Insights Digital Camera Buyer's Guide Introduction - Comments

[Andy Resnick]

Andy Resnick submitted a new PF Insights post

Digital Camera Buyer's Guide Introduction

Continue reading the Original PF Insights Post.

 

[jtbell]

What makes a digital camera special? [...] A second key difference is what happens when you press the button to take a picture. In manual cameras, not much happens- a mechanical shutter opens for a set amount of time, exposing the film, and then closes. Digital cameras do a lot of things when that button is depressed, including light metering and focusing,

Prior to the digital era, most film SLRs had automatic light-metering and many had autofocusing. When I bought a Pentax K1000 around 1990, it was one of only a few completely-manual SLRs still available, if I remember correctly.

 

[Andy Resnick]

I had a Nikon FG; no autofocus, but it did have metering (no autoexposure- I had to set the shutter speed and/or aperture manually). In addition, it had a full manual mode- if the battery was dead, for example- that provided a fixed 1/90 s shutter speed.

 

[Andre]

I still have my Minolta XD7 and some lenses. So when I decided to go serious on photography again, selecting a DSLR, I first considered Sony/Minolta because my lenses may still be useable.

But nope. Also lenses seem to age. Nothing automatic in there. I finally went for the Canon EOS 450D back in April 2008 because of superior image quality at that time.

 

[Greg Bernhardt]

Just saw this now. Great work Andy!

 

[Andy Resnick]

Just saw this now. Great work Andy!

Thanks, but I can't take all the credit- call me the 'lead author'. It was a collaborative effort from 3 or 4 of us (who can choose to identify themselves if they wish).

 

[turbo]

Prior to the digital era, most film SLRs had automatic light-metering and many had autofocusing. When I bought a Pentax K1000 around 1990, it was one of only a few completely-manual SLRs still available, if I remember correctly.

I had a large stable of Olympus OM-1s and a single OM-4 with Zuicko prime lenses. I had two OM-1s that were dedicated to astrophotography. They were both dirt cheap at pawn shops because the meters didn't work. I think I paid $20 each for them. Still, they had manual shutters, mirror lock-up, interchangeable focus screens, and other features that made them perfect for astro-imaging. Plus, the OM series was smaller and lighter than about any other pro-grade SLR.

 

[Andre]

Just wanted to tell that the new micro four thirds system (cropfactor 2) is maturing. A big leap ahead is made with the Panasonic GH2, from preliminary results I see that the image quality is excellent, on par or better than most of the APC-C format cameras (crop factor 1.5 / 1.6) like the CANON EOS7D or the Nikon D7000, that is on low ISO's. On higher ISO's -as expected- the image noise is clearly worse albeit that the image remains tack sharp.

Micro four third cameras are hybrids half way between upper end point and shoot cameras and DSLR. The lens is exchangable but there is no mirror.

It's not cheap though.

 

[Borek]

Hyperfocal distance: The hyperfocal distance is calculated by maximizing the depth of field: when a lens is focused at the hyperfocal distance, objects from infinity to half the hyperfocal distance are rendered in focus. The analytic result is:
H = f( \frac{f}{Fc}+1),
where H is the hyperfocal distance, f the focal length, F the f-number, and c the diameter of the circle of confusion. The hyperfocal distance also forms a series solution: focusing the lens at 1/2 the hyperfocal distance renders objects from the hyperfocal distance to 1/3 the hyperfocal distance in focus; focusing at 1/3 the hyperfocal distance covers objects from 1/2 to 1/4 the hyperfocal distance, etc. For example, the hyperfocal distance for a 28mm lens set to f/16 on a 35mm camera is about 1.6m. Everything from 0.8m to infinity will be sharp in a photograph taken with this lens focused at an object 1.6m away.

I understand value of 1.6m for 28/8 can be calculated assuming circle of confusion of 0.03 mm and the 'real' focal length. Thats for a full frame camera. What values would be reasonable for APS-C?

 

[Andre]

I understand value of 1.6m for 28/8 can be calculated assuming circle of confusion of 0.03 mm and the 'real' focal length. Thats for a full frame camera. What values would be reasonable for APS-C?

I guess if you don't change other facters that the crop factor 1.6 for Canon applies, hence 0.02 (0.01875).

Mind that a CoC of 0.03 mm on a 24mm height compares to a resolution of 800 LPH (lines per height). However modern full frames easily get to 2500 LPH. So it seems that you get the oxymoron that the better the camera resolution, the less the depth of field and the farther away the hyperfocal distance.

 

[Borek]

So it seems that you get the oxymoron that the better the camera resolution, the less the depth of field and the farther away the hyperfocal distance.

That's why I asked for "reasonable"...

 

[Andre]

I think that is worth some empirical trial and error [strike]work[/strike] fun.

 

[Andy Resnick]

I understand value of 1.6m for 28/8 can be calculated assuming circle of confusion of 0.03 mm and the 'real' focal length. Thats for a full frame camera. What values would be reasonable for APS-C?

Sorry- I didn't see this when you posted it. Varying the image size means the circle of confusion must be scaled (just as the focal length must be scaled). If the APS-C has a crop factor of say, 1.5x, the CoC is also divided by 1.5x so it would be 0.02 mm.

The CoC specification was originally determined by using a standard enlargement of 8x viewed at 25cm; the reality is that viewing conditions and acuity affect the result.

One counter-intuitive result (for me, anyways) is that larger format images (4" x 5" or even 8" x 10") appear sharper and have a more narrow depth of field than 35mm images.

 

[Topher925]

What is everyone's opinion of the Sony SLT A65 & 77 being released next month? I'm in the market for a new camera and so far have been looking at the Pentax k-5 for its low light performance and star tracker GPS thing. But with the A65 having such impressive specs for such a low price I'm really considering it instead.

My only concern is low light performance. Given that its an SLT with ~2/3 of the light actually hitting the sensor coupled with the sensor having a very small pixel pitch makes me think that low light performance/noise will be rather poor compared to the k-5. Sony also limits the ISO to 16000. Whichever camera I buy a lot of its time will be spent doing some lite astrophotography so high noise is a big concern.

 

[Andy Resnick]

I don't recommend specific products- nobody pays me or provides equipment to use- but if your highest priority is low-light-level imaging, then you are right to carefully examine that spec.

 

[Andre]

What is everyone's opinion of the Sony SLT A65 & 77 being released next month? I'm in the market for a new camera and so far have been looking at the Pentax k-5 for its low light performance and star tracker GPS thing. But with the A65 having such impressive specs for such a low price I'm really considering it instead.

My only concern is low light performance. Given that its an SLT with ~2/3 of the light actually hitting the sensor coupled with the sensor having a very small pixel pitch makes me think that low light performance/noise will be rather poor compared to the k-5. Sony also limits the ISO to 16000. Whichever camera I buy a lot of its time will be spent doing some lite astrophotography so high noise is a big concern.

yes you are right noise is an issue with the new Sony's. Obviously with a fixed translucent mirror you lose some light going to the main sensor. So it's plain physics that signal to noise ratio is worse. However the 25 mp of the A77 reduces that problem somewhat if you reduce the output to 16MP. Another issue would be hot pixels, that emerge at long shutter speeds and high ISO, same problem there.

The best signal to noise ration in relation to its price in DSLR's comes from the Canon 5D mkII (21 mp) especially at high ISO. The best absolute high ISO S/N ratio comes form the Nikon D3S (12mp) but at about double the price.

 

[khemist]

yes you are right noise is an issue with the new Sony's. Obviously with a fixed translucent mirror you lose some light going to the main sensor. So it's plain physics that signal to noise ratio is worse. However the 25 mp of the A77 reduces that problem somewhat if you reduce the output to 16MP. Another issue would be hot pixels, that emerge at long shutter speeds and high ISO, same problem there.

The best signal to noise ration in relation to its price in DSLR's comes from the Canon 5D mkII (21 mp) especially at high ISO. The best absolute high ISO S/N ratio comes form the Nikon D3S (12mp) but at about double the price.

The Nikon D700 has much better noise reduction for the same price point as a mark 2. However, they are both outstanding cameras.

Interestingly enough, the higher MP the mark 2 has contributes to the increased noise. My friend will shoot at ISO 6000 with the mark 2 and still get relatively noise free shots, but in comparison tests I have seen the d700 has less noise at almost all ISO levels.

 

[Andre]

Yes the D700 is very good at high ISO and outperforms the 5D MKII on some other things as well, except image detail/resolution due to only 12.1 MP versus 21.1MP. If you reduce the image size to the same 12.1 MP, you also reduce noise, while maintaining detail.

I also should add that Andy's Sony A900 has the best resolution/detail but cannot keep up at higher ISO


To Topher925, you can find unlimited discussions here about the Sony A65/77
http://forums.dpreview.com/forums/forum.asp?forum=1037

 

[Andre]

What is everyone's opinion of the Sony SLT A65 & 77 being released next month? ... Whichever camera I buy a lot of its time will be spent doing some lite astrophotography so high noise is a big concern.

Here are more answers, dpreview has made test shots available with the A77. So if you go to any test, like this of the Canon 60D then you can compare crops of many cameras on this page (in RAW).

In the individual drop down menus you can select available cameras (I selected Sony A77, Nikon D5100, and Canon 5D mkII) you can also select ISO and the crop to compare.

So I went to ISO 3200 and selected a grey area for better noise comparison:

I would consider the Sony A77 a bit underwhelming. The Canon 5D mk II clearly leads the pack, while the Nikon D5100 is slightly better than the three year older sensor of the Canon EOS 60D.

Maybe the Nikon D5100 should be in the very short list of least noise for the best price.

 

[Topher925]

I spent at least a good hour the other day checking out the A77 photos on dpreview and comparing to other camera's I'm looking at. While the JPEGs are questionable, the RAW files definitely bring out the shortcomings of the Sony 24MP sensor. I think I can cross that one off the list given the type of photos I want to take.

I've been looking closely at the d5100 for a long time now, it truly is a fantastic camera for money. The fit and finish could be a bit nicer but for the price its acceptable. If I end up not getting the k-5 I'll most likely pick up the d5100.

 

[Andre]

As I said earlier, buying an SLR is like marrying to that particular brand. Sooner or later more lenses are added in the inventory and then it's getting real tough to switch brands. Therefore looking at possible future desired lenses could assist, taking decisions.

 

[Andre]

Anyway, this is one of my favorite crops:

Notice that the K-5 is excellent for noise but moderate on resolution detail. Notice also that the Canon 600D (Rebel T3i), sharing the sensor with the 60D and 7D, is just outperforming the D5100 with half a notch on details, despite the slightly worse performance on noise.

 

[khemist]

Just another tip. The MP doesn't really matter. Any camera nowadays will be able to perform fine. Unless you blow up your photos to poster size, no one will be able to tell the difference.

To be honest, get nice glass over a nice body. A sharp lens can make an aging camera come back to life.

And in the end, whatever camera you buy will work. It will have the ability to take good pictures, but its really who is behind the camera that really matters.

 

[Andre]

Just another tip. The MP doesn't really matter. Any camera nowadays will be able to perform fine. Unless you blow up your photos to poster size, no one will be able to tell the difference.

The disadvantage of high MP is noise, but the advantage is that you can crop more, getting more effective focal length from your glass.

To be honest, get nice glass over a nice body. A sharp lens can make an aging camera come back to life.

Nicer bodies will still make that picture, where others can't, especially in high dynamic situations, you'll soon learn the difference.
Edit: Like these for instance. and this...

And better glass is not necesarily more expensive glass. Moreover, if you want from excellent glass to splendid glass, you pay a nice body more.

but its really who is behind the camera that really matters.

Absolutely

 

[turbo]

To be honest, get nice glass over a nice body. A sharp lens can make an aging camera come back to life.

And in the end, whatever camera you buy will work. It will have the ability to take good pictures, but its really who is behind the camera that really matters.

All true, and as Andre said above, one should select a system, not just a camera. Once you start adding high-end lenses to your kit, and maybe another camera body, you'll have an "investment" that is hard to recoup if you decide to switch brands.

I decided that I liked the Canon brand, and liked their line-up of lenses so I started out there. Now, I have two 30D bodies and a selection of lenses, including a 100-400 L. I have not felt a driving need (or even want) to upgrade. The quality of this equipment is incredible, and the biggest limitation in the system is the "nut behind the wheel" (myself). Yes, there are shutterbugs who will try to stay on the bleeding edge, but that is expensive, even if you stay within your brand and chase body upgrades so you can keep using that expensive glass. Andre's shots have tempted me to consider a 100mm macro, but I shoot macro so infrequently that I have resisted the urge to splurge.

 

[Andre]

To be honest, get nice glass over a nice body. A sharp lens can make an aging camera come back to life.

Why not put that to a test.

Suppose you own a Canon 350D with the too-cheap-for-words standard 18-55mm IS lens and you have something a bit short of a grand to spend. What would give you a better resolution (sharper pictures)?

Get the legendary super 17-55mm F2.8 lens for your 350D or get a new Canon 600D body (less expensive) with the same cheapo standard lens?

You can find the answer here.

 

[turbo]

The too-cheap-for words 18-55 (no IS) lens on my 30D is remarkably sharp. I was initially turned off by the light-weight plastic body, etc, in comparison to my old Olympus lenses (OM system), but I was quite pleasantly surprised by the performance of the actual glass.

 

[Andre]

so let's continue the test. this what you'd have, if you scroll roughly halfway you see that the 350D (8mp) with the 18-55 averages about 1850 lines per height.

Now you buy the super 17-55 to get this. See that the center is clearly better getting to 2100 lines per height. But the average is not increasing very clearly.

Now we know from DPreview that the image quality of the 600D with 18MP is practically identical to the 60D and 7D, since it's the same sensor. The 50D here is slightly inferior to that, so we know that the 600D is certainly not going to give worse results here. But as shown here, the cheapo 18-55mm on the 50D gets us average results around 2200 lines per height.

Consequently better glass is not dogmatic better than a better body. On the contrary, if you start comparing all the other nice improvements between the 350D and the 600D.

 

[Andy Resnick]

I think this discussion is a little too simple-minded. For example, my luminar lenses are 50 years old and I'm willing to bet they will outperform any contemporary macro lens, period- not just the optical quality but also the fact that they can be used on a 4" x 5" view camera (and possibly an 8 x 10). In terms of high-ISO/low light imaging, I'll put my EMCCD camera up against any DSLR. And it's not true that you are locked into a particular manufacturer, either.

From my perspective, I have never been disappointed when I get the best camera/lens that I can afford: not the most expensive, but the equipment that best meets my needs, regardless of who made it. Of course, getting high-end equipment also (initially) showcases my inability to take advantage of the full performance- all flaws are glaringly obvious.

 

[turbo]

You may be right about some of that. Over the years, I have owned a stable of Bronica Zenzanon primes, and Olympus Zuicko primes. I shopped carefully, and ended up with some superior lenses. Now, I can stroll around with a couple of Canon 30Ds and a few zoom lenses, to cover ranges from 18mm to 400mm. That's a whole lot handier than packing around two Bronicas and 4 primes, or 4-5 Olympus bodies and a corresponding number of primes. Another little wrinkle is that when you head out with bodies loaded with color negative film, color transparency film, Tech-Pan, etc, you can always end up wishing you had loaded films with different capabilities, different speeds, grain-size, etc. I can get a lot of flexibility on the fly with the DSLRs, and post-processing. I don't long for the days of film.

 

[Andre]

I For example, my luminar lenses are 50 years old and I'm willing to bet they will outperform any contemporary macro lens, period- not just the optical quality but also the fact that they can be used on a 4" x 5" view camera (and possibly an 8 x 10).

That's quite a statement. Would that imply that 50 years of innovations like aspherical, low dispersion elements and diffractive optics not could prevent that lens making deteriorated?

 

[Andy Resnick]

That's quite a statement. Would that imply that 50 years of innovations like aspherical, low dispersion elements and diffractive optics not could prevent that lens making deteriorated?

I stand by it. For whatever reason, Zeiss stopped making those lenses in the 1970's- none of those technologies (except possibly aspherical surfaces) which I agree could improve the optical performance of those lenses were ever incorporated. The fact that each lens has only three elements makes the optical performance even more astounding.

I'd be happy to put my claim to the test- all we need is a 'standard object' that we can all photograph using whatever tools we wish. My only constraint is that I need to work indoors. Normally I would suggest a coin or piece of newspaper, but given the international participation in this forum, it's not obvious what the best choice would be.

 

[Andre]

Sure you can compare with what we can and I would love to see results of that lens but it's not a game/match. I know that my 7D/100mm macro is not the best in the world, Borek beats it already with the 100mm L-version, but with a 5D mk2 body, results would even be better which would be topped by the A900, since the resolution is a function of both lens resolution and sensor resolution.

But we can compare pix maybe of a yardstick, or matches or playing cards, or a certain common brand of batteries.

 

[Borek]

Andre and I can take a picture of Canon's lens cap

 

[turbo]

Andre and I can take a picture of Canon's lens cap

From the inside? Your 100mm macros should deliver equivalent quality, I would think.

 

[Andre]

yes here is mine

[PLAIN]http://dl.dropbox.com/u/22026080/lenscap.jpg

Anyway, I would also be interested to see results compared to the Canon MP-E 65mm

 

[Andy Resnick]

Here's some preliminary results using two resolution targets. The first is from Edmund optics and is a chrome on glass variable bar target:

[PLAIN]http://img191.imageshack.us/img191/9345/targeto.jpg

and the second for higher-magnifications is a Richardson Gen III test slide:

[PLAIN]http://img97.imageshack.us/img97/9681/targetda.jpg

The images below are taken from that small region in the dead center.

There's a few things to keep in mind- first, although the targets are specified in terms of a length (for example, 5 line pairs per mm), the proper way to compare lenses is by the angular resolution: line pairs per radian. I didn't measure the object distance so I can't convert the units properly.

Second, the pixelated nature of the CCD results in imaging artifacts, especially aliasing:

[PLAIN]http://img638.imageshack.us/img638/7117/aliasing.jpg

In this case, it shows the effect of angular misalignment between the bars and the pixels. A related issue is the Bayer filter, but from the point spread images I took earlier, that issue may not be a significant one.

Lastly, for whatever reason I set the camera to 6 MP instead of the full resolution. This is probably a fatal flaw. In any case... From top to bottom are images from the 100mm, 63mm, and 25mm at full aperture, at the two useful extremes of reproduction ratio. For the 100mm, this is 1x and 1.78x (45 lp/mm and 75 lp/mm), while the 63mm images are at 2.5x and 4.5x (105 lp/mm and 190 lp/mm) and the 25 mm is 10.5x and 28.5x- the largest bars are 2 microns wide (4 microns per light/dark pair):

[PLAIN]http://img204.imageshack.us/img204/3050/topdi.jpg

What I should probably do is re-take these at full resolution...

 

[Andre]

Impressive Andy and yes, resolution is a function of lens resolution and sensor resolution. But maybe this should go in another thread?

 

[Topher925]

I ended up pulling the trigger on a Nikon d5100 w/kit lens and 55-300 DX VR lens last night. Amazon was having a killer deal where I got the stuff above and a camera bag for a little over $900 shipped. While I was leaning more towards getting the K-5, it just seemed like too good of a deal to pass up.

 

[Borek]

Can we find or prepare some kind of test that each one of us could print to take pictures of?

 

[turbo]

How available are bicycle-brand playing cards in Europe? When the cards are new, they are very flat, and the printing quality on the backs is generally excellent and consistent to foil cheats.

 

[Andy Resnick]

I ended up pulling the trigger on a Nikon d5100 w/kit lens and 55-300 DX VR lens last night. Amazon was having a killer deal where I got the stuff above and a camera bag for a little over $900 shipped. While I was leaning more towards getting the K-5, it just seemed like too good of a deal to pass up.

Excellent- I think you will be very happy with your decision!

 

[Andy Resnick]

Impressive Andy and yes, resolution is a function of lens resolution and sensor resolution. But maybe this should go in another thread?

Probably- I think some sort of "shoot out at the PF corral" thread makes sense for this.

Can we find or prepare some kind of test that each one of us could print to take pictures of?

All we need is a 'standard object' (or a few objects of different sizes if we want to go beyond macro shooting)

How available are bicycle-brand playing cards in Europe? When the cards are new, they are very flat, and the printing quality on the backs is generally excellent and consistent to foil cheats.

This is a good suggestion- there does not appear to be very much fine detail, but it's probably good enough. How about the hologram sticker on credit/debit cards? Just a suggestion for something commonly available...

Edit: I think it's worth saying that this isn't a "mine's bigger than yours" competition. There's a lot of optics knowledge that goes into setting up and quantitatively evaluating images, and this should be treated as an educational resource.

 

[turbo]

No, Andy, there is not a lot of really fine detail on the card backs, but they could make pretty decent long-distance targets. Just trying to think of something cheap and consistent. I long ago abandoned my wall-sized resolution chart (Edmunds, I think). When I was considering buying a zoom lens for my Olympus kit, I borrowed several from friends, including one who owns a camera shop, and used that chart to put the lenses through their paces. I gave up on the zoom idea right after developing and printing that film.

It was with a bit of trepidation that I jumped in with a Canon 30D and 100-400mm L, because of the expense, but I have a friend on another forum who shoots wildlife/birds from a blind with the 100-400, and the quality is scary-good. Zooms have come a LONG way in the last ~40 years.

 

[Andre]

I ended up pulling the trigger on a Nikon d5100 w/kit lens and 55-300 DX VR lens last night. Amazon was having a killer deal where I got the stuff above and a camera bag for a little over $900 shipped. While I was leaning more towards getting the K-5, it just seemed like too good of a deal to pass up.

Congrats. I hope you enjoy it. Actually a few years ago I had a similar experience, I was charmed by the multigadget Sony A350 and I had already ordered it (tentatively), I thought that the Canon 450D, the other one of the shortlist, was way too simple and a bit expensive in comparison, then I saw the test results and I never looked at the Sony again.

But that was just a moment in time and Nikon really had no competative model at that time that made it to my shortlist.

 

[Andy Resnick]

No, Andy, there is not a lot of really fine detail on the card backs, but they could make pretty decent long-distance targets. Just trying to think of something cheap and consistent. <snip>

I was thinking more and realized that there doesn't have to be fine detail, as long as the pattern edges are sharp- it's still possible to extract the modulation transfer function.

Cards sound good to me- consider that a second. How about the overseas folks?

 

[turbo]

I was thinking more and realized that there doesn't have to be fine detail, as long as the pattern edges are sharp- it's still possible to extract the modulation transfer function.

Cards sound good to me- consider that a second. How about the overseas folks?

Well, I guess a pack of cards isn't that much of an investment, so we could spend a few bucks to split a deck and send one or two cards to others who want to use those as test targets.

I'll send you some money to buy a deck of cards and pay for postage if you want to supply targets to Andre, Borek, and others. I can't get out too much due to my sensitivity to fragrance chemicals, so if you can handle the logistics, I'll pay the cost. It might be better to use cards within a single pack, to keep printing-variables as even as possible.

 

[Andy Resnick]

Well, I guess a pack of cards isn't that much of an investment, so we could spend a few bucks to split a deck and send one or two cards to others who want to use those as test targets.

I'll send you some money to buy a deck of cards and pay for postage if you want to supply targets to Andre, Borek, and others. I can't get out too much due to my sensitivity to fragrance chemicals, so if you can handle the logistics, I'll pay the cost. It might be better to use cards within a single pack, to keep printing-variables as even as possible.

I'm not sure I want to open "Andy Resnick's Playing Card Emporium" just yet :) Interested people can buy them directly online, anyways:

http://www.playingcardsandmore.com/bicycleplayingcards.aspx?gclid=CISzw92Py6sCFcjc4AodPgS63w

As it happens, I'm teaching the Advanced Optics Lab class next semester, and I've been trying to come up with a few labs to replace the (ahem) useless ones. For example, I'm thinking about a lab that centers on this:

http://www.skymall.com/shopping/detail.htm?pid=203769555&c=10323

and a "imaging performance" lab could also be a nice addition. I've come up with a preliminary procedure that I can debut here and see if it's worthwhile to fully develop for the course.

So, for anyone interested in quantitatively measuring the imaging properties of a lens, I'll work to get a rough draft started in a new thread. Details to follow, but all you need to start is a greyscale photo containing a light-to-dark transition.

 

[Topher925]

I know most people here shoot Canikon, but I can't say enough about how much I love my new Pentax. I finally got some time this morning to really try it out and while I'm still learning, I think the results are fantastic. Can't wait to get some nice lenses for it.

 

[Andy Resnick]

Excellent- hope you are having fun!