1 Feb, 2008 - Quality vs. Value - When is Enough Enough?

[Ray]

Ray,

Take care on this one. Have you seen high ISO results from the Sony A900?

Mark,
The internet has been abuzz with conflicting reports of performance ever since the A900 was released. On balance, I get the impression from numerous sources that the A900 is a bit lacking at high ISO. DXOMark seem to confirm this impression.

Comparing the 5D with the A900 on the DXOMark website, it seems that at ISO 3200 the A900 has the same noise as the 5D, but actually 2/3rds of a stop less DR, at the normalised print size of 8x12".

On the screen, pixel for pixel, the 5D has more than 1 stop greater DR than the A900, at ISO 3200. This is not encouraging for me. On the screen at 100%, the 5D also shows less noise than the A900, although not as much as one stop.

Edit: However, my new standard for low noise and high DR is the Nikon D700. Whether at a normalised print size of 8x12" or at full screen resolution, it either beats or equals the A900, mostly beats. I think the A900 matches the D700 only at base ISO in respect of SNR, DR, Tonal Range and Color Sensitivity.

Of course, resolution is better. But I was hoping for more.

[Mark D Segal]

Mark,
The internet has been abuzz with conflicting reports of performance ever since the A900 was released. On balance, I get the impression from numerous sources that the A900 is a bit lacking at high ISO. DXOMark seem to confirm this impression.

Comparing the 5D with the A900 on the DXOMark website, it seems that at ISO 3200 the A900 has the same noise as the 5D, but actually 2/3rds of a stop less DR, at the normalised print size of 8x12".

On the screen, pixel for pixel, the 5D has more than 1 stop greater DR than the A900, at ISO 3200. This is not encouraging for me. On the screen at 100%, the 5D also shows less noise than the A900, although not as much as one stop.

Edit: However, my new standard for low noise and high DR is the Nikon D700. Whether at a normalised print size of 8x12" or at full screen resolution, it either beats or equals the A900, mostly beats. I think the A900 matches the D700 only at base ISO in respect of SNR, DR, Tonal Range and Color Sensitivity.

Of course, resolution is better. But I was hoping for more.

Ray,

Here's where we get to the nub of an issue about user needs and results versus measurements. Doubtless there will be more to say about this generic question on L-L. However, in the context of the specific discussion here, there are a number of questions:

1) How often would you use ISO 3200 (yes I know good to have high quality functionality when you need it, but still.......)?
2) How likely is it that you would actually see and be disturbed enough by this difference ("not as much as 1 stop") in a print of the size you would normally print?
3) Do you really believe that seeing noise at 100% on your display is a reliable indicator of the noise you will see in the print? (I'm not saying no, but I'm a bit skeptical.)
4) The Nikon D700 lacks the resolution of the A900. In all these things there are trade-offs - sad to say - as you know. What do you need more: the nth degree of performance on noise or the higher resolution? With an Epson 7600 and if I make big prints with it, I'd definitely opt for the latter because I can reduce noise any number of ways, but I can't manufacture more resolution than the camera provides. Makes sense?

I come back to my basic view that the measurement stuff is really useful for general guidance, but must be complemented by visual inspection in print for the images and workflow we the photographers would use.

[Panopeeper]

I have a paractical concern only with the  benefits at the output stage of such factors as DR, SNR and high ISO performance

Of course, the only thing what counts at the end of the day is, how the real shots come out.

My claim is, that the capability of the camera can be objectively measured in shots targeted to that purpose better than judged from "real life" shots. If the DR of one camera is relevantly higher than that of another camera, then it depends only on you if that difference is taken advantage of.

However, I would not replace my camera for only 1/3 EV difference (other aspects aside, of course).

[Panopeeper]

Please look at the article and the example of sensor bloom, which shows quite clearly that there are very real, very practical effects of highlights that must be considered, not only by photographers, but also by theorists like yourself

Sensor blooming is caused by not only reaching but exceeding (usually by a lot) the wells' saturation. That is outside of the dynamic range.

I wasn't embarrassed on my behalf. I was - and still am - embarrassed on your behalf

I did understand your post.

[jani]

Sensor blooming is caused by not only reaching but exceeding (usually by a lot) the wells' saturation. That is outside of the dynamic range.

Okay.

So why do you then think that sensor bloom is immaterial?

[Panopeeper]

So why do you then think that sensor bloom is immaterial?

Immaterial for what? Blooming is not immaterial, because the effects range from ignorable to devastating. However, it is immaterial for the dynamic range, because the DR ends exactly at the pixel saturation at base ISO. At higher ISOs the range is even more limited simply by the numerical range of the pixel values, thus saturation is not a factor at all.

Back to your question: if you think blooming affects the dynamic range, then pls explain, how.

[ErikKaffehr]

Hi!

I have the A900 and it is a fine camera for landscape photography. From my experience I would agree that it should be used at low ISO and my impression is that DR is decaying with ISO. My experience doesn't conflict with the data on the DxO-mark site. On the other hand I don't have experience with other cameras than Sony/Minolta. Some raw converters may yield better results than ACR. Noise Ninja cleans up noise but cannot help with the limited DR.

Best regards
Erik

Mark,
The internet has been abuzz with conflicting reports of performance ever since the A900 was released. On balance, I get the impression from numerous sources that the A900 is a bit lacking at high ISO. DXOMark seem to confirm this impression.

Comparing the 5D with the A900 on the DXOMark website, it seems that at ISO 3200 the A900 has the same noise as the 5D, but actually 2/3rds of a stop less DR, at the normalised print size of 8x12".

On the screen, pixel for pixel, the 5D has more than 1 stop greater DR than the A900, at ISO 3200. This is not encouraging for me. On the screen at 100%, the 5D also shows less noise than the A900, although not as much as one stop.

Edit: However, my new standard for low noise and high DR is the Nikon D700. Whether at a normalised print size of 8x12" or at full screen resolution, it either beats or equals the A900, mostly beats. I think the A900 matches the D700 only at base ISO in respect of SNR, DR, Tonal Range and Color Sensitivity.

Of course, resolution is better. But I was hoping for more.

[Ray]

Ray,
Here's where we get to the nub of an issue about user needs and results versus measurements. Doubtless there will be more to say about this generic question on L-L. However, in the context of the specific discussion here, there are a number of questions:
1) How often would you use ISO 3200 (yes I know good to have high quality functionality when you need it, but still.......)?

Mark,
Perhaps the question should be, "How often would I use ISO 3200 if image quality were better?" The answer would be, surprisingly often, particularly when using the Canon 100-400 IS for wildlife. In practice I'm reluctant to go beyond ISO 1600 with my current Canon cameras. However, I would consider it a worthwhile upgrade if a new camera could give me the same noise and DR at ISO 3200 that I currently get with my 5D at ISO 1600.

The 100-400 is effectively an F8 lens at 400mm for decent image quality. F8 at ISO 1600 usually produces sharper results than F5.6 at ISO 800 with the same shutter speed.

I find when choosing a camera body one should always consider the options for available lenses. The availability of a particularly good lens in the range that one uses frequently should affect one's choice of camera body.

I see nothing from either Nikon or Sony that is an improvement on the Canon 100-400mm. The 400mm prime lenses are too expensive, or too heavy or too inflexible for my style of shooting. I think there's a big gap in the market for an affordable, lightweight, F5.6 zoom in the range of 100-400, that is sharp at full aperture.

2) How likely is it that you would actually see and be disturbed enough by this difference ("not as much as 1 stop") in a print of the size you would normally print?

Possibly not much, comparing the 5D with the A900. The additional resolution of the A900 would always be appreciated. However, I'm reminded of the performance of my first DSLR, the Canon D60. I never went above ISO 400 because the results were so obviously degraded. In fact, I was often reluctant even to use ISO 400 on some occasions when I probably needed to, for a fast shutter speed with a lens like the Canon 100-400. In such situations I sometimes found myself in a state of indecision. Is a 200th sec exposure at ISO 200 likely to produce better results than a 400th at ISO 400?

You can imagine how pleased I was when I discovered that the 20D at ISO 1600 could actually produce better image quality at ISO 1600 than the D60 at ISO 400; better in terms of noise, color saturation, tonality, resolution and dynamic range; all the things that count.
The fact is, if a camera doesn't produce good results at high ISO, then you don't use it at high ISO. I don't want to feel limited in that respect.

3) Do you really believe that seeing noise at 100% on your display is a reliable indicator of the noise you will see in the print? (I'm not saying no, but I'm a bit skeptical.)

No. One usually has to reduce the magnification on the monitor to 67%, or 50% or less, depending on the print size one intends making. But I'm not happy that an A4 size print from a 5D file would have about 2/3rds of a stop greater DR than the A900, according to DXO tests.

4) The Nikon D700 lacks the resolution of the A900. In all these things there are trade-offs - sad to say - as you know. What do you need more: the nth degree of performance on noise or the higher resolution? With an Epson 7600 and if I make big prints with it, I'd definitely opt for the latter because I can reduce noise any number of ways, but I can't manufacture more resolution than the camera provides. Makes sense?

It's sometimes a difficult decision. When the pros and cons seem about even with regard to the camera bodies, one should consider the lenses. There are too many issues regarding the A900, about which I'm not exactly happy. In addition to the 'less than stellar' performance at high ISO, it has no Live View and no Sony or Zeiss wide angle zoom on a par with the Nikkor 14-24/2.8.
As far as I can tell, the A900 also doesn't have that excellent feature of auto-ISO in manual mode. You select the aperture and shutter speed, and the camera selects the ISO for a correct exposure.

The A900 also doesn't have the wide range of autobracketing options that the D700 has. I believe it's either 3 frames up to a 2EV interval, or 5 frames up to 0.7EV interval. The D700 has up to 9 frames at up to 1EV intervals. I believe much of the criticism levelled at Adobe's 'Merge to HDR' is due to people trying to merge too few shots with a too great an EV interval. I could be wrong, though.

[Dan Vincent]

There is the new Sony 16-35 f/2.8 ZA and 70-400 f/4-5.6 G. They are just newly available, though, and might be hard to find until supplies ramp up in the coming months.

[BernardLanguillier]

It's sometimes a difficult decision. When the pros and cons seem about even with regard to the camera bodies, one should consider the lenses. There are too many issues regarding the A900, about which I'm not exactly happy. In addition to the 'less than stellar' performance at high ISO, it has no Live View and no Sony or Zeiss wide angle zoom on a par with the Nikkor 14-24/2.8.

I have found Live view on th D3x to be one important feature when one is relly trying to get the peak focus exactly where it should be.

AF is often not accurate enough for shots like the following, and it is next to impossible to focus manually at the exactly right location.



Cheers,
Bernard

[Ray]

I have found Live view on th D3x to be one important feature when one is relly trying to get the peak focus exactly where it should be.

AF is often not accurate enough for shots like the following, and it is next to impossible to focus manually at the exactly right location.
Cheers,
Bernard

Bernard,
Isn't this a stitched image? You've shown it before. I think it's an excellent technical example of what's now possible with stitching programs. The straight lines are straight and even the focussing looks accurate after stitching, although one can't really tell without 100% crops, However, I have to say that the subject and composition leave me somewhat underwhelmed.

Below is a shot of fine detail taken with the 5D. No stitching required. This is how I found her when I came home for dinner   .

[attachment=11483:9717.jpg]

[Rob C]

Below is a shot of fine detail taken with the 5D. No stitching required. This is how I found her when I came home for dinner   .

[attachment=11483:9717.jpg]

Lies, more lies and damned lies! I can see the stitches from here - all along the hemline.

Rob C

[Mark D Segal]

Bernard,
Isn't this a stitched image? You've shown it before. I think it's an excellent technical example of what's now possible with stitching programs. The straight lines are straight and even the focussing looks accurate after stitching, although one can't really tell without 100% crops, However, I have to say that the subject and composition leave me somewhat underwhelmed.

On the contrary, I think it's a fine photograph. It has real "atmosphere", good formal composition and the control of focus is masterful.

Turning to your counter-example Ray, well, what to say about that!....................

Mark

[Ray]

On the contrary, I think it's a fine photograph. It has real "atmosphere", good formal composition and the control of focus is masterful.

Turning to your counter-example Ray, well, what to say about that!....................

Mark

Well, Mark, you could say "Cute!"   . Bernard's image attempts to demonstrate mastery of stitching, but fails because the image presented is so small. It could, for all we know, be just a single shot from any camera with a wide-aperture lens.

Mine actually is a single shot. The stitching is implied, not of multiple images, but of the fabric in the lady's dress ( as Rob noted).

Bernard's image demonstrates great mastery of foussing. Mine was autofocussed at F8. But there's a different type of mastery at work here. Do you realise how difficult it was to train this lady to greet me in this fashion when I return home from work each day?  

Okay! Enough silliness! Back to the theme of the thread.

If we consider 3 major issues regarding image quality; resolution, dynamic range and focussing accuracy; we find that resolution can be increased to taste through stitching, provided the subject is static. Dynamic Range can be increased through exposure bracketing, provided the subject is static. Focussing accuracy can be increased either through Live View, or through the autofussing micro-adjustment of the latest Canon models.

The advantage of any higher pixel count camera in respect of resolution, dynamic range and autofocussing accuracy, therefore only applies to subjects that are moving.

If the subject is stationary, you simply don't need an expensive, high resolution camera. And certainly not a ridiculously expensive MFDB. You don't even need a D3X. A D700 will be more than adequate, provided the subject lends itself to the stitching process. Focussing accuracy is perhaps in a different ball park. Liveview and micro-adjustment of autofocus with respect to any lens, seems the best solution for this factor. But again, manual focussing is more difficult if the subject is not static.

Agreed?

[Mark D Segal]

Well, Mark, you could say "Cute!"   . Bernard's image attempts to demonstrate mastery of stitching, but fails because the image presented is so small. It could, for all we know, be just a single shot from any camera with a wide-aperture lens.

Mine actually is a single shot. The stitching is implied, not of multiple images, but of the fabric in the lady's dress ( as Rob noted).

Bernard's image demonstrates great mastery of foussing. Mine was autofocussed at F8. But there's a different type of mastery at work here. Do you realise how difficult it was to train this lady to greet me in this fashion when I return home from work each day?  

Okay! Enough silliness! Back to the theme of the thread.

If we consider 3 major issues regarding image quality; resolution, dynamic range and focussing accuracy; we find that resolution can be increased to taste through stitching, provided the subject is static. Dynamic Range can be increased through exposure bracketing, provided the subject is static. Focussing accuracy can be increased either through Live View, or through the autofussing micro-adjustment of the latest Canon models.

The advantage of any higher pixel count camera in respect of resolution, dynamic range and autofocussing accuracy, therefore only applies to subjects that are moving.

If the subject is stationary, you simply don't need an expensive, high resolution camera. And certainly not a ridiculously expensive MFDB. You don't even need a D3X. A D700 will be more than adequate, provided the subject lends itself to the stitching process. Focussing accuracy is perhaps in a different ball park. Liveview and micro-adjustment of autofocus with respect to any lens, seems the best solution for this factor. But again, manual focussing is more difficult if the subject is not static.

Agreed?

First on the image: Bernard should clarify the nature of the subject matter and why he approached it with a stitching solution. There may be more to this than meets the eye simply looking at an end result posted in a web forum.

Second, on the main theme of the thread: quality/value/ ideal specs for the purpose: Whether or not the subject matter is in motion I think is one variable amongst others to be considered in the appropriate choice of specs. I don't think it is necessarily determinative one way or another, but I agree that more technical flexibility is helpful for this kind of image. You may still want to use a hi-res camera for stationary subjects - it depends on how much resolution you want at what print size and the extent to which you are comfortable with resampling data.

[Ray]

Whether or not the subject matter is in motion I think is one variable amongst others to be considered in the appropriate choice of specs. I don't think it is necessarily determinative one way or another, but I agree that more technical flexibility is helpful for this kind of image. You may still want to use a hi-res camera for stationary subjects - it depends on how much resolution you want at what print size and the extent to which you are comfortable with resampling data.

Mark,
I think the point I'm trying to make is that one has to be careful with one's own weighted assessment of a camera when there's any obsessional factor involved. Resolution is perhaps the most obsessional of all camera specs. It's the most easily appreciated aspect of camera performance because, however small the improvement, one can usually see it. If it's not visible at 100% on one's monitor, then it might be at 200%. If it's not visible at 200%, then it might be at 400%. The relevance to the print of such increases is another matter.

The fact is, provided that an increase in pixel count doesn't have any downside, then it has to be a good thing, but by itself it might not be sufficient reason to upgrade. At least, that is what I have found to be the case. When I look back on all my purchases of DSLRs, I find that resolution was only a major issue with my very first purchase, the 6mp Canon D60. The 3mp of the D30 was just too few pixels to persuade me to part with my money, so I waited for the upgrade. Apart from a doubling of pixel count, the D60 had no significant performance advantage over the D30.

However, that has not been the case with subsequent upgrades. With each upgrade, the increase in pixel count has been fairly low on my list of performance considerations. My latest Canon purchase, the 50D has fully 50% more pixels than my 40D which I bought on impulse in Bangkok about a year earlier. That increase in pixel count alone would not have been sufficient reason for me to buy the 50D. In conjunction with the autofocus micro-adjustment feature and the high resolution LCD screen, which is just amazing at 10x magnification with a 400mm lens, that 50% increase in pixel count tipped the balance in favour of my being able to justify a purchase.

If I can justify the purchase of a 50D as an upgrade from the 40D, then surely I can justify the purchase of a 5D2 as an upgrade from the 5D1? Oddly enough, I'm having difficulty doing this, despite all of the obvious advantages of the 5D2. However, the increase in pixel count is only slightly greater, as a percentage, than the increase of the 50D over the 40D. We're looking at about 65% as opposed to 50%.

In terms of lp/mm resolution, those figures represent a 22% increase for the 50D (over the 40D) and a 28% increase for the 5D2 (over the 5D1). But these increases in resolution can only be achieved if there's a corresponding and equal increase in lens resolution, which there rarely is. Using the same lenses with the 5D2 and 50D as one would with the 5D1 and 40D, one could cut those resolution increases in half. That is, an 11% increase for the 50D and a 14% increase for the 5D2, approximately.

I'm sure you'll agree, the true benefit of those increase in pixel count are now beginning to look a lot less persuasive.

I shan't mention the water resistant issues and the limited functionality of the 5D2 video, which are also cause for hesitation.

When dealing with such relatively small increases in 'system' resolution, the quality of the type of lenses one uses most frequently should surely be a major concern. In my own case, the 3 lenses I have found, over the years, that I use most often are the Sigma 15-30, the Canon 24-105 IS (before that the Canon 28-135 IS), and the Canon 100-400 IS.

The Sigma 15-30 is simply not good enough for the 5D2 at the borders and corners which are bad enough on the 5D1. The 5D2 would just accentuate that poor performance.

The 100-400 IS I use mostly on a cropped format DSLR, such as the 20D or 40D, but in future the 50D. I think there can be no doubt that a 50D image with that lens at 400mm will show superior detail to a 5D2 image cropped to the dimensions of the 50D sensor.

So where does that leave me? Perhaps on my next photographic trip I shall be carrying 3 cameras. The D700 with 14-24, the 5D2 with 24-105, and the 50D with 100-400.

[jjj]

Jeremy and Gabor,

The use of expletives and recourse to personal insults is an unwelcome intrusion into an otherwise useful discussion. Both of you are serious enough professionals to sustain your reputations and enhance your impact on your readers by being moderate on the keyboard. I would suggest there is merit to keeping this discussion polite, dispassionate, impersonal and technical. Otherwise you annoy other readers and do yourselves a professional disservice.

What has Gabor got to do with Ray's repeated and innaccurate insistence I'm constantly off my head on drugs, as there is no other reason I could possiblyI post the things I do? Simply having a different point of view based on my drug free experiences, never occured to him.
And if I'm more than a little annoyed with such made up, nonsense that is being used instead of rational argument, is that surprising?
I also find it baffling that Ray escapes censure for his inflammatory comments.

[BernardLanguillier]

First on the image: Bernard should clarify the nature of the subject matter and why he approached it with a stitching solution. There may be more to this than meets the eye simply looking at an end result posted in a web forum.

Thanks for your kind support Mark.

I selected this image because IMHO it would have been very hard to get the exact focus point with anything but live view. Considering the proximity of the subject, the AF focus points are just too large to enable accurate focus on the exact point where it needs to be set. This matters because I am planning to print this large.

I decided to use stitching for this image for 4 main reasons:

- aspect ratio, 3:2 was just too square for this subject per my taste,
- resolution, this one is pretty self explanatory,
- quality of bokeh. Out of the lenses I own and had with me during that trip, the Zeiss 100 mm f2.0 wins hands down bokehwise, and it is obviously too long to get a wide angle feel with a single frame,
- there was no reason whatsoever not to stitch.

Second, on the main theme of the thread: quality/value/ ideal specs for the purpose: Whether or not the subject matter is in motion I think is one variable amongst others to be considered in the appropriate choice of specs. I don't think it is necessarily determinative one way or another, but I agree that more technical flexibility is helpful for this kind of image. You may still want to use a hi-res camera for stationary subjects - it depends on how much resolution you want at what print size and the extent to which you are comfortable with resampling data.

Absolutely. Adding to what you are saying, it might be interesting to think about the following question "what is a static object?...".  Indeed, many of my subjects are static, but the light changes in intensity, color and direction. The following image was hard to stitch because of the fact that the sun moved enough in the sky between the different rows of frames that the shadows of the trees were not matching... this is due to the fact that each row is in fact a depth of field stacking made up of 5 to 9 frames...



Cheers,
Bernard

[Mark D Segal]

Mark,
I think the point I'm trying to make is that one has to be careful with one's own weighted assessment of a camera when there's any obsessional factor involved. Resolution is perhaps the most obsessional of all camera specs. It's the most easily appreciated aspect of camera performance because, however small the improvement, one can usually see it. If it's not visible at 100% on one's monitor, then it might be at 200%. If it's not visible at 200%, then it might be at 400%. The relevance to the print of such increases is another matter.

The fact is, provided that an increase in pixel count doesn't have any downside, then it has to be a good thing, but by itself it might not be sufficient reason to upgrade. At least, that is what I have found to be the case. When I look back on all my purchases of DSLRs, I find that resolution was only a major issue with my very first purchase, the 6mp Canon D60. The 3mp of the D30 was just too few pixels to persuade me to part with my money, so I waited for the upgrade. Apart from a doubling of pixel count, the D60 had no significant performance advantage over the D30.

However, that has not been the case with subsequent upgrades. With each upgrade, the increase in pixel count has been fairly low on my list of performance considerations. My latest Canon purchase, the 50D has fully 50% more pixels than my 40D which I bought on impulse in Bangkok about a year earlier. That increase in pixel count alone would not have been sufficient reason for me to buy the 50D. In conjunction with the autofocus micro-adjustment feature and the high resolution LCD screen, which is just amazing at 10x magnification with a 400mm lens, that 50% increase in pixel count tipped the balance in favour of my being able to justify a purchase.

If I can justify the purchase of a 50D as an upgrade from the 40D, then surely I can justify the purchase of a 5D2 as an upgrade from the 5D1? Oddly enough, I'm having difficulty doing this, despite all of the obvious advantages of the 5D2. However, the increase in pixel count is only slightly greater, as a percentage, than the increase of the 50D over the 40D. We're looking at about 65% as opposed to 50%.

In terms of lp/mm resolution, those figures represent a 22% increase for the 50D (over the 40D) and a 28% increase for the 5D2 (over the 5D1). But these increases in resolution can only be achieved if there's a corresponding and equal increase in lens resolution, which there rarely is. Using the same lenses with the 5D2 and 50D as one would with the 5D1 and 40D, one could cut those resolution increases in half. That is, an 11% increase for the 50D and a 14% increase for the 5D2, approximately.

I'm sure you'll agree, the true benefit of those increase in pixel count are now beginning to look a lot less persuasive.

I shan't mention the water resistant issues and the limited functionality of the 5D2 video, which are also cause for hesitation.

When dealing with such relatively small increases in 'system' resolution, the quality of the type of lenses one uses most frequently should surely be a major concern. In my own case, the 3 lenses I have found, over the years, that I use most often are the Sigma 15-30, the Canon 24-105 IS (before that the Canon 28-135 IS), and the Canon 100-400 IS.

The Sigma 15-30 is simply not good enough for the 5D2 at the borders and corners which are bad enough on the 5D1. The 5D2 would just accentuate that poor performance.

The 100-400 IS I use mostly on a cropped format DSLR, such as the 20D or 40D, but in future the 50D. I think there can be no doubt that a 50D image with that lens at 400mm will show superior detail to a 5D2 image cropped to the dimensions of the 50D sensor.

So where does that leave me? Perhaps on my next photographic trip I shall be carrying 3 cameras. The D700 with 14-24, the 5D2 with 24-105, and the 50D with 100-400.

Ray,

I take your general point that lens quality can be the binding constraint, but let's look at some details of the options you discuss. Your Canon 5D body was worth round-about 1500 USD on eBay last week when several sold. The 5DMk2 is 2700 USD (NYC), so in the USA, which I know is not Australia (but Bangkok is competitive with NYC for Canon gear - better get back there so you can stock-up on gear and photograph some more of those pretty ladies waiting to serve you tea   ), the 5Dmk2 would carry a net incremental cost of about 1200 USD plus some sales tax or shipping, whatever. For that you would be getting, as you say, a 28.6% increase of resolution allowing to print one more sq.ft. of paper at 240 PPI without any resampling. IF you use your 24-105 L lens at its optimal aperture, the effective resolution of the lens should be adequate for the 5DMk2 sensor. Hard to know for sure without detailed testing of both bodies using the same lens, but having been there between the 1Ds and the 1DsMk3, this is my impression - remembering the aperture is important. According to the DxO results, there is what I would estimate to be a moderate improvement in sensor quality between the two models. These factors, combined with all the other improvements between the two cameras suggests that there may be more of a case to up-grade than you give it credit for. Even if you don't sell your 5D, it would give you a full-frame reserve camera, and you could probably sell all the rest of your cameras unless one of them particularly suits a special purpose not equally well-fulfilled by the FF gear. There's nothing preventing you from using a cropped portion of your FF 5DMk2 where you would have used the 50D, and you should get the same IQ - if not - better - as you would from a 50D. There's a substantial improvement of DxO rating for a 5DMk2 sensor compared with the 50D sensor.

Mark

[Mark D Segal]

Thanks for your kind support Mark.

I selected this image because IMHO it would have been very hard to get the exact focus point with anything but live view. Considering the proximity of the subject, the AF focus points are just too large to enable accurate focus on the exact point where it needs to be set. This matters because I am planning to print this large.

I decided to use stitching for this image for 4 main reasons:

- aspect ratio, 3:2 was just too square for this subject per my taste,
- resolution, this one is pretty self explanatory,
- quality of bokeh. Out of the lenses I own and had with me during that trip, the Zeiss 100 mm f2.0 wins hands down bokehwise, and it is obviously too long to get a wide angle feel with a single frame,
- there was no reason whatsoever not to stitch.



Absolutely. Adding to what you are saying, it might be interesting to think about the following question "what is a static object?...".  Indeed, many of my subjects are static, but the light changes in intensity, color and direction. The following image was hard to stitch because of the fact that the sun moved enough in the sky between the different rows of frames that the shadows of the trees were not matching... this is due to the fact that each row is in fact a depth of field stacking made up of 5 to 9 frames...

So if a single frame were good enough for the target print size, this image would be a lot easier to come up with. Using a Phase one P65+ would make sense here.

Cheers,
Bernard

I knew there would be a logical explanation and you didn't disappoint. Thanks for the insights and nice shot of the log in the snow.

Mark