ChrisDM wrote:
Well, it's not so much sensor size as it is pixel size/density, and the 1Ds3 has relatively small/dense photosites. Granted the 50D will probably be even moreso, but the point is that diffraction is more a function of the lens than the pixels/sensor, and is only a consideration when producing very large prints anyways.
I understand that diffraction is caused by the lens - more specificity, the aperture blades. However as I understand it, the larger the image circle the lens projects, the less diffraction is an issue. So pixel density aside, it would seem to me that the 50d or any cropped body would be more prone to diffraction than a larger sensor. I think the concern is that the high pixel density of the 50d may make diffraction very visible and has 15mp, 1.6x reached the point of diminishing returns on quality due to lens issues. You example, though great, does not really relate to the 50d as it was done with a much larger sensor( image circle).
Paul Tessier wrote:
Isn’t it true that the aperture that diffraction begins to affect image quality has to do with the sensor/film size. So comparing the 50d crop to a 1ds 35mm sensor is not a fair comparison, bigger is more forgiving. That is the reason large format shooters routinely use apertures like f32 and f64 without diffraction issues.
Paul, Chris was not comparing the 50D and the 1Ds3 in his example. His point was that diffraction is not as much as killer as some people have been led to believe. In his example, both shots were done with the 1Ds3 - the one at f/11 showing a blurrier image than the one shot , with the same focus point, at f/18. He's saying that the effect of diffraction at f/18 is less than the loss of DoF incurred by "reducing diffraction" by shooting at f/11. The effect of diffraction on those flowers (or whatever) seen at f/18 is not bad, but f/18 is where all the "experts" say diffraction is ruinacious. The "sharpness" of the objects in the plane of focus at f/18 was probably not quite as good at f/18 as at f/11 (or even at f/8) but Chris didn't care about that - he wanted the image in focus and was prepared to accept a little softening in exchange. The extent to which the effect of diffraction is visible does indeed depend on how much the image is magnified "geometrically" - i.e. mm of sensor (not number of pixels) to mm of print. The absolute f values where diffraction loss and DoF benefit balance depends on the subject distance range and sensor format. Thus in large format situations the geometric enlargement often isn't as great so smaller apertures can be used for the same balance. As in most photography, it's a series of compromises.
AJSJones wrote:
Paul, Chris was not comparing the 50D and the 1Ds3 in his example. His point was that diffraction is not as much as killer as some people have been led to believe. In his example, both shots were done with the 1Ds3 - the one at f/11 showing a blurrier image than the one shot , with the same focus point, at f/18. He's saying that the effect of diffraction at f/18 is less than the loss of DoF incurred by "reducing diffraction" by shooting at f/11. The effect of diffraction on those flowers (or whatever) seen at f/18 is not bad, but f/18 is where all the "experts" say diffraction is ruinacious. The "sharpness" of the objects in the plane of focus at f/18 was probably not quite as good at f/18 as at f/11 (or even at f/8) but Chris didn't care about that - he wanted the image in focus and was prepared to accept a little softening in exchange. The extent to which the effect of diffraction is visible does indeed depend on how much the image is magnified "geometrically" - i.e. mm of sensor (not number of pixels) to mm of print. The absolute f values where diffraction loss and DoF benefit balance depends on the subject distance range and sensor format. Thus in large format situations the geometric enlargement often isn't as great so smaller apertures can be used for the same balance. As in most photography, it's a series of compromises. ...Show more →
Right. And the geometric enlargement of a 1.6 crop is far more than a 35mm sensor is my point. As we reach new heights in pixel density the geometric enlargement is going to come in to play and not just the pixel numbers.
His example is great and shows a lot relevant to 1ds but, not 1.6 crop. The 50d will have different DOF/ diffraction behavior at similar apartures and need more enlargement due to the smaller image size.
The question here is how will the 50d perform? and I just don't think a 1ds image proves much about 1.6x preformance is all.
Geert Koning wrote:
There is more to it than only SNR
Diffraction will kick in at about f 8 with this pixel density, that means stopping down to f 16 for depth of field is useless. Also this sensor will outresolve a lot of lenses. I am not happy with this pixel race at all, I hope the 5DII if it ever comes will not have more than 16mp.
under f/8 and with top lenses it will be better universally, above it will be better to get more DOF (een if losing some overall sharpness to difraction- at the per pixel level)
plus you can always resize it down to 8-10MP and have the same as before.
and they can use a weaker AA filter (don't know if they did)
it doesn't hurt to better than the lens can give, although returns are diminishing
drgreenberg wrote:
Your logic is somewhat flawed. For a given per-sensel fill factor (percentage of each sensel's area that is optically active), increasing sensel count has never impacted noise at the image level. Only sensel-level noise increases and the image-level noise becomes more fine grained.
I thought the point was that if you have gaps, then more pixels mean a higher proportion of gap in the frame, since gap is a fixed cost per pixel. One huge pixel would have very little gap/border, whereas a gazillion pixels could mean most of the area is taken up by gaps. So the 'fill spector' or whatever, gets worse if there are gaps and you increase pixel density. Where was my logic flawed, exactly?
> Unfortunately, keeping fill factor constant isn't easy...
OK - I see now - you're agreeing with me while pretending not to, and substituting the correct jargon.
> The gapless microlens design of the 50D sensor doesn't imply any benefit toward future scaling per se. It's more of a one-time improvement.
Since there are no gaps, there is no longer any noise penalty if you decide to increase pixel density. There used to be a penalty for pushing up the number of pixels in a given area. Now there isn't. Zero gap remains zero gap, and sensor size is final. That was my point, and I still don't see how the logic is flawed. The result is that it no longer makes any sense to say that increasing the number of pixels will lead to more noise. That's not inevitable any more because we're gathering all the light that a sensor of that size could ever possibly gather. Now it's just the job of the semiconductor people to find a way of improving the efficiency of the electronics.
SoundHound wrote:
Mr. Brainiac:
What do you make of the, apparent, resolution limitation of present day DSLR lenses!!
Not sure why you're asking me, but my experience is that as your pixel density increases you only get extra detail if the lenses are up to it. Better technique, careful choice of lenses, and a slightly narrower range of apertures are required if you want to see a 21 megapixel camera surpass a 13 megapixel one in detail. I am sure the 50D will do wonders with really sharp lenses, but some users may not routinely get the full benefit of all those pixels. Having said that, at f8 quite a few lenses ought to do the business. 4.7 microns is new territory though. We won't know which lenses can do it until we try. We're also bound to hear many experts complaining about the strong AA filter and how they have to sharpen more than usual, just like we do with the 1Ds3.
danmitchell wrote:
The "amount" of diffraction is identical on every 1.6x cropped sensor. The absolute "size" of the diffraction is a function of aperture and focal length, and the also-important "relative size" is a function of sensor dimensions.
There is no more diffraction on a sensor with more photosites.
If you make a print from a camera with X photosites using a particular lens/aperture combination, and then make another print from a second camera with the same sensor dimensions, aperture, and lens but with, say, X times 2 photosites...
... diffraction in the print will be the same.
Dan
Dan, what has to be considered is to make a print you have to enlarge the image from a 1.6x sensor a lot more than a FF sensor, so any deleterious effects will be magnified more. Also the 50D will resolve the effects of diffraction better (assuming it's AA filter is not very strong). A large print form the 50D say made @ f/16 should show the effects of diffraction more clearly than from a 20D say and from a 5D. You cannot isolate talk of diffraction from CoC and print sizes. This is why you can get away with f/45 aperutres on medium format since the enlargement required for a large print is much less than for 35mm.
Still the effects are quite often overblown. For example say you shoot at f/11 and f/22 on your 5D and compare the 100% view - you'll definitely see the f/22 shot looks softer and many will then use that as all the proof they need not to stop down. But you'll find that nearly all the detail is still there and judicious use of sharpening coupled with low ISO so as not to exaggerate noise will reveal the f/22 shot can look almost as good from a detail pov, and the shot itself overall may be much better since the DoF will be much larger. I know which one I'd prefer to print large if DoF was most important. This is not to say always shoot at the smallest aperture, but not to be scared to shoot at the aperture you need for the job.
Edited by Pixel Perfect on Aug 27, 2008 at 09:43 AM GMT
Pixel Perfect wrote:
till the effects are quite often overblown. For example say you shoot at f/11 and f/22 on your 5D and compare the 100% view - you'll definitely see the f/22 shot looks softer and many will then use that as all the proof they need not to stop down. But you'll find that nearly all the detail is still there and judicious use of sharpening coupled with low ISO so as not to exaggerate noise will reveal the f/22 shot can look almost as good from a detail pov, and the shot itself overall may be much better since the DoF will be much larger. I know which one I'd prefer to print large if DoF was most important. This is not to say always shoot at the smallest aperture, but not to be scared to shoot at the aperture you need for the job.
Edited by Pixel Perfect on Aug 27, 2008 at 09:43 AM GMT...Show more →
This was exactly ChrisDM's purpose in showing the f/11 vs f/18 comparison above...
Can anyone summarize the various aspects of the discussion into something that makes sense to the rest of us? I will phrase the question again:
Does the increase in megapixels on the 1.6 sensor in any way compromise IQ? Or are we going to have to wait for real world samples and experience to answer this question?
The reason I asked this question initially was that in discussions about P&S cameras, I have often heard it said that increasing the MPs was reaching the point where IQ was being compromised. I wondered if the case was the same with the crop DSLRs
AJSJones wrote:
This was exactly ChrisDM's purpose in showing the f/11 vs f/18 comparison above...
Chris's photo however, is not full of fine detail that is the type to suffer most from diffraction, but it is a nice illustration of how the extra DoF more than outweighs any slight loss of detail. However, I'm saying even when we have a scene with far more fine detail it's surprising how much detail can be extracted at small f-stop if you are careful.
Pixel Perfect wrote:
Chris's photo however, is not full of fine detail that is the type to suffer most from diffraction, but it is a nice illustration of how the extra DoF more than outweighs any slight loss of detail. However, I'm saying even when we have a scene with far more fine detail it's surprising how much detail can be extracted at small f-stop if you are careful.
Agreed - the effect of diffraction softening, a bit like AA softening, will be more or less noticeable depending on subject matter and the eye/brain's expectation of the presence of fine detail (distant tree leaf detail gone is much more noticeable than soft skin in a portrait getting even softer).
Similarly both effects are global in the image and can , as you say, be recovered to a good degree by appropriate sharpening (such as Canon's recommendation of 300%, 0.3 pixel radius for "undoing" the effect of their 1 pixel horiz/1 pixel vert AA filter) The expansion of a point image by the AA filter and its expansion by the growth of the Airy disk could probably be considered similar when considering the choice of sharpening algorithms. At some point in the MP race, the diffraction softening will obviate the need for the AA filter and the whole system will be limited by the lens
John Power wrote:
Can anyone summarize the various aspects of the discussion into something that makes sense to the rest of us? I will phrase the question again:
Does the increase in megapixels on the 1.6 sensor in any way compromise IQ? Or are we going to have to wait for real world samples and experience to answer this question?
The reason I asked this question initially was that in discussions about P&S cameras, I have often heard it said that increasing the MPs was reaching the point where IQ was being compromised. I wondered if the case was the same with the crop DSLRs...Show more →
John.... as I understand it the gapless diodes essentially give the sensor a little more elbow room to accommodate the increase in pixels. I've tried to follow the technical discussion that is going on but feel I need another four year degree to follow it seamlessly.
While I believe the APS-C sensor does have a limit on how many pixels it can handle, I think the proof will be in the pudding whether or not the 50D is pushing those limits. I'm a little concerned that we may be seeing the start of another "how many megapixels can you fit on an APS-C sensor" battle much like we saw years ago with P&S. The result did jeopardize image quality.
rsg_1 wrote:
I'm talking about ADC characteristics. 16-bit monolithic ADC should provide theoretically 96 dB DR and 14-bit 84 dB DR. The SFDR improves considerably. The additional 2-bits should provide at least another stop in sensitivity.
rsq_1 take a dose of this: http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html#SNR and report back in the morning. It is hard to get a 4x improvement in ADCs at this range, a 16 bit ADC will not be better than a 14 bit unless the noise is reduced to account for the higher resolution or smaller measurement bins of the ADC.
Mike V wrote:
Also the actual micro lenses could always be improved even if they are gapless.
Better light transmission, less vignetting, less distortion etc. etc.
I think the only parameter you need to consider for the microlenses is efficiency : (Photons reaching photodiode)/(photons falling on microlens)
Geert Koning wrote:
Diffraction will kick in at about f 8 with this pixel density, that means stopping down to f 16 for depth of field is useless. Also this sensor will outresolve a lot of lenses.
Actually, it will begin above f7, where the size of the Airy disc equals four pixels (RGBG set). The Airy disc is the smallest circle to which a perfect lens will project an incoming pinpoint light source. In practice, it will be hard to distinguish this effect with non-test targets until maybe f11 and then only through flip-comparing two equivalent images.
With a 30D (8MP), my experience shows that f16 is still very good but f22 no longer looks good. The theoretical limit for the 30D is f9.6. This would mean that f11 would still be OK with the 50D.
