I think that the chnces of the NEX7 beating hte IQ of the M9 are essentially zero. But if it improves materially from the NEX5, it will partly close the gap. And, with the Zeiss 24mm it should open one relative to the very successful Fuji X-100. So it looks promising...
Jerry_R wrote:
I think - it would matter more, if people would use M lenses at apertures like f/5.6 or f/8.
But very frequently - they use them wide open. And that will not be reproduced by NEX or RICOH.
It really just depends on the user, like you said. It's only about a stop difference between 135 and aps-c, so, if you have a style dependent on working with the shallowest dof possible, and you're shooting lenses on 135 in the f.95-f1.2 range, then the larger format is obviously the better choice. However, if you're someone who stays nearer the large end of the aperture scale, but not necessarily wide open all of the time, the difference may not be noticed too much. With my 135 cameras, I tend to shoot in the f2-f4 range, so adjusting to the f1.4-f2.8 on NEX is no big deal for me. If I was just dying for shallower DOF, I'd get a couple of f1.2 or larger lenses, but I haven't needed it...although for fun I may get that new 35/1.2 coming from Voigtlander.
All of this being said, for me, m4/3's sensor is small enough where I would notice the difference, but to each his/her own limit.
I thik there's actually more than a stop difference between 135 and ASP-C. I'm using the DOF calculator for this, so the values are obviously approximations:
Suppose we have a 35mm f/2 lens on full frame and we shoot a subject that is at a distance of 100 cm. We get a DOF of 9.5 cm.
So:
35 f/2, FF -> DOF 9.5 cm
On a 1.5 cropper we would need a 23.3mm lens to match the angle of view. What aperture would we then need to match the DOF of the FF shot?
23.3 f/2 -> DOF 22 cm
23.3 f/1.4 -> DOF 15.5 cm
23.3 f/1 -> DOF 10.9 cm
so after two stops we're still not there. We have to go one half stop more to f/0.95 to get a match. So there's a two and a half stop of difference in the example above.
DOF is however only part of the equation. You also have the sharpness->blur transition which does depend on focal length. That's why you with a FF camera can't really get that medium format look. Furthermore it's much easier and cheaper to make a good 35mm f/2 lens than a good 23.3mm f/0.95. So even if you could get faster glass for the cropper to compensate, image quality wide open will subsequently suffer.
Anyway, I was out shooting video today with the NEX-C3. I'll put something together later and post it. It was kind of silly having that miniscule camera on a big video tripod but I didn't really have a choice as my video gear uses different quick release clamps from my photo gear. I did not bother with an external monitor though (although the C3 has a HDMI output port) which was possible thanks to the peaking function.
I've looked at a couple of video files and the good thing is that they're very flat which is great for PP. The bad thing is that the compression is atrocious. Obviously the C3 isn't intended as a serious video camera as it has no manual controls but I was still curious how it would perform and I wanted to shoot some video with Leica M glass. That last thing wasn't all that brilliant though. The focusing tabs on some of the Leica lenses means that you can't attach a follow focus to it. This in turn means that you can't get good repeatability for focusing moves and you lose some stability when you pull focus directly on the lens.
denoir wrote:
I thik there's actually more than a stop difference between 135 and ASP-C. I'm using the DOF calculator for this, so the values are obviously approximations:
Suppose we have a 35mm f/2 lens on full frame and we shoot a subject that is at a distance of 100 cm. We get a DOF of 9.5 cm.
So:
35 f/2, FF -> DOF 9.5 cm
On a 1.5 cropper we would need a 23.3mm lens to match the angle of view. What aperture would we then need to match the DOF of the FF shot?
23.3 f/2 -> DOF 22 cm
23.3 f/1.4 -> DOF 15.5 cm
23.3 f/1 -> DOF 10.9 cm ...Show more →
those numbers sound pretty crazy to me, and when i've used DOF calculator i've gotten values that are clearly way off from actual dof. i wouldn't make any conclusions based on what you get from them. using a rokkor 45/2 on my film camera and 28/2 on my NEX 1 f-stop of difference seems to be about right, probably closer to 1.2. i think at macro distances that fall apart though...
There is a handy lens equivalent calculator here, at the very bottom of the page:link
35mm f2 on 135 = 22.7mm f1.3 on NEX.
I prefer the 50 (equivalent) focal length, myself, and 50mm f2 on 135 = 32.5mm f1.3 on NEX.
Either way, it's not as if I only shoot my medium format cameras wide open, either, so it's not a huge adjustment. Shooting only wide open gets a little one-note, to me. In fact, I'm not sure I've ever looked at any of my pictures and thought they were ruined because I couldn't open up just one more stop, at least in terms of dof.
But, if one is banking personal style on an f1.4 lens wide open on 135, surely aps-c is not the way to go, unless someone is willing to drop the cash on a Noctilux for NEX.
Both calculators can't be right. So I decided to do a test - Zeiss 100 MP @ f/2 on FF vs Rokkor 58/1.2 @ f/1.2 on 1.6 crop. The distance was about a meter. I barely had to move the tripod to get an identical FOV (the Zeiss is acutally shorter than 100mm and the Rokkor seems to be a bit longer than 58mm).
Open images in separate tabs in the browser and flip between them:
Lens design plays a large role in how the DOF will be, so this is not conclusive, but at least in this case the Rokkor at f/1.2 has waay more DOF than the Zeiss at f/2. And that's a stop and a half difference. So I reckon at least two stops difference in this case, probably more.
Denoir, your usage of a DoF calculator shows very clearly why I recommend that people stay away from stuff like DoF calculators until they actually know what they're doing.
So, you'd need a lens that was ~23.3mm F1.3. Which lines up perfectly with optical theory.
What would hurt the smaller sensor performance at large apertures is of course that the bigger lens at the bigger sensor is used at F2.0. On the smaller sensor you BOTH have a higher optical strain (to get the same sharpness in the resulting picture, you need higher optical resolution since the enlargement factor is higher) and you're using the lens at the APS sensor at a larger aperture > harder to correct, probably a lot lower in both contrast and detail sharpness.
1) I pick a focal length L and aperture A and a distance D. From the DOF calculator I get a total DOF, D_1.
2) I calculate a second focal length L2 = L/cf where cf is the crop factor. So I get the focal length L2 I would need to use on a crop camera to get the equivalent angle of view as L on a full frame camera at the same distance D from the subject.
3) I enter L2 and D into the DOF calculator and change the aperture parameter until my calculated DOF = D_1. That aperture is the 'equivalent' one for the cropped sensor.
What am I missing? Assuming you know your optics, and I assume you do, I have either done a logical error in the process above or the DOF calculator doesn't work correctly.
denoir wrote:
Both calculators can't be right. So I decided to do a test - Zeiss 100 MP @ f/2 on FF vs Rokkor 58/1.2 @ f/1.2 on 1.6 crop. The distance was about a meter. I barely had to move the tripod to get an identical FOV (the Zeiss is acutally shorter than 100mm and the Rokkor seems to be a bit longer than 58mm).
Open images in separate tabs in the browser and flip between them:
Lens design plays a large role in how the DOF will be, so this is not conclusive, but at least in this case the Rokkor at f/1.2 has waay more DOF than the Zeiss at f/2. And that's a stop and a half difference. So I reckon at least two stops difference in this case, probably more....Show more →
first: i'm not seeing much difference between how much is in focus. i am seeing quite a difference in how blurred and what type of blur the oof regions are.
second: if you want to do this test well, you should have both lenses stopped down at least a bit where apertures tend to be more accurate (max aperture labels can be a bit optimistic) and blur transitions more consistant.
edit: also use a camera with 1.5x crop factor rather than 1.6x.
When I had NEX, I did attach 35 f/1.4 to NEX and 50mm 1.4 to M9. The difference was as on above samples, it is not only a question of resolution, dynamic range, other sensor properties...
35mm lenses on FF sensor act differently, than wider counterparts on cropped sensor.
sebboh wrote:
second: if you want to do this test well, you should have both lenses stopped down at least a bit
That would make no sense to me... When I also wanted to answer myself question, to what extent NEX can be a backup for M9, I was MOTSLY interested in wide open image quality.
If I would need sharp pictures with large DOF - then u43 acts much better than NEX - taking into account its existing lenses.
Since you're comparing "same frame", i.e you're trying to find a lens that will do the same thing on APS a another does on FF, the CoC's change. A CoC of 0.03mm is 1/1200 of the frame width on FF. To get the same on a Canon APS you divide sensor width by 1200 > 22.2mm/1200 = 0.0185mm
"Same Frame" for me means:
Same picture vantage PoV
Same picture angle (view angle)
Same DoF
-i.e, standing in the same spot, getting the same picture coverage and getting the same DoF.
How big a blur disc is is always related to the output, the finished picture. So the CoC needs to be correlated as in "parts per picture width, height or diagonal", not as a constant.
I don't doubt your test results with the Rok/Zeiss, but spherical aberrations do very greatly affect close range DoF - and the character of the oof areas. The Zeiss, being close range corrected is probably a LOT sharper in the DoF than the Rok is? When you look at the pictures scaled to the same size (both at 10MP or whatever).
I think you own a 70-200/2.8IS2 too. That would be a more "natural" way to compare, since it doesn't change in "character" as much as two different primes, and you can dial in a perfect "same frame" just by zooming in with the larger sensor 5D2. It also has 1/3 Ev stops of aperture.
If you do the "book" test again, with the 70-200 bolted to the tripod and just change cameras without moving the tripod, then you get exactly what you're supposed to get: if you compare 7D with 5D2 you get 70mmF2.8 = 110mmF4.5. At least we did the last time I demonstrated the "FF effect" to someone... :-)
sebboh wrote:
first: i'm not seeing much difference between how much is in focus. i am seeing quite a difference in how blurred and what type of blur the oof regions are.
You're not seeing that you can't read half the text in one shot and not in the other?
second: if you want to do this test well, you should have both lenses stopped down at least a bit where apertures tend to be more accurate (max aperture labels can be a bit optimistic) and blur transitions more consistant.
Possibly, the problem is that even at these large apertures you need to get close up for a simple layout like that to work. I'd basically have to take macro shots.
edit: also use a camera with 1.5x crop factor rather than 1.6x.
It would only lead to greater inaccuracies. The 5DII and 7D have the same form factor, same focusing method, same RAW developer etc. Using the NEX would just complicate things. And you're not going to see any drastic differences between 1.6 and 1.5 crop..
douglasf13 wrote:
I think your missing piece is that the circle of confusion for aps-c is different than 135.
You may be right, but in that case there is more that I don't understand - something fundamental.
I can't see how the circle of confusion would be different unless you took into account the pixel pitch of the sensor. We're talking about a projected image on a surface - how on earth could there be a change in the projected image just because the surface is larger?
Take the M9 and M8 sensors as an example - they have identical pixel pitch. The M9 sensor is just larger but otherwise they're the same. How could a pixel in the center, or at the edge for that matter care how much sensor there is left on the sides? The underlying assumption when you say that ASP-C has a different CoC than FF 135 must be that they have the same number of pixels on different areas - i.e that the ASP-C has a higher pixel density. Am I correct?
theSuede wrote:
Since you're comparing "same frame", i.e you're trying to find a lens that will do the same thing on APS a another does on FF, the CoC's change. A CoC of 0.03mm is 1/1200 of the frame width on FF. To get the same on a Canon APS you divide sensor width by 1200 > 22.2mm/1200 = 0.0185mm
As I said to Douglas - doesn't this assume the same total number of pixels on both sensors? I can't see how the light projected from the lens on the sensor could change if you change the size of the surface it is projected on... Am I missing something?
theSuede wrote:
I think you own a 70-200/2.8IS2 too. That would be a more "natural" way to compare, since it doesn't change in "character" as much as two different primes, and you can dial in a perfect "same frame" just by zooming in with the larger sensor 5D2. It also has 1/3 Ev stops of aperture.
If you do the "book" test again, with the 70-200 bolted to the tripod and just change cameras without moving the tripod, then you get exactly what you're supposed to get: if you compare 7D with 5D2 you get 70mmF2.8 = 110mmF4.5. At least we did the last time I demonstrated the "FF effect" to someone... :-)...Show more →
That's a good suggestion, I'll do that. I would however like to know what I'm doing before I start - i.e get the theory right
douglasf13 wrote:
I think your missing piece is that the circle of confusion for aps-c is different than 135.
That's exactly what he missed. For identical CoC's denoir's calculations were correct, but acceptable CoC also changes with format, thus theSuede's calculations.
Kind of a moot point, since there are no APS-C 23mm f1.3 lenses. Closest is going to be the Leica 24mm Summilux-M and if you're going to spend like that, an M9 is probably in the budget anyways.
If you look at a picture - the finished picture - you think that "sharp" is when something has good contrast at pixel level (at the scale you're looking at the picture with).
-"What do you need from the lens to arrive at this point?"
Say that your output needs to be 10MP. That's roughly 3900 pixel at the long side (assuming a 3:2 format).
So, your lens(es) needs to be equally sharp at:
FF long side = 36mm >>> 36mm/3900lines = 0.0092mm >>> 1/0.0092 = 108 lines per mm = 54lp/mm
APS-C side = 22mm >>> 22mm/3900lines = 0.0056mm >>> 1/0.0056 = 177 lines per mm = 88lp/mm
-if you want the results from the cameras to be equal.
Jerry_R wrote:
That would make no sense to me... When I also wanted to answer myself question, to what extent NEX can be a backup for M9, I was MOTSLY interested in wide open image quality.
If I would need sharp pictures with large DOF - then u43 acts much better than NEX - taking into account its existing lenses.
i mean to test what the dof conversion factor is. if you just want to test whether the NEX will serve your purpose there is no reason to compare at all – just take the NEX and see if it can do what you want it to.
denoir wrote:
You're not seeing that you can't read half the text in one shot and not in the other?
i certainly see that, when i count the number of readable lines there is a big difference, but when i count the number of lines that actually look to be in focus rather than the number that are readable i get the same number from both shots.
This scenario - that the APS lens is actually that much sharper - is very unlikely to happen at large aperture values, especially since you need to use even larger apertures at the APS sensor.
So you (almost) always will get better contrast and sharper result when you shoot "short DoF pictures" with FF/FX format cameras - compared to APS or smaller formats.
