Upgrading from XSi to 5D Mk2

Discussion in 'Canon EOS' started by anunes, May 2, 2009.

  1. Hi everybody, I'm new to this forum (my first post) but already impressed with the level of cooperation I could see in other posts!
    I used to take photos when I was a young adult with an old EOS 1000F which I still have. I have recently resumed this passion and bought a Rebel XSi for begining - to be very honest I started to study and read about every aspect of modern photograph AFTER I bought the camera and rapidly regreted about not going to a better body - so I decided to invest a little more and went straight to the 5D Mk2.
    My doubts are:
    - Should I keep the XSi as a backup? Is there any benefit of having a full frame as well as a crop sensor body? Can I use my lenses taking advantages from the different sensor sizes?
    A. Nunes (please take a look at my portfolio at www.anunesphotography.com and live your impressions)
  2. I'd say yes.
    For one it's lighter. (To me using a rebel and a small prime is great fun!)
    Plus it gives a tele lens a little more reach because it has smaller pixels.
    (But not a lot so the difference is not that big. 5DII has 6.4u pixels and the XSi has 5.2u pixels so lenses are 23% longer on the XSi, never mind what the crop factor says... If it was an XT or a 30D the difference would be 0.)
    Plus in some situations (especially if you're going to shoot something important / for money) it's best to have a backup body.
  3. Keep the XSi. It's an excellent camera, and the smaller size and wieight will make for a nice travel camera, or whenever you don't want to lug the 5D II. No brainer here.
  4. Alexandre:
    If you like it, keep it.
    I had an XTi at one point. Bought it because it was small and lightweight. In the end, I hated not having a control wheel on the back. Setting manual exposures was a pain. So I sold it.
    The crop factor between the XSi and 5D2 is 1.6. The size of the individual pixels doesn't affect how much of the image circle is projected on the sensor. How much of the lens image is resolved on the sensor determines the crop factor. Think of it this way: if the individual pixels were exactly the same size, would you say there is no crop factor?
  5. Eric,
    Matthijs is right on this point. The pixel size does actually matter more than the crop value if you are interested in telephoto length for a given resolution . If the pixels are the same size, then the full-frame 5D has an advantage , because you can always crop the image to give the exact same result as a 1.6x crop image, whereas if you used a crop body, you'd lose the option of keeping the extra area of the image circle and you wouldn't actually be recording any more detail. It is only when the pixel size gets smaller that you yield a higher apparent magnification for a given resolution (at the cost, of course, of things like noise).
    People are used to the fallacious thinking that because a crop body image seems "zoomed in" due to the crop, then that means essentially longer telephoto performance on crop bodies. That is only true if you think in terms of the overall image, and not in terms of cropping a full-frame image to match a crop body image - assuming, of course, identical physical pixel densities. But just to repeat the point one last time, yes, a camera body does in fact "enhance" telephoto length for a given lens when it has a higher pixel density.
  6. Guys, I'm kind of lost here. Lets use an example to have it cleared:
    XSi with a 100mm lens
    5DMkII with the same 100mm lens
    I understand that the Xsi has a crop factor of 1.6x making the lens above appears like a 160mm used in a FF body. I would have more zoom using the XSi and consequently I would loose some of the overall image when compared to the same photo (taken from the same position) with the 5DMkII, although the resolution (# of pixels per line) of the XSi would be 23% better than the FF. Is that correct?
  7. Bodies may come and go, but lenses are forever.
    Well, unless you drop them. But let's not speak of such horrors, shall we?
    So get yourself some really good glass, maybe two L zooms, and some primes around the focal lengths you tend to shoot at. All of them should be EF, not EF-S (which are specific to APS-C format as seen in the EOS Rebels and 20/30/40/50D series). You don't mention how much money you have, but since you basically told us you bought two camera bodies totaling around 3000 USD plus whatever lenses might go with those, I'm taking that as a hint that you can afford some really high quality glass. Some good choices are 24-70/2.8L, 24-105/4L IS, 70-200/2.8L IS for your zooms. Don't ignore the non-L primes, either, some are very impressive like 50/1.4 and 85/1.8, for what they cost.
    I'll take a different angle on your question--considering that some of these lenses are easily 2-3x the resale cost of a lightly used XSi, you're not recovering a significant amount of value by reselling it, as opposed to the potential convenience of having multiple bodies. It's not really about which body is better than the other--truth is, the 5DmkII is superior in nearly every way, save for the weight.
    Sensor pitch comparison (linear distance between centers of adjacent whole RGB pixels, not subpixels)
    EOS Digital Rebel XTi, EOS 50D = 4.6717 micrometers/pixel
    EOS Digital Rebel XSi = 5.1966 micrometers/pixel
    EOS 5D Mark II, EOS 1Ds Mark III = 6.4103 micrometers/pixel
    EOS 5D (Mark I) = 8.2418 micrometers/pixel
    The difference is significant, with each pixel of the 5DmkII being approximately 52% larger in sensing area than each equivalent pixel of the XSi. Ignoring for the time being the fact that the increased signal-to-noise ratio leads to less noisy images, if you were to center crop the 5DmkII to the same pixel dimensions as the XSi, the same lens on an XSi would have a 1.233x longer effective focal length than the cropped 5DmkII. That is to say, on a pixel-by-pixel basis (not on a sensor size basis), the XSi would have a focal length conversion ratio of 1:1.233. On a sensor size basis it would be about 1:1.622, the more familiar number (which would be the case if the 5DmkII image were scaled, not cropped, to match the XSi pixel dimensions).
    This bit of math demonstrates why the APS-C sensors, with their smaller linear pitch, are more demanding of a given lens than a 35mm sensor. You need much sharper glass.
  8. Alex,

    You seem to have the idea. If you used the same lens on the XSi and the 5DII, cropped the 5DII image to give identical subject framing (field of view), and then printed both images, the XSi image would actually have more detail (either printed larger at same dpi or have more dpi if printed same size) since it has a higher pixel density. But as Peter mentioned, the higher pixel density might not matter if all it does is lead to noisier images or images that reveal lens flaws rather than record more real detail. And frankly, the many other considerations of sensors (crop factor / wide-angle performance, low-light performance, features of the body the sensor is actually housed in, etc.) vastly outweigh this consideration for almost everyone. The only reason it was ever brought up was to point out that it could potentially be one reason to reach for the XSi over the 5D in certain situations, if you already owned both. Namely, if you were not going to print very large at all, and you were trying to squeeze every ounce of reach out of your current equipment, and the light was good, and you had a very good lens... then the XSi will give a bit more ultimate detail.
    I am totally with Peter's suggestion to invest in lenses first, bodies later. And I also think that it makes complete sense to keep a crop body around as a backup - or even as a first choice for certain kinds of shooting. If it's within your budget, there's no reason not to. And if it IS a question of budget, well, that's something only you can really decide.
  9. Peter, thanks for your post - it was really useful!
    As you've touched in the lenses subject, I'd like to have opinions about the right set of lenses to have, taking in consideration what I already got, which is:
    - a very old Canon EF 35-80mm/1:4-5.6
    - a very old Canon EF 75-300mm/1:4-5.6
    - a Sigma DC 18-125mm/1:3.8-5.6 HSM
    - a Sigma APO Macro 70-300mm/1:4-5.6 (I've got as a gift from a friend)
    - a Canon Macro EF 100mm/1:2.8 USM
    I really don't have specific photos I like to take, basically I'd like to have a set that enables me a good performance in any field (portraits, macros, landscape, etc)
  10. Alexandre:
    "although the resolution (# of pixels per line) of the XSi would be 23% better than the FF. Is that correct?"
    dpreview's tests show both the XSi and 5D resolve about the same detail. The 5D2 resolves about 30% more. Pixel density doesn't translate into real world resolution improvements.
    You may have 23% more pixels. But you're not resolving 23% more detail. You have more pixels. If the smaller pixels aren't resolving more detail, you're not gaining anything by using smaller pixels. In fact, you're getting a noisier image at the same ISO and losing dynamic range.
  11. Okay, I'm seeing a bit of confusion about this subject that maybe my last post did not help to clarify. For demonstration purposes, we will ignore the effects of noise and signal (light) strength and assume an ideal optical system throughout.
    Suppose we have two identical 100mm lenses, each attached to a different camera. Camera A has a (hypothetical) sensor size of 50x50mm and Camera B is 25x25mm (for simplicity of numbers; we'll assume the image circle of the lens covers the entire sensor regardless of its size). However, A captures a 500x500 pixel image, and B captures a 250x250 pixel image. That is to say, both cameras have the same size pixels.
    There are two intuitive ways you can compare the resulting image from each camera. The first way is to center crop A's image to 250x250, to compare it to B (you clearly can't "anti-crop" B's image to compare it to A--the information isn't there!). It should be obvious that if you did this, the resulting two images are equivalent, because all you did was crop out the exact part of A's image that was not captured by B's sensor.
    The second way to compare is to resample A's image down to the same pixel dimensions as B. But as you do this, the effective magnification changes, which in turn can be expressed as a change in equivalent focal length. What this means is if you were instead to have used a 50mm lens on B, then the resulting image would be identical to the resampled A image:
    A: 100mm lens @ 50x50mm sensor @ 500x500 px -> downsampled to 250x250 px
    = B: 50mm lens @ 25x25mm sensor @ 250x250 px.
    Now, suppose you had a camera C, with a 25x25mm sensor, but it captures a 500x500 px image, same size as camera A. That means each pixel of A is FOUR times the area of each pixel of C. Using either comparison method (crop or scale), there is nothing to do, since both images are captured at 500x500 px. But what do you see ? Clearly the image captured by A's sensor must encompass a larger portion of the image circle cast by the same 100mm lens than what was captured by C's sensor. So the result is C has a narrower field of view than A. It would appear that C's image was taken with a 200mm lens; that is to say, if we had placed a 200mm lens on camera A, we would see the same result as C:
    A: 200mm lens @ 50x50mm sensor @ 500x500 px
    = C: 100mm lens @ 25x25mm sensor @ 500x500 px.
    Now let's compare B and C. In this case, the sensor sizes are equal (25x25mm), and hence they capture the same exact portion of the image circle, but C resolves 500x500 px compared to 250x250 px for B. We clearly have to either crop or downsample C's image. In the former case, cropping now effectively multiplies C's focal length by 2:
    B: 200mm lens @ 25x25mm sensor @ 250x250 px
    = C: 100mm lens @ 25x25mm sensor @ 500x500 px -> cropped to 250x250 px.
    But if you use the second method and downsample C's image, it should again be obvious that the resulting images are equivalent (after all they captured the same part of the image circle):
    B: 100mm lens @ 25x25mm sensor @ 250x250 px
    = C: 100mm lens @ 25x25mm sensor @ 500x500 px -> downsampled to 250x250px.
    Bear in mind, none of this discussion considers real-world effects of noise and its impact on perceived image quality. That is something not easily mathematically quantifiable, which is why we have so many debates about APS-C vs 35mm sensors.
    So now I close with an exercise for the reader: can you (1) identify each variable that contributes to the angle of view in each of the above scenarios, and (2) determine a mathematical relationship among those variables that tells you when they are equivalent?
  12. Alexandre,
    You want to take portraits, macro, and landscape? Surely you are not talking about trying to do this with one lens. That's at the bare minimum two lenses (wide angle prime, 100mm macro) and in order to do each one well, that kind of range is best covered by at least five lenses (wide angle prime, wide-to-normal zoom, normal prime, normal-to-telephoto zoom, and one dedicated telephoto macro that might double as a portrait lens).
    Of all your lenses, only your 100/2.8 macro has an aperture significantly wider than f/4 (the f/3.8 doesn't really count, that's almost an f/4, it's less than 1/3 EV more open--unless it's a typo?). If you went with a 50mm/1.4 you just gained +3 EV over anything else you have in that focal range, and your DOF will be significantly shallower leading to more creative possibilities. Plus, most of your lenses are zooms. It seems to me that you're missing out on some spectacular sharpness and contrast that a prime lens can offer.
    I'm afraid to say this, but I think you just spent a lot of money on a body that almost none of your existing lenses can do justice. In fact, I think perhaps you may have been dissatisfied with the performance of your XSi, thinking the problem is the body, when the reality is that you haven't really begun to see what kind of images are possible with some sharp glass! Now don't get me wrong, by all means stick with the bodies you've got, but please please do yourself a favor and just try a nice open prime, take the time to become familiar with it. Given what you have already I would suggest the 50/1.4 as it is reasonably cheap (your wallet must be much lighter after that 5DmkII--by comparison this lens is a drop in the bucket!), insanely sharp (compared to your zooms) beginning at f/4 and quite lovely at wider apertures, and is not too close in length to your 100/2.8 macro. I would hesitate to drop an additional 1000-2000 USD on an L zoom right now. If you are absolutely serious about photos and you have cash burning a hole in your pocket right now, then buy an L prime, say the 24/1.4LII, 50/1.2L, or 135/2L.
  13. Unless you shoot professionally I would consider selling the Rebel body. I know this sound like heresy and in my case is hypocritical as i own 5 different EOS bodies, 7 FD and a raft of medium format gear. My suggestion of selling it is that unlike film bodies (when film was the only option) digital bodies devalue and become obsolete very quickly. The XSi will lose it's remaining value very quickly and you will find that you much prefer shooting with the 5D II. While the Rebel is smaller it is not dramatically so - especially if you end up using full frame lenses on the Rebel as you are buying them for the 5D II. I am with Peter sell the XSi and buy a lens.
  14. On lenses:
    I'd hazard a guess that the 100/2.8 Macro will work perfectly and the Sigma APO 70-300 might do o.k.
    Your other lenses might need improvement but you're the only judge of that. (Maybe the 35-80 is good enough.)
    That said:
    The 50 1.4 would be a very nice addition to the 5DII.
    If you're on a budget now the 24-85 might be an o.k. choice as well.
    A combination of 17-40/4L plus 50/1.4 plus the lenses you already have covers a nice range of focal lenghts too.
    (Note that the 17-40 is best when stopped down a little.)
    Kind regards, Matthijs.
  15. A. Nunes, I'm kind of in the same situation and I'm keeping my XTI. It is nice to have two bodies laying around. The rebel will be nice for kids parties, and such, that need a small amount of flash. As the 5D II does not have a pop-up flash and putting on a canon flash makes the 5D II that much larger and harder to weild in some situations. I'm concerned with the interface and switching between them as they are so different, but I think the XTI will not be used very often so it should not be a huge problem. I think your in a better position, as the XSI has a larger lcd and is a little nicer camera.
    To the people who argue for the 5D II in the crop vs crop factor debate. I took a photo of the full moon with both bodies and the 100-400 L at 400. The resulting size of the moon in pixels is larger from the XTI. Otherwise it is hard to tell the two appart. But, I was hand holding so it is not a good test. Next time I'll use a tripod and see what happens, but I suspect not much will change.
  16. To the people who argue for the 5D II in the crop vs crop factor debate. I took a photo of the full moon with both bodies and the 100-400 L at 400. The resulting size of the moon in pixels is larger from the XTI. Otherwise it is hard to tell the two appart. But, I was hand holding so it is not a good test. Next time I'll use a tripod and see what happens, but I suspect not much will change.​
    Well, of course you will see this result. You'd know this if you went through my (very long) previous post. When you put the same lens in front of both cameras, you've got one sensor with a 4.67 micron versus another with a 6.41 micron pixel pitch. The former will obviously capture more pixels per degree of view than the latter. That's perfectly fine. But the incoming light is the same in both instances, and the amount of light hitting each sensing unit is proportional to its area, so there is greater noise. Furthermore, the center portion of the lens must have a resolving ability beyond what either sensor is able to capture, or else the limiting factor then becomes the lens.
    Higher pixel density does not automatically translate into better image quality. I keep saying this over and over again when people erroneously think that the APS-C 1.6x crop factor automatically means you just gained 1.6x magnification for free. That increase in apparent resolution comes at the cost of greater noise, sometimes to the extent that the true resolution is in fact lower than the 35mm frame.
    Don't get me wrong--APS-C has distinct advantages. The most obvious ones are less weight and lower price. But another one that cannot be overlooked is that you can get significantly easier lens design because now your image circle needs to only be 62% of the diameter of a corresponding design for 35mm. That means smaller lens elements which are easier to grow defect-free and have better optical characteristics. That's what some people are missing about the EF-S lenses: why buy them if you eventually plan to upgrade to 35mm? The point is with the smaller design you can afford to make fewer compromises. Unfortunately, I don't see a whole lot of offerings in this regard. Instead APS-C owners are buying up super expensive L glass designed for 35mm, and as a result, are paying a high cost up front for light that will never be captured by their sensor.
  17. Alexandre
    You have some very nice cameras but, with the exception of the EF 100 macro, some very ordinary lenses. Most people at photonet would recommend spending most of your photography dollars on lenses rather than camera bodies.
    Depending on your budget I would suggest selling your current lenses, except the EF 100 macro and consider getting the Canon 24-105 f4L and Canon 70-300 f4-5.6 IS USM or 70-200 f4L.
    If you advise what you budget can withstand we can make more suitable suggestions.
    Don't get caught up in the craze that you need the latest camera. Lenses make more difference and the photographer makes much more difference still.
  18. Guys
    I really appreciated your cooperation along the discussion. Just as a matter of curiosity I'll keep my Xsi and probably start collecting some better glasses as long as my budget allows.
    Thanks everyone.

Share This Page