tom_m2

"Tom, it might be helpful if you mentioned what those "right tools" are now since you've got me curious, and maybe others as well."

 

It's called "flat-field correction" and is widely used in scientific and astronomical image processing. Lots of professional image processing tools support it (check Google and Freshmeat), although you can probably script even Photoshop to do it.

The answer depends on the lens. The 5D sensor has bigger pixels, so it's actually less demanding of the optics than the 20D/XT sensor. OTOH, corner sharpness, vignetting, chromatic aberration, and distortions could get worse, depending on the lens.

 

The real reason why you'll probably be getting new lenses is because the lenses give you a different FOV. OTOH, the smaller APC lenses are really nice, too. I view FF and APC as largely different systems where I occasionally share a lens.

Image quality of the two is about the same. The 20D has more features, the AF sensors are different, and it can take shots at a faster rate. The XT is cheaper, lighter, and smaller, which also makes it nicer for hiking and travel.

 

Why not go with the XT and spend the extra money on another/better lens?

Serguei: "Tom, the only thing that matters for the perspective is the subject distance, not the FOV."

 

Luiz asked about the "perspective of a 17mm lens". While that's a sloppy shorthand, most people understand that "the perspective of a lens" is what you get when you keep the subject at a constant size in the frame for different focal length. So, yes, perspective depends on subject distance, but subject distance depends on focal length.

 

I just answered the original question in the form in which it was asked, and the answer is that a 17mm+APC combo will have "the same perspective" as a 28mm+FF combo, which is the answer Luiz was looking for.

With the right tools (we have some in-house), all 300 images could be corrected automatically. Don't know whether Photoshop is up to it, though.

"The 17mm will always be a 17mm regardless of the FOV, a 28mm on a full frame body has much less distortion."

 

Sorry, but that's just wrong. The only thing that matters for perspective is the angle of view, and a 17mm on APC has the same angle of view as a 28mm on full frame. Sensor size has consequences for DOF, noise, and diffraction.

"22mp must be very lens limited, and 25mp is likely the logical limit of future 35mm DSLR."

 

Keep in mind that because of the Bayer patterns, resolution from those sensors is based, in part, on interpolation. It may well make sense to go up to about 4x the limits of optical resolution (with two separate green channels for better color).

<i>It is interesting to see that everyone is comparing film and digital and no one is discussing color...That is a topic of interest for me, usually working with Kodak E100VS and GX and some Velvia I am curious as to whether I will be happy with the colors of digital or the photoshop plug-ins which duplicate the films vs film itself. Maybe should be a seperate topic but it also deserves to be discussed here. Any thoughts?</i>

<p>

Change the contrast and saturation settings on your camera presets to match your preferences; that's the equivalent of choosing a particular film type. Alternatively, you can take RAW images and make the choice when you convert. Many cameras allow you to have multiple groups of presets and switch between them (the equivalent of interchangeable film backs).

The ISO comparison for megapixels is misleading. High ISO films do have lower resolution, but they also have high noise and poor tonality. That makes high ISO films useless in many applications even when their resolution would be sufficient.

 

Low megapixel cameras have lower resolution, but noise and tonality aren't affected (if anything, they get better as the pixel count drops).

 

4-5 megapixels is sufficient for 8x10 prints (or larger prints at larger viewing distances) of normal photographic subjects and without cropping. That's generally sufficient for most applications. 8 megapixels gives you room for significant cropping, a bit of extra quality, or the occasional big enlargement. But extra megapixels aren't usually worth it if they are bought at the expense of noise or tonality.

If the EOS 3 works for you, you probably can't go wrong with the 1D or 5D; they handle similarly, take the same lenses, and have excellent sensors. I'd personally go with the 5D at this point (have a 1D).

 

To me, the H1 and H2 only make sense if I still wanted to do MF film. You also have the option of large format digital (look for "betterlight" on Google).

Things can always go wrong, and it's better to know as long as you're still there and can retake something. Of course, you wait until you have a bit of quiet time.

 

In any case, they may just have been fiddling with menu settings rather than checking their shots.

"But there is tons of things that I already know or are useless for me, since I'm shooting digital only. I mean I went thru film processing and development in the past. I used to have my own darkroom. The digital course is too Photoshop specific."

 

As a professional photographer, you should know how to produce predictable, reproducible results without a lot of digital manipulation. That's not a question of purity, it's a question of money: fixing bad images after the fact is wasted time (=money). I think film is a good way to learn that; fancy Photoshop manipulations are a crutch that let you avoid learning that.

 

I'm not a wedding photographer, but the 17-85 isn't what I would pick as my primary lens for a social event (nice lens otherwise). But since you already have it, in your situation, I would get a second body (XT or 20D), and carry both a 28/1.8 and a 50/1.4. That gives you low DOF and redundancy. For weddings, if you don't get the shots, your customers will kill you.

I recommend Keeble and Shuchat in Palo Alto as well: great selection, very helpful.

Between the two, I'd personally go with the Canon, simply because it's a little smaller and because I've had better luck with Canon lenses. If you are wondering about the 300-400mm range, ask yourself whether cropping the 300mm shot won't give you similar quality to using 400mm. The 70 vs. 80 should make no difference whatsoever.

 

I've been happy with the 75-300mm IS lens, which I carry for the occasional telephoto shot. You might also want to consider the Canon 70-300 DO IS, which is smaller and optically better, but apparently prone to flare. It costs just a little more than the Sigma.

 

To people recommending a 70-200L, you aren't being helpful. Someone who is trying to decide between those two lenses has already decided he doesn't want a 70-200L. Many people (amateurs and professionals alike) just don't have a use for that kind of lens even if money is no object.

If you are asking whether a monopod gives equivalent stabilization to IS, the answer is that it probably depends on circumstances.

 

The more important difference is that there are many situations where a monopod is impractical or impossible to use, so it's either IS or nothing.

I tried two more Sigma 30/1.4 lenses (at the store) and they weren't much better than the one I had--unacceptable: fuzzy in the corners with strong chromatic aberration. The only thing that was noticeably better was that focus was more consistent.

 

Side-by-side, the Sigma 24/1.8 and Canon 50/1.4 were excellent performers in comparison. The Sigma 24/1.8 seems like a decent choice, except that it is rather big.

"Is the author correct in his central assertion that FF sensors will never work correctly with 35mm lenses because of *allegedly* unavoidable light fall-off and vignetting?"

 

Even with current sensors, light fall-off and vignetting aren't problems on digital cameras because they can be corrected in post-processing. Some cameras, like the Olympus, even do this correction in the body, based on information communicated by the lens.

 

Furthermore, while currently, sensors happen to be regular grids of flat pixels, in the future, you will likely see pixels towards the edges of the sensor tilted. That fixes both the light fall-off problems and other problems with shallow angles of incidence.

 

So, there are no intrinsic problems with full-frame sensors. In fact, Leica may be pursuing this path with their rumored full-frame Leica M digital camera.

 

However, the smaller size of APS sensors is an advantage in itself, and I suspect that both APS and FF will continue side-by-side.

I don't view the 5D as an upgrade, it's just a different kind of camera: in order to get the same quality and range of lenses I got on the 20D/XT, I would have to lug around probably twice the weight and volume in lens gear. For a lot of travel photography, that's not worth it.

 

I may get one eventually in addition to the 20D/XT, together with a couple of fast prime lenses, to cover other kinds of photographic situations. The fact that the FF and APS systems can share some lenses is a bonus.

 

So, I think they are just different cameras for different purposes. APS is always going to have a size and weight advantage, just like 35mm always used to have a size and weight advantage over MF.

I did all the tests on a Rebel XT (I prefer it to the 20D body because of the smaller size, and it's what I would be using a standard prime lens on).

"I had the feeling that while they were promoting it as a digital APS normal, it's much greater size, weight, cost and complexity, suggested that Sigma may have been trying to cover FF as well. But that's a pure guess on my part and it's not clear if that is the case or not."

 

No, it's definitely not a FF lens (white dot). It's not surprising, though, that a lens like this would be complex: digital sensors are very high resolution, and they have additional requirements compared to film about how light strikes their surface.

 

"[The question is academic, since the lens needs to work with AF] Not quite academic. The crucial question is whether the lens is at fault or an inadequate AF! That's two separate issues."

 

The lens I got clearly has both defects. Either one of them would be sufficient to make it unusable.

 

I suspect this is a manufacturing problem, not a design problem. It wouldn't make sense for Sigma to go through all this trouble for a poor lens, and Sigma's pricing isn't usually off this much.

 

The question is whether good copies of the lens ever come out. Maybe the best thing to do for Sigma to fix whatever is wrong an release it as a "Mark II" or as a "28/1.4 DC" so that there is no confusion. The reputation of this lens may be shot even if they start shipping (more) good ones.

"Im curious. Why would you want a fast lens for shooting outdoors in bright sun light? An f/1.4 lens has its purposes, but outdoor landscape at f/8?"

 

Why wouldn't I want that? It's what I used to get with a traditional standard prime lens, like a 50/1.4 or 50/1.8. Every manufacturer has those for FF. Those kinds of lenses have excellent optical performance stopped down, and they are also fast, giving them a lot of versatility. Furthermore, as an additional bonus, something like a 50/1.4 often reaches its peak performance at somewhat larger apertures than slower lenses. It seemed natural to expect that Sigma's 30/1.4 would fulfill the same role for APS bodies; it's disappointing that it doesn't.

"Just how consistent is the AF focusing of the XT or 20D themselves? Scott Eaton and others, have commented that they are having trouble getting consistent AF from the cameras, even with Canon's own lenses."

 

I did a side-by-side comparison. Canon's lenses show some variability on scrutiny, but no serious outliers; it's not great, but I can live with that. The Sigma 30mm, however, was much worse and would frequently yield unusable results.

 

"Were tests made with the Sigma manually focused?"

 

The question is academic, since the lens needs to work with AF; the Rebel XT doesn't even have usable manual focus (neither does the 20D really). But if you are asking whether the lack of sharpness outside the center may be due to focus, it's pretty clear that it isn't: the center is focussed well in the sample. Furthermore, because of the AF difficulties, the Sigma yields many differently focused images, and none of them are sharp in the corners. Finally, focussing at infinity in the center, then recomposing putting the subject in the corner yields the same results.

 

"How good is the Sigma 30F1.4 over the full frame or with film?"

 

It's designed only for APS sensors. That's what makes it an attractive lens, since it should have a lower size, weight, and cost than an equivalent FF lens.

Luke, thanks for sharing your experiences. Could you perhaps take an outdoors landscape photo in bright daylight with it and share it, please? It would be nice to see whether there are better examples of this lens. How consistent do you find the focus to be on yours? I found the focus on mine too inconsistent to trust the lens even for street photography or portraits.

I've already returned it. The question is whether there are any Sigma 30mm lenses out there that are better and I should try again, how the Canon 28mm compares, or whether there are other alternatives (well, other than the Canon 35/1.4L).

A 50mm lens on a FF body used to be good for many situations:

portraits, landscapes, street photography, low-light photography, etc.

When wide open, it would give good results, and when stopped down, it

would generally be one of the best lenses around.

 

I was hoping the Sigma would fulfill the same function on a Rebel XT

or 20D, but the Sigma 30mm I got was not optically good enough for

that purpose; in particular, sharpness outside the center was pretty

poor. Furthermore, focus on the Sigma was inconsistent: a significant

number of shots would be out of focus even in a series of shots taken

from a tripod (all my Canon lenses focus consistently under the same

conditions). Color and exposure also seem to be off, but that at

least one could correct for.

 

So, the question I have is: is the lens I got the best I can expect

from a Sigma 30mm, or did I just get a bad lens?

 

The following image shows the Canon 17-85 vs. the Sigma 30/1.4, both

at f5.6 and 1/250; the crops are from the left margin, the top right

corner, and the center.

 

http://www.photo.net/photo/3768976&size=lg

 

This image is pretty representative; I get the same kind of blurriness

under lots of different conditions.

 

If the Sigma is this inconsistent by design, how does the Canon 28/1.8

compare? I'm really not looking for L-level performance, just

something that is consistent across the frame, focuses predictably,

and generally does a little better than my non-L zoom.