Will the Epic 645 hava a OLPF?

[Michael Lindsay]

I ask this as many in the stills world like the look of false detail (crunchy) and I know of no medium format digital system with a OLPF...

Additional questions:

with 6micron photo sites and stops above f11 is the fuzziness due to diffraction enough?
Am I wrong and is there a MFDS with a OLPF?
Is it possible to make a OLPF big enough for the 645...
Has pulling the OLPF away from the sensor (red 1) allowed it to be thinner?

Personally I hate false detail and hope it does have a OLPF (and slightly assume it will). That said I know this world a bit and am curious how red will deal with the pixel peeping assistants and their 1 to 1 digital loupes?

kind regards

Michael L

[Tim Griffith]

If I recall correctly I think the Mamiya ZD & ZDb back had a detachable OLPF. really pushing the memory now but I think the anti-aliasing could be corrected with software? either way, I think you would get that false detail in both, not sure what they do now days.

Tim.

[Pawel Achtel]

with 6micron photo sites and stops above f11 is the fuzziness due to diffraction enough?


At f/11 the diameter of Airy disk (the smallest point on which a beam of light can be focused) is 17 microns.

[Michael Lindsay]

Hi Pawel


At f/11 the diameter of Airy disk (the smallest point on which a beam of light can be focused) is 17 microns.

Do you think that is enough to not need a OLPF?

Tim


If I recall correctly I think the Mamiya ZD & ZDb back had a detachable OLPF. really pushing the memory now but I think the anti-aliasing could be corrected with software?


Do you mean that the softness from a OLPF can be compensated for with edge enhancement?

Michael

[Pawel Achtel]

Do you think that is enough to not need a OLPF?

It's more than enough. The smallest detail the lens can focus at f/11 would be 3x3 pixels large. :idea:

[Michael Lindsay]

I aked not because of the maths you gave but because I've seen lots of medium format stills which seems to have aliasing but are shot at f11+. Maybe I'm wrong and the crunchiness I've seen in the Raw files is the result of something else.

I will dig out some examples and check..

I am excited by the Epic 645 but am wondering about its viability!

regards

Michael

[Tim Griffith]

Michael,
Sorry, I'm really not sure, I haven't had much experience at all with the ZD's, i just remember there being a detachable OLPF, and the software did some funky stuff to correct for this (I never went down this path), but I assume that the software would enhance the edges or something similar.

Tim.

[Stefan Christou]

Pawel are you sure you are not talking about f11 on a S35 lens? I'm asking because the relative aperture of f11 might be a larger physical aperture size on medium format than on s35 glass.

[Tim Griffith]

it wouldn't matter, f number is only affected by the focal length of the lens and the apertures diameter. so any 50mm lens at f11 will ALWAYS have an aperture diameter of 4.5mm.

Tim.

[Pawel Achtel]

Pawel are you sure you are not talking about f11 on a S35 lens? I'm asking because the relative aperture of f11 might be a larger physical aperture size on medium format than on s35 glass.

The sensor size is irrelevant.

[Daniel Browning]

with 6micron photo sites and stops above f11 is the fuzziness due to diffraction enough?


For most people it would be enough, IMHO.



Is it possible to make a OLPF big enough for the 645...


Yes. It's only money. (Cost rises exponentially with area for the good crystal ones.)


At f/11 the diameter of Airy disk (the smallest point on which a beam of light can be focused) is 17 microns.
[...]
It's more than enough. The smallest detail the lens can focus at f/11 would be 3x3 pixels large.


I think MTF is the best way to measure the effect of diffraction. The higher the frequency, the lower the MTF (i.e., smaller pixels have lower MTF at Nyquist due to diffraction).

To avoid aliasing, an MTF of 0% would be the theoretical ideal, but in practical situations I don't think it's necessary to go that low. I think 10% MTF is sufficient to prevent aliasing. Using the formula for MTF of diffraction , I calculate that MTF drops to 10% for green light (550nm) at f/11 at 3.7 micron pixels. Using red light (750nm) I get 5 micron pixels. Both of these are slightly smaller than the 6 micron pixels in the EPICs.

Again, it's a question of what MTF you consider aliasing taken care of.

2/pi*(acos(f)-f.*sqrt(1-f.^2)) where f is 1/(w*f-number).

[Pawel Achtel]

Thanks Daniel,

I thought the MTF would reach 0% at the size of Airy disk (17 microns) corresponding to f/11. The reason the MTF would reach 0% at this frequency is because the size of Airy disk is the smallest detail the lens can produce. Thus, the frequency limit would be around 17 microns.

Are you sure that at f/11 the MTF would be 10% at frequency of 3.7 microns? Even from practical experience, lenses on most DSLRs (6 micron pixels for argument's sake) start to be diffraction limited passed f/5.6.

[Daniel Browning]

I thought the MTF would reach 0% at the size of Airy disk (17 microns) corresponding to f/11. The reason the MTF would reach 0% at this frequency is because the size of Airy disk is the smallest detail the lens can produce.


Have you considered the fact that the intensity of the Airy disk varies from the center to the edge? Or the effect of alignment on a grid of square pixels for sampling? That to resolve two lines requires a minimum of three samples: one for each line and a third to show a change in intensity (half the Rayleigh criteria spacing)?



Are you sure that at f/11 the MTF would be 10% at frequency of 3.7 microns?


I think I got my math right. Here's an online source for the diffraction formula:

http://www.mellesgriot.com/products/optics/os_2_2.htm



Even from practical experience, lenses on most DSLRs (6 micron pixels for argument's sake) start to be diffraction limited passed f/5.6.

Agreed. Most of us have all of our practical experience with diffraction in similar situations: noticing when diffraction starts to affect the image, and trying to avoid it. But "starting" to be diffraction limited is not enough to eliminate aliasing. For example, diffraction can cause contrast to drop from 80% to 40%, but that's still enough to cause aliasing.

Few have practical experience in the area covered by the thread: finding out how small pixels can be before MTF drops below 10%.

[Pawel Achtel]

Thanks Daniel,

Here is the source I was referring to:

http://www.schneideroptics.com/industrial/singer/MatchingLenses.htm

which, I think, is exactly what we are discussing here.

[Daniel Browning]

Here is the source I was referring to:

http://www.schneideroptics.com/industrial/singer/MatchingLenses.htm


As far as I can tell, they're only talking about the typical situation, which is avoiding diffraction:

What is required is a lens system with a fairly low f/# to even theoretically achieve the sensor limited resolution

In this thread we're talking about the extreme opposite end. First diffraction starts. Then it gets annoying. Then it gets bad. Then it gets really bad. Then it finally drops all the way down 10% MTF. Only then is aliasing finally taken care of.

They are talking about trying to be "sensor limited", which means avoiding the "start" of diffraction, or at least avoiding the "annoying" phase of diffraction, but certainly not going anywhere near 10% MTF. HTH.

[Pawel Achtel]

Ahhh, OK, it makes sense now.

[Stefan Christou]

it wouldn't matter, f number is only affected by the focal length of the lens and the apertures diameter. so any 50mm lens at f11 will ALWAYS have an aperture diameter of 4.5mm.

Tim.

I stand corrected. Very educating posts BTW :thumbsup:

[Tim Griffith]

What is required is a lens system with a fairly low f/# to even theoretically achieve the sensor limited resolution[/I]

In this thread we're talking about the extreme opposite end. First diffraction starts. Then it gets annoying. Then it gets bad. Then it gets really bad. Then it finally drops all the way down 10% MTF. Only then is aliasing finally taken care of.

They are talking about trying to be "sensor limited", which means avoiding the "start" of diffraction, or at least avoiding the "annoying" phase of diffraction, but certainly not going anywhere near 10% MTF. HTH.

Great information, thats going to give me something to chew on for a while, thanks. :thumbup:

Tim.

[Dan Hudgins]

The circle of confusion is not just a fuzzy blob with a bell curve, depending on the lens formula it can have some structure, and different structure at different wave lengths, more so off center because of comma.

Also aliasing is not the only issue, you also need to reduce chroma moire.

A good OLPF will cut divide the light over four pixels and help kill those issues, more than just stopping the lens down because the sharpening used in the de-mosaic can be over 400%, which counteracts some of the diffraction making the center of the bell curve brighter than it would be at 100% sharpen (i.e. no sharpen). Also the sharpen needed brings out other shapes in the circle of confusion, which is not a circle off center but looks at bit like a bird on some lenses with the comma as its tale.

I've been stopping the lens down in our scanner to reduce the grain in film scans, but when you sharpen most of the smaller details come back, so it is not an on or off solution. Also you get smaller details in the UV, so if you try it you might want to use a UV+Violet cut filter.

[Rodney James]

The sensor size is irrelevant.

Wrong.

[Pawel Achtel]

Wrong.

Show how. :confused5:

The size of Airy disk (diffraction limit) is the same for the given aperture. Sensor size doesn't matter. How can it?

[Shane Betts]

Hey, accuteoptical, I tried looking at your site and it appears to have been hacked into, with malware warnings popping up all over. You might want to clean up your HTML and get rid of any new additions. Just thought you'd like to know. Been there, done that.

[Jeff Kilgroe]

Our website works well.If you can't see anything,pls refresh page.

That's not what Shane is saying. I'm getting the warnings too. Your site appears to have been hacked or compromised by an intruding web bot, virus or malware. There seems to be malware or a virus embedded that could be infecting anyone who visits your site. Seriously, this is something you or your web people need to look into. Just trying to help.

Also we have a real name policy here now -- see the real names thread in the Recon section.

[Andrew McCarrick]

it wouldn't matter, f number is only affected by the focal length of the lens and the apertures diameter. so any 50mm lens at f11 will ALWAYS have an aperture diameter of 4.5mm.

Tim.

Sorry... but I'm a little confused by this. Wouldn't a medium format lense be physically larger and therefore have larger diameter then a 35mm lens?

Take a 65/70mm (IMAX) lens for example, is that not bigger in diameter than a 35mm lens?

[Jeff Kilgroe]

Sorry... but I'm a little confused by this. Wouldn't a medium format lense be physically larger and therefore have larger diameter then a 35mm lens?

Take a 65/70mm (IMAX) lens for example, is that not bigger in diameter than a 35mm lens?

In some situations, yes, the lenses are larger. However this is not always the case. Once again, focal length is focal length, regardless of the lens coverage format. F number is a representation of the aperture diameter in relationship to focal length. So a 100mm lens at f2.8 always has a 35.7mm diameter aperture, regardless of whether the lens is designed to cover 16mm, 35mm, or 65mm, etc..

[Andrew McCarrick]

In some situations, yes, the lenses are larger. However this is not always the case. Once again, focal length is focal length, regardless of the lens coverage format. F number is a representation of the aperture diameter in relationship to focal length. So a 100mm lens at f2.8 always has a 35.7mm diameter aperture, regardless of whether the lens is designed to cover 16mm, 35mm, or 65mm, etc..

ahh... okay. That makes a little more sense to me.

[Christopher Polis]

In some situations, yes, the lenses are larger. However this is not always the case. Once again, focal length is focal length, regardless of the lens coverage format. F number is a representation of the aperture diameter in relationship to focal length. So a 100mm lens at f2.8 always has a 35.7mm diameter aperture, regardless of whether the lens is designed to cover 16mm, 35mm, or 65mm, etc..

The difference being that with 65mm you get a much larger field of view than you do with 35mm. So to end up with the same image you have to use a longer lens (not sure of the actual number here, but 65/35 x 100 will end up somewhere around 185mm...)

And a 185 / F2.8 has a significantly larger aperture. This is where 'sensor size matter' - if you keep f ratio constant and field of view constant. If neither of those premises hold, then sensor size doesn't matter.

[Jeff Kilgroe]

Right you are Christopher, I didn't go into that much detail... It's a topic that has been beat to death on these forums. As film or sensor sizes increase, so does the field of view for a given focal length. A 90mm lens on 35mm is considered somewhat long. On the 645, it would be considered rather wide.

[Jeff Vogeding]

I ask this as many in the stills world like the look of false detail (crunchy) and I know of no medium format digital system with a OLPF...




First post from a "Stills" Guy. Their is a reason that Phase and Hasselblad Medium format backs do not have an OLPF. Image results are superior without them (still wise). The higher the pixel the count the less likely the chance of Moiré and other ugliness. Have you used such a camera? Have you actually seen files? Are you joking with the "false detail" bit? Canon files are Mushy with too much OLPF and never look as good to me compared to medium format. Red ONE files? Sooooo mushy that they border on unusable for many still applications. They make the Canons look sharp. An epic without a removable OLPF is a non starter for many people I know unless its very weak. -Jeff

[Michael Lindsay]

Hi Jeff

I know your opinion is common it is why I asked the question.

Have a little internet search on 'false detail' as it might help you understand a little.

Yes canon files (and red even more so) are mushy at a 1:1 inspection. But pixel detail at this level can't ever be trusted.

Thanks for posting as I suspect many on this list (and Red) might not realise what many stills guys seem to think about OLPFs.

regards

Michael L

PS I'm guessing your questions of me are rhetorical and for effect. But if your actual interested in the answers: Yes (love mf for stills despite false detail), yes and obviously no.


First post from a "Stills" Guy. Their is a reason that Phase and Hasselblad Medium format backs do not have an OLPF. Image results are superior without them (still wise). The higher the pixel the count the less likely the chance of Moiré and other ugliness. Have you used such a camera? Have you actually seen files? Are you joking with the "false detail" bit? Canon files are Mushy with too much OLPF and never look as good to me compared to medium format. Red ONE files? Sooooo mushy that they border on unusable for many still applications. They make the Canons look sharp. An epic without a removable OLPF is a non starter for many people I know unless its very weak. -Jeff

[Daniel Browning]

Their is a reason that Phase and Hasselblad Medium format backs do not have an OLPF.


Yes. It's the same reason why they have have high noise, low dynamic range, poor sensitivity, slow frame rate, and a host of other deficiencies compared to 35mm DSLRs: volume. They have some great advantages too, but unfortunately not enough to stem of death of Medium Format manufacturers (from over a dozen down to just 4 now). They can't afford to do their own sensor design, so they have to buy off-the-shelf parts from Kodak and Dalsa, which have far lower performance.

A good OLPF is lab-grown, high-grade, ground, and polished Lithium Niobate crystal. The cost scales exponentially with area because even the tiniest defect will show up on the image, thanks to being so close to the sensor. Even though MFDB is only three times more area than FF35, the cost can be an order of magnitude (or more) higher. More importantly, Canon/Nikon ship millions of units a year, compared to less than 6,000/year for all MFDB combined, so economies of scale is a huge factor.

There is also momentum. Some people have been practicing unsafe imaging (with no protective anti-alias filter) for so long that they have actually become addicted to the aliasing artifacts (jaggies, stair-stepping, sparkling, "snap to grid", wavy lines, bands, patterns, fringing, popping, strobing, noise, false detail, and of course moiré). You can't rip away their artifacts from them too suddenly, or they will experience extreme withdrawals. Not a good move for a market that is already shrinking.


Image results are superior without them (still wise).


How could you possibly explain the fact that over 99% of digital cameras are built with anti-alias filters? Why would manufacturers go to all the extra fixed and variable costs of anti-alias filters, even for 36x24mm sensors, and live with the reduced contrast caused by them, if there was no benefit?



The higher the pixel the count the less likely the chance of Moiré and other ugliness.


That's not true at all. If you have soft lenses, or let diffraction soften the image, or have motion blur, then you can avoid moiré and dozens of other aliasing artifacts. But if you make a sharp image, then you will get aliasing.



Have you used such a camera?


I have not.



Have you actually seen files?


I've seen lots of raw files from MFDB, although they are a little hard to come by.



Are you joking with the "false detail" bit?


It's not a joke. Aliasing is false detail. MFDB have severe aliasing. Therefore MFDB have lots of false detail. Jagged eyeglass rims when they were smooth in real life. Colored sparkles in the hair where there was none. Blobs of the wrong texture in clothing.

I've tried C1P to see how it does at dealing with the aliasing in software, and I was not impressed. Making speckles into blobs may be an improvement, but it's far worse than if the speckles weren't there in the first place.



Canon files are Mushy with too much OLPF and never look as good to me compared to medium format.


That makes sense, since you like aliasing aliasing artifacts.



Red ONE files? Sooooo mushy that they border on unusable for many still applications. They make the Canons look sharp.


Instead of getting the aliasing artifacts direct from the camera, why don't you just add them in post? They have plenty of software now for introducing aliasing artifacts. It's like methadone, you can slowly reduce the software until there is no more aliasing left.

[Jeff Vogeding]

That makes sense, since you like aliasing aliasing artifacts.



Daniel and False detail haters ;-)

Seems some of you are a bit eager to dismiss medium format digital and their "false detail". The way you describe MFD images they are unusable stair stepped monsters. Nothing could be farther from the truth. I have looked at tens of thousands of images from 8x10 tango drum scans to just about every digital camera ,so my opinion comes from years and years of real world experience and not theory, and not specs. I'm talking about actual prints in my hands, and big prints at that. I promise if I show you near identical shots from a canon/nikon and a Phase you would pick the Phase. The Fact is MFD produces a superior image at this time compared to 35mm digital when printed large (that is what counts to most photags,not all). Your talking about stuff you've never used and seen some examples. That seems a dangerous way to draw a conclusion. The false detail you speak of doesn't rear its ugly head near as much as you would think in the real world. You can talk theory, and you can talk specs but reality lies somewhere else. For example, If every lens I had performed to its printed MTF curve I'd be a very happy man. But thats not reality, now is it? Reality is that too much filter on a camera is like a haze in front of your image. You have to over sharpen to get detail back which introducing artifacts, and they are the very thing that seem your enemy. I'll take the detail up front, and blur (rarely) to take away if I please. I guess many many professional photags that shoot MFD are just idiots and know nothing of image quality, yet you who has not used said tech knows better. ;-)

Will the next generation of Canon's and Nikons further depress MFD? Yes. Will they come closer in quality? Yes. However, MFD is a mature technology. A company like Phase has been around since the early 90's. They have honed image quality for many years to take adavntage of the Kodak and now Dalsa CCD's. They use the same Kodak chip's that Hasselblad uses in their backs, yet The Phase can do hour long exposures (the Hassy just a few seconds and much noisier) that are amazing. A testament that it is both the chip and the software that makes a great image. There are caveats to MFD of course as you mention. You need to shoot near base ISO (50,100), you can't shoot many frames per second, etc. For certain jobs (high ISO/frame rate) the canon is the better tool. But this has always been the way. Is the RED the perfect tool for all jobs? Horses for courses as many say. But the latest Phase backs have 12.5 stops of DR (to low for you?) and no noise at or near base ISO. Pixel peep at 100% or beyond as some do and you will see nasty stuff for sure. Does it show up in a 24x36 inch print? Almost never. But I see more "bad" images out of a canon/nikon, and don't get me started on RED still images. I understand a need for an OLPF. I understand the need for one in moving imagery is most likely even greater. However, I want the right balance of detail retention and anti aliasing. My guess would be an OLPF on the Epic would be tough call. Like you said they are pricey, but do have benefits. I guess we will see. I look forward to what red will do with this Brain as it seems the most ambitious to me (save the 6x17). I've heard Jannard is a camera buff, I wonder what he feels on the matter in regaurds to stills.

[Daniel Browning]

Thanks for the well-reasoned response, Jeff. And welcome to the forum.

Two people can look at the exact same image and each see something different. Take an aliased image for example. Where one sees overly harsh and sudden transitions from black to white in just 2 pixels, another sees microcontrast. Where one gets the impression of fakeness, another gets the feeling of sharpness. Where one is jarred by the conformation of small details into a slightly different location than they exist in nature, another is awestruck by the high acuity.

Same thing with anti-aliased images. Where one sees slow, smooth, and careful transitions from black to white, over 3 or more pixels, another sees mushy detail. One gets the impression of natural, life-like renditions, another gets the feeling of haze and low contrast.

So you may percieve a MFDB image as having high microcontrast, sharpness, and acuity; while I see the same one as harsh and unnatural. The OLPF'd images you perceive as mushy, haze, lowcon images are, to me, smooth and natural. So it makes sense that it would be difficult for us to come to agreement on how big of an issue aliasing is.

You can look at the success of the 5D2 video to realize that many people are in your camp. It has far worse aliasing than MFDB, yet is going out on tons of projects.

If camera manufacturers like RED were shrewd enoguh, they would stop wasting so much money and effort to fight aliasing. It's not like they get appreciated for it, they're more often lambasted (IMHO). I think the market of people that like aliasing is large enough. But since I happen to belong to the smaller part that prefers anti-aliased images, I'm glad they haven't (yet) given up on the expensive AA filters, and hope they never do. (At least until we start hitting diffraction cutoff frequencies).



The false detail you speak of doesn't rear its ugly head near as much as you would think in the real world.


I think you and I are thinking of different things when we say false detail. I think it's one of the things I described above: we just literally see the same image in a different way. For example, if we look at a transition from black to white, and on your camera it takes 2 pixels, to me that is false detail. On my camera it takes (at least) 3 pixels, and to you that is mushy. So we could look at the same image and have a different idea about whether it contains false detail or not.



You can talk theory, and you can talk specs but reality lies somewhere else. For example, If every lens I had performed to its printed MTF curve I'd be a very happy man. But thats not reality, now is it?


Agreed.



Reality is that too much filter on a camera is like a haze in front of your image. You have to over sharpen to get detail back which introducing artifacts, and they are the very thing that seem your enemy.


True. The unfortunate reality of OLPF is that it lowers contrast at far more spatial frequencies than the ideal. It is a haze, and sharpening can introduce artifacts. The way I prefer to see this problem dealt with is by increasing resolution. For every increase, the spatial frequencies (e.g. 60 lp/mm) affected by the OLPF move to higher and higher levels, and so affect any given print size less and less.



I'll take the detail up front, and blur (rarely) to take away if I please.


Generally, I agree with that, but not in this one case.



I guess many many professional photags that shoot MFD are just idiots and know nothing of image quality, yet you who has not used said tech knows better. ;-)


You've got me there. Not only have I not used MFD, but to make matters worse, I'm not even a pro. The many pros that shoot MFD like that look, (along with most of the general populace), and that's OK, but I do not. I still think that anti-aliased looks better. (And for now, 99% of camera makers seem to agree.)



But the latest Phase backs have 12.5 stops of DR (to low for you?) and no noise at or near base ISO.


Yeah, my criticism of their DR was a little unfair for several reasons. One is that there are other things that matter more. For example, even the older backs had higher SNR (signal-to-noise ratio) over the far majority of the dynamic range: that translates to a more pleasing image in several ways, including color depth and tonal gradations. That is visible even when you're only using 6 or 7 stops of dynamic range, so for many shooters it's more important than any difference in the "engineering" definition of dynamic range.



My guess would be an OLPF on the Epic would be tough call. Like you said they are pricey, but do have benefits. I guess we will see.


Agreed. RED does things differently, so maybe they'll offer an option. What would be really great in the far-off future is an electronically-adjustable AA filter. One way is to take the anti-dust "sensor shake" system and use it during the capture to blur the image. The amount of movement determines the strength of the filter. Another possibility is "microscan", where you take multiple actual images (moving the sensor each time) to reduce aliasing (and/or increase resolution). I doubt either option will happen soon, but I can just imagine dialing a different AA strength into the camera depending on the shot.

Thanks again for the post.

[Peter Moretti]

First post from a "Stills" Guy. Their is a reason that Phase and Hasselblad Medium format backs do not have an OLPF. Image results are superior without them (still wise). ...

Jeff, I think you touch upon a very significant difference: the implications of still vs motion uses. And it actually has nothing to do with motion. It has to do with viewing size.

If iPhone video were invented fifteen years ago, but big screen tv's and movie theatres had never been conceived of, video cameras would have little use for OLPF's. But for almost all of its history, video has been viewed at much larger size than what stills have been viewed at.

This orders of magnitude of additional enlargement makes the "evils" of the sensor's grid photosite arragement more apparent.

[Michael Lindsay]

Anyone involved in any kind of photography would be well advised not to dismiss MF digital camera on any theoretical grounds. There is a quality to the possible photography that is often compelling (putting aside false detail and pixel count) and unachievable with smaller sensors cameras (I know DB will tell me I'm wrong).

The next time I have a stills shoot I will test like for like (DOF,FOV etc) with a 5d try to be a bit more scientific.

I do worry that Red will find it difficult to continue with their OLPF strength when producing the larger format cameras. I hope they at least offer the option.

regards

Michael L

[Michael Hastings]

Yes. It's the same reason why they have have high noise, low dynamic range, poor sensitivity, slow frame rate, and a host of other deficiencies compared to 35mm DSLRs: volume. They have some great advantages too, but unfortunately not enough to stem of death of Medium Format manufacturers (from over a dozen down to just 4 now).

If MF dies I'll be able to buy all those great lenses cheap for use on the Epic 645! :biggrin:

[Daniel Browning]

It has to do with viewing size.

Good post, Peter; however, I think that motion does also play a very large part. The reason is that aliasing artifacts move, dance, sparkle, and strobe with motion. For example, jaggies on someone's eyeglass rims are one thing, but when there is motion, they jaggies look like a tiny waterfall of motion, distracting the viewer from the eyes. Sparkling in hair, in a still image, could be interpreted as glint from hair spray, or something else, but in a moving image, the sparkles pop and dance as if the hair was filled with hundreds of tiny fireflies.

Probably the most important reason is the effect of aliasing on compression. Aliased images already have lots of high frequency content, which is hard enough to compress on its own, but to make matters worse, they move in the opposite direction to that of the subject, which throws off the compression codecs, causing all sorts of compression artifacts, even in non-aliased parts of the image. That is the reason why the 5D2, even with 48 Mbps MPEG-4, has more compression artifacts than 25 Mbps MPEG-2. When you get around to encoding it to 5-10 Mbps MPEG-2 for TV distribution, the compression artifacts get much worse with aliasing than with an anti-aliased image.


There is a quality to the possible photography that is often compelling (putting aside false detail and pixel count) and unachievable with smaller sensors cameras (I know DB will tell me I'm wrong).


You're right. In addition to the advantage in color depth I mentioned above, they also have some great advantages in lenses. For one, the possibility for higher contrast at any given display size (spatial frequency), and more control over DOF for any given level of contrast/aberration. Their sensor tech is generally a bit behind Nikon/Canon (e.g. D3X is 13.5 stops DR right now), but they can make up for it with size.

Personally, I think 645 is about the perfect sensor size for me. When you forgo the optical viewfinder and shrink the electronics, it can be made as small as the current DSLR. I'm just waiting for it to drop to around $5,000 (in a decade or two).



The next time I have a stills shoot I will test like for like (DOF,FOV etc) with a 5d try to be a bit more scientific.


I would be interested to see the results of that.

[Michael Lindsay]

I would be interested to see the results of that.

I might be able to do this on Wednesday.

5dmrk11 vs Hasselblad digital back (lenses would be Canon 24mm mrk11 and a 50mm Hasselblad lens)..

It will be very real world as it will be a human subject.

Hope I have the time.

Michael

[Daniel Browning]

5dmrk11 vs Hasselblad digital back (lenses would be Canon 24mm mrk11 and a 50mm Hasselblad lens)..


My favorite lens! The 24mm f/1.4 II almost never leaves my 5D2. I hope you'll have the chance to test thin DOF as well as deep DOF.