IQ180 with rm3di - can the lenses keep up?

[Jim2]

hi, just a curious question for those who are technically knowledgeable in this area. I'm probably not asking this question right so please pardon my ignorance. Can the current line up of lenses for Arca rm3di still resolve to the fullest ability of IQ180 sensor? What about tilt / shift? Thanks for your info.

[yaya]

hi, just a curious question for those who are technically knowledgeable in this area. I'm probably not asking this question right so please pardon my ignorance. Can the current line up of lenses for Arca rm3di still resolve to the fullest ability of IQ180 sensor? What about tilt / shift? Thanks for your info.

Some answers here: http://www.luminous-landscape.com/forum/index.php?topic=52064.0

[design_freak]

Lucky you   Nice toys
The Answer is Yes

 

[AlBowers]

Jim,
Not a direct answer to your question, but more a general statement about lens resolution vs digital capture resolution. Because of the nature of light, you really do want an imbalance between the full resolution of the lens vs the back. In the digital signal processing world, you'd ideally want the capture to be at least a factor of 5 above the signal you're trying to capture. Its when you get close in resolution between the two that you get lots of aliasing of the signal (moire). As a general statement, more resolution in the capture is good, and we should hope to overwhelm the lenses (as long as we can stand the file sizes).
I have done some digital signal processing as a part of my "day job" for almost 3 full decades. And image capture is no different than trying to characterize very high frequency modes in gas turbines, flutter, or any other vibratory frequency analysis. [OT: I work at a flight research facility...]
Al Bowers

[Jim2]

Jim,
Not a direct answer to your question, but more a general statement about lens resolution vs digital capture resolution. Because of the nature of light, you really do want an imbalance between the full resolution of the lens vs the back. In the digital signal processing world, you'd ideally want the capture to be at least a factor of 5 above the signal you're trying to capture. Its when you get close in resolution between the two that you get lots of aliasing of the signal (moire). As a general statement, more resolution in the capture is good, and we should hope to overwhelm the lenses (as long as we can stand the file sizes).
I have done some digital signal processing as a part of my "day job" for almost 3 full decades. And image capture is no different than trying to characterize very high frequency modes in gas turbines, flutter, or any other vibratory frequency analysis. [OT: I work at a flight research facility...]
Al Bowers

Correct me if I'm wrong please, but I was under the opposite impression that currently the lenses should have more resolving power than the available sensors, up until 80mp - which is why I'm curious.

[Christopher]

I'm pretty sure that all my lenses will work nicely with 80Mp and would even work on 120Mp. These are HR-W 32mm Rodenstock, HR-W 40mm Rodenstock, 43mm Schneider, HR-W 90mm Rodenstock. Sure at some point you will loose some sharpness when shifting to the limits, but I wouldn't worry about it to much.

[Wayne Fox]

Correct me if I'm wrong please, but I was under the opposite impression that currently the lenses should have more resolving power than the available sensors, up until 80mp - which is why I'm curious.

There seems to some disconnect by many about the relationship of the ability of a lens to resolve detail and the resolution of the sensor. I'll freely admit I'm not an engineer, but as with any analog data, there is no real correlation between that data and a digital representation.  As with pretty much anything, to obtain the highest quality requires substantial oversampling of of the data. There is no 1 to 1 relationship of the analog detail and the sensels on the sensor. It's interesting that we've been talking about lens outresolving sensors for many years now, and most thought we would hit that point years ago.

I would assume there is a point where sufficient oversampling can be obtained that going further would gain nothing, but I don't think we are there yet. It's interesting that many testers of the IQ180 claim the resolution difference is more significant moving from the p65 than it was when moving from the p45 to the p65, even though it represents a lower percentage increase in resolution.

I see this debated frequently.  I personally agree with Al that the goal is to get to where the lens is the limiting factor, not the sensor.  Having too much sensor resolution won't affect image quality, but not enough certainly does.  I hope we can get to the point that sensor resolution is high enough we can eliminate AA filters without issues, and maybe even create sensors that work differently and not use Bayer filters and algorithms.

Of course this assumes lens won't get any better ... but how can anyone assume anything anymore?

 Ctein over at TheOnlinePhotographer.com always comes up with some interesting viewpoints.  He's a smart guy so his opinion certainly has some merit (I won't debate how much, but he certainly knows a lot more than I do about things)

Metamaterials and Photonic Crystals

Why 80 megapixels won't be enough

[AlBowers]

Just a quick thought experiment. Let's just for sake of argument say a MF lens can resolve 60 lpmm (that's 120 lines total, because every line PAIR is a black line and a white line). So we're at 120 lmm. So across a 54mm frame, that's a total number of 6480 lines. So we'd need 6480 pixels to capture those 6480 lines (assuming we had perfect alignment to prevent any aliasing/moire).
The 80 megapixel sensors are approaching double that number (and remember the lenses tend to fall off in resolution near the edges too). It sounds to me like we'd see the increasing resolution stop somewhere in the 160-200 megapixel range with the current lenses we have. Even if lenses were to improve significantly (to say 100 lpmm) we'd start to see the fall-off in improvement near the upper end of that range (~200 mega pixels).
Now this is an engineer's approach to a very subjective criteria. So it might be that we are not "close enough" even if the above were to come true for many users. I believe there is still significant room for improvement...
Al Bowers  (...who exposed a significant number of silver-halide crystals to light this weekend...)

[Jim2]

So across a 54mm frame, that's a total number of 6480 lines. So we'd need 6480 pixels to capture those 6480 lines (assuming we had perfect alignment to prevent any aliasing/moire).

Don't you at least need twice the sampling i.e. you need 2x6480 pixels to properly sample 6480 lines?

[AlBowers]

Jim,
I think I got the factor of two already. As I originally stated "Let's just for sake of argument say a MF lens can resolve 60 lpmm (that's 120 lines total, because every line PAIR is a black line and a white line). So we're at 120 lmm." So I already captured the "factor of two" that we need (I think). I think the number of 6.5k is about right for the total number of pixels. Of course, if the resolution is much higher from the lens, say 200 lmm (that's 100 lpmm) then the you need about 10800 pixels (albeit perfectly aligned) to resolve across the field of the lens.
Keep in mind:
lpmm = line pairs per mm or lp/mm
lmm = lines per mm or l/mm
At my suggested Nyquist number above, that's 87.5 megapixels. At 350 mega pixels we have twice that resolution. Perhaps my original estimate was a bit low. So we might see a significant flattening of the IQ vs pixel count curve in the 500 megapixel range? Those pixels would be about 1.13 x 1.13 micron sized (about 48,000 pixels across a 54mm frame). That's 4x smaller than our current best 5.2 x 5.2 micron pixels...
Al