if this is of any interest, I could post a similiar LCC sequence for the Schneider 43mm, shifted from 0 to 30mm).
I have no first hand experience of IQ180 files but, at the time being and with my limited experience, I don't think we see a new or up to now unknown problem.
It actually is the same thing no matter what back assuming there are no micro lenses. The critical angle is a function of the sensor pixel pitch. LCC "correction" code also neds to deal with the fact that the cells are not actually uniformly spaced and of consistent pitch; DALSA sensors have additional poly-silicon conductors spaced through the array which the raw processors need to take care of especially as acute angles cause them to cast long shadows.
My pet theory is that there is a specific angle which becomes problematical at a given pixel pitch. That angle determines the widest angle fov that can be accommodated. Note that this FOV is almost independent of sensor size since it is the angle that determines criticality. Thus a large sensor will go critical at a longer focal length than a small sensor but both will go critical at about the same fov for the given pixel pitch.
So what all the hubbub really is all about is folks adjusting to a 5.2 micron pixel pitch and trying to make sense of lenses based on favorites determined with a sensor of different pixel pitch and size.
I propose that a proportional slide-rule would do an adequate job of predicting the results.
I spent some time yesterday shooting "real world" landscapes with the Arca RM3D and Schneider 43 in front of my IQ180 back.
I won't bore you with the "ugly" LCC frames since they looked just like the others posted here other than to say I found that LCC using "Analyze Technical Wideangle" in C1 6.2 worked perfectly well up to 20mm of shift on this lens. At 20mm however, the lens itself starts falling apart to where the corners are very visibly bad. Probably about 17mm is the maximum practical shift with this lens on full frame, regardless of sensor. IMHO given the fact LCC is totally usable there with this lens on the IQ180, I think the combination issue is pretty much nullified. However, falloff is pretty extreme in this lens at 15mm of shift, showing about 2 stops at the shifted edge. Since falloff with that much shift can be in a significant portion of the image that includes highlights, the LCC falloff correction can cause an otherwise good full frame histogram to blow those highlights after correction. (Falloff in technical wideangle lenses can act like a Split Neutral Density filter for skies, and thus influence the histogram giving you false confidence. This is favorable for things like bright skies, but not necessarily for the rest of the tones in the image below the 3/4 mark.) Thus for this lens, I'd recommend getting a physical center filter to help reduce the load on the LCC for making the extreme correction by itself, also rendering a more realistic histogram spread. Other option is to leave about a full stop of exposure headroom if you shift 10mm or more.
Usage Note: Setting focus mask at 40 makes for a comfortable indication of accurate focus. Moving it up to 45 renders a tighter and narrower indication, and my preference might be 48 or so for those wanting maximum criticality. Confirming with a double-tap to 100% is still recommended until you get comfortable with reading the system.
I am trying to get my hands on a Rodenstock 40 HR-W in Arca mount to test this coming week.
Luckily I need the movements for the brute work of dealing with converging verticals, rather than stitching, so I can potentially sacrifice some resolution at the edges. I'd rather not, but them's the breaks.
The Schneiders' greater freedom from distortion still wins for me, with the inescapable trade-off of sensor casts. There's always some element of post-capture work to be done regardless of lens, camera, back or user.
Thanks very much Jack, Guy, Woody, Christopher and the guys at CI (have I forgotten anyone?) for all the testing.
Okay, here you go Woody:
This is pretty extreme ugly capture conditions from a technical standpoint for noise and LCC correction combined. 43mm Schneider shifted 12mm left, the max lateral shift in the RM3D I am using, shooting into a shady area so we get the max boost from the LCC at the dark corner. This is ISO 400 on the IQ180, and 1/15th at f11 on the lens. Note too that the Sun is just off camera right behind the cabin to the right basically shining on me as I shoot. Even though it isn't visible to the sensor, it is inside the lens' IC and tossing a boatload of light into the camera. That generates a strong flare in the upper LH corner and less in the upper RH corner. Noise is overly significant as I suspected, probably around net ISO 1600 in the corner after LCC boost. I would advise to use ISO 100 max if you plan on shifting -- and obviously, using ISO 35 would be even better. Note I used my standard (slightly warm) outdoor WB for all of these examples and did NOT dropper the passport! In addition to the noise making it unreliable at that small size, I have a bunch of light reflecting in on the passport from trees, dirty deck and parked cars.
Edit note: Since the Sun is shining basically on me and the camera form the front right, the LCC for this frame required the exact same exposure as the image to render a centered histo from the LCC. Using my normal 2 stops over, the LCC was pretty much all high gray.
First we have the un-corrected ugly to show I have the passport in a pretty bad area from an LCC PoV:
Now here is the LCC corrected frame. I am actually surprised (and sort of impressed) the LCC did not try to correct out the flare:
Here is the passport crop after the LCC. As you can see, noise is bad at probably around 1600 or more effective net ISO after LCC boost, and renders the passport basically unusable as a WB tool in that area of the frame. However, the overall saturation remains very respectable to my eyes, and not "killed" by the LCC as we feared it might be:
So I'll go on record with my conclusion:
If you push to the extreme outer limits of the lens IC or use extremely high ISO's with moderate shifts, you may not be satisfied with the LCC result. However, when used in realistic shooting conditions with proper technique for highest image quality, LCC corrects even significant lens and sensor color cast anomalies exceptionally well.
Thanks Woody -- yes, there is zero conversion noise I can find at base ISO.
Here is another exlemplar of how good LCC is. You see how bad the casts are in the above lateral shift. Here is two frames at each end of the full 12.5mm L&R shift in each direction. These look just as bad as the one above before LCC. I purposely put the tree in the center front so we could see any hint of parallax shift if there were any -- there isn't, it's a perfect stitch as we expect from tech cameras with rear shifts and lens remains stationary. Note it's a blustery day here, the wind is blowing, rain off and on and the Sun is moving in and out of clouds, so I had to work really fast to get relatively even lighting. My process was, capture far L, add 2 stops, capture LCC, shift full R and capture image, then add 2 stops and capture LCC. Process each LCC in C1 and apply to each image. Make sure each image has identical settings and their proper LCC, process out as full 16-bit tiffs. Talk about stressing the laptop LOLOL! I then brought them into CS for a merge and got this perfect result. Folks, this is pretty impressive, I did not use the center frame capture; it simply isn't needed. Final file would print out natively at 21 x 44 inches at 360 PPI, and is roughly a 17mm lens FoV in 35mm terms:
Jack- this one stitch has done wonders to put my mind at ease. For me my typical use of the tech camera would involve stitches like this along with minor shifts to help my framing.
Last edited by Terry; 29th May 2011 at 14:22.
Purple fringe -- good call. Maybe just the barest hint?:
Another useful part of the LCC correction tool in C1 is that there is a slider to control the amount of Light Falloff correction you wish to apply (0 - 120 %) which may be useful in striking a balance between the amount of "acceptable noise" and exposure boost from the LCC correction. This may may not be useful for all types of stitched images, however, but on single frame images it could help and I often prefer some vignetting anyway.
Some of us will recall that the early LCC algorithyms for the P65+ needed more work.
I earn 90% of my income with a tech cam, so I will be watching closely and testing lots.
"In the end, it's all about the pictures"
Sorry, I do not. A friend had one, but sold it to get the 43 they loaned me for this test.
Also, I want to thank Rod Klukas with Arca Swiss for loaning me his personal RM3D to do this test with. I am about 3mm shy of pulling the trigger and going with a tech kit for myself, and have decided if I do make the move, it will definitely be with the Rm3Di -- a truly beautiful camera, clearly designed for shooters to work quickly and accurately! Thank you again for the loan Rod, you may not get it back!
For those interested, Rod has posted a video of the RM3D features here -- advise good net bandwidth for viewing though: http://vimeo.com/24366528
Never mind Jack, thanks for the reply - I expect my back soon so I will run the tests on the 35 XL if no one else has managed it by then, and report back here.
A coda on the Rodenstock: I managed to install the Alpa Lens Corrector, a PS plugin. This was as slight challenge - the installation instructions say that it works with CS 3 and CS 4. What they mean by that is that it doesn't work with CS 5. I've seen this in at least one other case with a plug in that shows up in the "automate" submenu. So you're in luck if you have a legacy version of CS on your computer - I'm not sure what you do if you were a first time buyer of CS 5.
Anyway, it works and it works very well. It corrected the complex linear distortion of the Rodenstock just fine. It's fast and had no discernible impact on image quality. Of course the size and shape of objects near the edge of the frame are distorted in very wide shots (for the same reason the Greenland looks bigger than South America in a mercator projection of the earth) even though (or actually because) the lines are straight but that's not Rodenstock's fault - just a fact of life when you go really wide.
Here's a screen shot from LR showing after and before for the right hand frame in the pano that I posted previously:
The version on the left was my blog post for the day - it reflects a little extra work to smooth the light fall off and bring highlights into range which I didn't bother with on the distorted version.
The issue that I have with this lens (and maybe the other Rodenstocks) is that the images seem lifeless to me - probably because the designers have traded local contrast for sharpness.
Woody, ALC does work on CS5/32, not on the 64 bit version.
Looking back through this thread I am really confused. The primary reason I ordered an IQ180 is to quickly check tech camera focus on a good display. That would really broaden my capabilities.
If I am using a tech camera, it is because I need both quality optics and movements. Now comes the gotcha. The 180 has limitations with movements and, no, it is not just with the Schneiders. Shift a 40HRW by 15mm and color cast correction and local color/contrast are often challenged (to my eyes) even with a 50MP back. The 90 will take 15mm but no more. If I go beyond this, skies for example, will often retain a cast - this after shooting an LCC for each moved shot. I am left with all sorts of painful post to do.
So I have set my own movement limits for the 50MP - 15mm on the 90HRW and 12mm on the 40. What I want from Phase is to know that I can get 75% of that with the IQ180. That is what I am still confused about. Is the IQ180 just a wonderful technology for DF shooters, or is it a real advance for tech camera users who can accept only a bit less shift than they had with the earlier CCDs?
Yes, I can clear all this up for myself but by then I will have taken delivery. Right now, I still have options.
I greatly appreciate all the information being posted here, but this is a fundamental performance issue. Given that Phase positions itself as the best back for tech cameras, I would like to them take some official position.
I think Doug is doing that today actually from what I read.
Just a observation from everything I have read and seen I think the bottom line is your limits on movements will be reduced some and you will need proper LCC for everything you do to make those corrections. Basically your image circles have gone down in size and you just need to work within them more closely. You get outside the box it will get ugly.
I disagree, sorry. I must not have been clear enough in my earlier posts. To reiterate:
I estimated 17mm of shift before I saw image degradation at the corners with the 43 on my IQ180 -- this was from lens anomalies, and not sensor issues. I'm talking visiby soft corners at 20mm shift that looked like they'd be totally acceptable to me with 3mm less shift, or 17mm total shift. If you add a little for the diagonal on the frame for that rise, you get say 20mm on the IC radius. Now double that for the full IC diameter and you get 40mm. Add that to the IQ180's diagonal of 68mm to that and you get 40+68=108mm, or very close to the manufacturer stated 110mm of usable IC! IOW, I got virtually 100% of the usable IC of that lens! And to be clear, what I was seeing at the very corners at 20mm shift was resolution degradation, not LCC difficulties. In fact, LCC worked all the way to 25mm of shift, but the image was unusable in the edge zones, and again from lens anomalies, NOT sensor issues.
Howeverbutt!!! I was using C1's LCC controls, not somebody else's, and I was using the new "Analyze Technical Wide Angle" to render the LCC -- and that is a MUST for the IQ180 to get the best LCC.
Is that more clear now?
Doug Peterson (e-mail Me)
Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More
National: 877.217.9870 *| *Cell: 740.707.2183
Newsletter | RSS Feed
Buy Capture One 6 at 10% off
Jack, am I right that you are talking about 17mm shift left or right ? Did you tried it up ? so that you can see how the LCC corrections handles the sky ? Here was the main problem for me. I could not get good sky back with 17mm shift. Not withe C1 6.2.1
the IQ180 frame is approximately 40mm x 54mm. If I add 17mm rise and fall to the horizontal frame, I get a 74mm x 54 mm frame. The diagonal of that rectangle would be ~ 91.5mm. So that is the USABLE IC of the 43 on the IQ180. 91.5 is approximately 83% of the stated 110mm IC.
HOWEVER! I was able to shift the full 25mm and correct it, and though the image degradation rendered not useful image data, LCC appeared to work fine as respected sensor anomalies. Image quality was so poor however, I do not want to say I'm 100% certain that LCC worked. So with that qualification, if you accept very soft and distorted corners, the new frame would be 90mm x 54mm with a diagonal of 105mm, or very near the full stated IC of the lens; 96% of it. I suspect that Schneider is using the full circle of illumination rather than usable IC for its 110mm spec, unless they consider 2 or 3 LPmm acceptable resolution for the extreme edges .
I did NOT see problems in the sky, but my skies were unfortunately gray. I did however shoot the Passport out in the trouble zone, and did not see significant desaturation at 12.5mm on the horizontal. 12.5mm shift on the horizontal gives a frame of 40mm x 79mm and would have put the color checker at about a 45mm radius or 90mm position on the 43's total IC. I would say given how much color remained, it is going to be a relatively trivial issue.
Re how Schneiders v Rodies render. I personally prefer Schneider's look, however I feel the Rodies are actually technically sharper in the center and show more local contrast than Schneiders -- but in a way that leaves them looking almost sterile for lack of a better word; like drinking water so pure it has no taste. So if my "sterile" = Woody's "lifeless" we are in agreement. However, I also would say this is a really, really subtle call based on my looking at many tens of thousands of different images from different lenses over the years. Moreover, I think a lot of life can be added during post using some artistic license and secret sauce editing
I you sure about that ? From the specs: "In der Horizontalen lässt sie sich um 40mm und in der Vertikalen (siehe Abb.7) um 30mm mittels der seitlichen Drehknöpfe verschieben. "
in English it means as much as "30mm shift left/right and 40mm fall" which would mean yours should have 15mm as well ;-)