I have to rethink my camera decision (7D vs 5d Mark II)

[dtaylor]

The camera itself is fantastic. I've had issues with shooting in the shade (forest) and high isos. I shot in the grocery store at 3200 iso and I swear it looked like some kid was taking crayons and stabbing it all over (that's how grainy and pixelated it looked at the iso). I'm not sure if it's the lens or the camera or if I'm doing something wrong.

I don't see the grocery store photo in your Flickr stream. Seeing an ISO 3200 example which you think is bad would help because we can see if it's underexposed.

Typically when people complain about high ISO images from 15-18 MP crop bodies they are either a) underexposing the image, or b) spending way too much time pixel peeping. Viewing a 7D image at 100% on a typical monitor is like viewing a 60" print at arm's length or less. It is not necessarily a good indicator of what a print will look like.

I've produced 8x10 and 11x14 prints from ISO 3200 7D shots that have tighter and more pleasing noise than the grain found in ISO 160 35mm portrait films. And I didn't do anything special like layering the image to allow greater NR on low detail areas. I just simply applied a touch of NR, did my other normal processing, and printed. Spending too much time pixel peeping can lead to processing mistakes such as heavy handed NR which results in a plastic look when printed, so be careful with the pixel peeping.

Having said that, the 7D has very little underexposure tolerance at high ISO. Make sure you nail your exposures when shooting in low light.

[Kernuak]

I've been reading for a while (mainly to keep up with what might happen in the near future), but I felt inclined to make some comments on noise with the 7D. The first comment, is that even at ISO3200, the level of noise is much less than the grain that would have been seen on old ASA 400 film. It is also much easier to deal with than previous Canon cameras (such as the 40D), as you don't get the banding and it is much more even. That said, I tend not to go higher than ISO 1600. If I need to go higher (or if I don't need the frame rate), then I switch to the 5d MkII. However, noise is variable. Some scenes, ISO 1600 is perfectly fine, while in others it results in disturbing levels of noise. Essentially, it depends on how many shadow areas there are and whether those shadow areas are part of the subject. If you have a textured OOF background, then it tends to hide the noise somewhat too, without losing much of the fine detail, while a smooth background will make it more evident. In fact, it was only after using the 7D for some time, that I realised that part of the problem is the resolution, almost as if it is finding some sort of texture that wouldn't have been seen with lower resolution cameras. Often, the out of focus areas have a speckled appearance that isn't like noise seen in other cameras I've used, probably because it is more luminance noise than chroma, which is easily removed with a low setting in Lightroom 3.
The biggest disadvantage I have found with the 7D, is softening at narrow apertures, due to the diffraction limited affects, resulting from the small pixel size. In fact, that was the reason I ended up getting the 5D MkII for my landscape work. Of course, then I ended up using that camera for macros too, as I found it easier to focus manually, plus the images were clearner overall.
As for the 17-40. Yes, it does distort a lot, but that isn't necessarily a bad thing. Many wide-angle zooms (which is what is was designed as after all, even if it isn't that much of a wide-angle on a crop body) suffer to a greater or lesser degree from this. The trick is to make creative use of that distortion and pick and choose what subjects you use it for and how you angle the camera. If you keep the sensor parallel to the subject, then the distortion is less noticeable for many subjects and is also more easily corrected in post. The distortion is really handy for accentuating converging verticals with tall symmetrical buildings and for giving the appearance of curvature of the earth in certain landscapes and seascapes. It's not everyone's cup of tea, but it does give you something different to what everyone else is doing.
Essentially, photography is always a compromise, whatever tools you use, you simply have to explore the weaknesses and strengths of each piece of equipment and work around them. For example, as Neuroanatamist has said, the 5D is weak at focus tracking, but good at focusing in low light and producing low noise (relative to virtually any camera except the D3s and possibly the original D3). For that reason, I usually use the 7D for wildlife, but when I was photographing roding woodcocks after sunset, I used the 5D in single shot mode for optimal focusing and as low noise as possible (at least for the equipment available to me). Working around the limitations (and spending more nights attempting it), I was able to get similar quality shots to a pro wildlife photographer on his D3s, I just had to apply a bit more noise reduction at ISO3200 than he would have needed.

[Kernuak]

Typically when people complain about high ISO images from 15-18 MP crop bodies they are either a) underexposing the image, or b) spending way too much time pixel peeping. Viewing a 7D image at 100% on a typical monitor is like viewing a 60" print at arm's length or less. It is not necessarily a good indicator of what a print will look like.

Having said that, the 7D has very little underexposure tolerance at high ISO. Make sure you nail your exposures when shooting in low light.

That's actually a point I meant to make in my reply, but got carried away. For any camera near it's limits, you really need to nail the exposure to get the least amount of noise possible.

[dtaylor]

The biggest disadvantage I have found with the 7D, is softening at narrow apertures, due to the diffraction limited affects, resulting from the small pixel size. In fact, that was the reason I ended up getting the 5D MkII for my landscape work.

You don't need and shouldn't be using such narrow apertures on crop for landscape shots. Remember that for an identical FoV crop has 1.6x more DoF, so you don't need to stop down as much. In fact, for a given FoV and DoF diffraction does not impact any format more than the others. This is true for everything from 4/3rds to LF.

On any format it is beneficial to know and use hyperfocal distances. A typical 35mm shooter might choose f/16 for a landscape. A crop shooter might use f/11 in that instance. But a hyperfocal crop shooter might end up with f/8 or even f/5.6. He would get more and sharper fine detail than either of the other two yet still have sufficient DoF.

[lady]

I thought you said you tested the 17-40 and already had the lens in question and found it superior than the 17-55 in "every way"?  F4, for most indoor environments was never claimed to be the best of both worlds... In most indoor environments, F2.8 is ONLY 1 stop more light and may not even be enough.  You can pick up cheap primes of 50mm 1.4, 35mm 1.4, etc (Even those would have incredibly shallow DOF and need a tripod to shoot with a lower F stop to get more DOF)  Low light situations is a constant battle... you can always get a faster lens but faster lenses have smaller DOF and that can hurt shots as well.  Then it's either raise ISO or get more light, hence strobes or flash... those would be more suited for indoors.  That being said, it appears that you are shooting handhold, am I correct?  For those macro shots which tend to lose light as is due to lighting environments, you almost need to use a tripod.  Using a tripod will allow you to knock down the ISO, longer more stable shots, sharper images, etc...  You can also get a ring light that goes around the lens to add more light in those macro settings.   In the forest you can get a neutral density filter, and lengthen the exposure, lower ISO and get a better shot overall.  I dont know if the grocery store will allow a tripod in their store unless you give them a few photos, but that's the route i'd go in.  Also bring a gray card with you to run a quick white balance in commercial buildings... florescent lights are nasty to work with.  In this Ultrawide area of lenses, distortion will be a fact of life, but it will just vary depending on your lens and camera.  The only way to really avoid distortion is TS lenses.

It was better in every way and I had no noticed any lens distortion during my testing. It wasn't until I was in a more linear environment (my house) that I noticed it. However, I paid a lot less for the 17-40mm than the I would have for the 17-55mm. I paid $600 for it which is about half the price. It's ridiculous to say I'm foolish for not paying twice the price that did not compete. The aperture is not something I tested with the lens when I compared the two, I was in a controlled lighting environment. On top of that, I already own a f/1.4 50mm which I mentioned several pages ago. The 2.8 is enough for indoors for me. No need to act condescending and rude about it, dude.

As for your other suggestions, they are very good and I'll consider trying them. I wish I could afford TS lenses, but that's for later on.

The camera itself is fantastic. I've had issues with shooting in the shade (forest) and high isos. I shot in the grocery store at 3200 iso and I swear it looked like some kid was taking crayons and stabbing it all over (that's how grainy and pixelated it looked at the iso). I'm not sure if it's the lens or the camera or if I'm doing something wrong.

I don't see the grocery store photo in your Flickr stream. Seeing an ISO 3200 example which you think is bad would help because we can see if it's underexposed.

Typically when people complain about high ISO images from 15-18 MP crop bodies they are either a) underexposing the image, or b) spending way too much time pixel peeping. Viewing a 7D image at 100% on a typical monitor is like viewing a 60" print at arm's length or less. It is not necessarily a good indicator of what a print will look like.

I've produced 8x10 and 11x14 prints from ISO 3200 7D shots that have tighter and more pleasing noise than the grain found in ISO 160 35mm portrait films. And I didn't do anything special like layering the image to allow greater NR on low detail areas. I just simply applied a touch of NR, did my other normal processing, and printed. Spending too much time pixel peeping can lead to processing mistakes such as heavy handed NR which results in a plastic look when printed, so be careful with the pixel peeping.

Having said that, the 7D has very little underexposure tolerance at high ISO. Make sure you nail your exposures when shooting in low light.

I had the ISO set to auto. That usually works well out doors, but indoors it can be a bit temperamental. I may have needed a tripod.

However, the results I got were similar to the review test (here) 12800 ISO. It was pretty bad. It looked like this, though honestly I'd complain if it was equal to the 3200 too. I do think I didn't have the shutter slow enough. It just shocked me to see that much noise.

[Kernuak]

The biggest disadvantage I have found with the 7D, is softening at narrow apertures, due to the diffraction limited affects, resulting from the small pixel size. In fact, that was the reason I ended up getting the 5D MkII for my landscape work.

You don't need and shouldn't be using such narrow apertures on crop for landscape shots. Remember that for an identical FoV crop has 1.6x more DoF, so you don't need to stop down as much. In fact, for a given FoV and DoF diffraction does not impact any format more than the others. This is true for everything from 4/3rds to LF.

On any format it is beneficial to know and use hyperfocal distances. A typical 35mm shooter might choose f/16 for a landscape. A crop shooter might use f/11 in that instance. But a hyperfocal crop shooter might end up with f/8 or even f/5.6. He would get more and sharper fine detail than either of the other two yet still have sufficient DoF.

It's a very blanket statement to say that you shouldn't use such narrow apertures on crop cameras, it depends on the actual landscape, focal length and to a lesser degree the actual lens used. Also, hyperfocal distance isn't always appropriate to use. While I started paying more attention to hyperfocal distance when I got the 7D (and still do), if there is a cliff smack bang in front the lens about half a mile away, it's pretty pointless trying to get infinity focus at the expense of nearby detail. I also do some micro-landscapes, where hyperfocal focusing is completely useless and DoF is even more critical, plus there is generally more fine detail which is affected by softness. Also, f/11 on a 50mm lens wasn't always narrow enough to get sufficient DoF, even using hyperfocal focusing. Additionally, there are few modern lenses that have detailed enough distance scales to accurately judge the focal distance selected. You could argue that I could have chosen different subjects to make better use of the abilities of the 7D and I did for a while to experiment and try new things, but for rocky seascapes, the wide open landscape is the best option, which requires a lot of DoF, with a sharp detailed foreground.
Since getting the 5D, I find that I tend to frame differently, tending to go for a wider FoV, so essentially using the same focal length as on the 7D. As I am restricted to the same standing points (due to the high tidal range and inability to swim with gear on), the DoF is actually greater on the 5D than the 7D. Also, I still have the same options as with the 7D (albeit at a longer focal length), so I have greater flexibility.
I did once see some quoted DLA figures for different cameras and for the 7D I believe it was wider than f/5.6, so it doesn't take long to start seeing the effects of diffraction when shooting landscapes, if that is the case. While it isn't the whole story and acceptable sharpness is quite subjective (and less of an issue for landscapes to some degree), it is something to take into account. I'm not one to get hung up on so called scientific data (I have a scientific background), but the results I was getting spoke for themselves.

[awinphoto]

It was better in every way and I had no noticed any lens distortion during my testing. It wasn't until I was in a more linear environment (my house) that I noticed it. However, I paid a lot less for the 17-40mm than the I would have for the 17-55mm. I paid $600 for it which is about half the price. It's ridiculous to say I'm foolish for not paying twice the price that did not compete. The aperture is not something I tested with the lens when I compared the two, I was in a controlled lighting environment. On top of that, I already own a f/1.4 50mm which I mentioned several pages ago. The 2.8 is enough for indoors for me. No need to act condescending and rude about it, dude.

As for your other suggestions, they are very good and I'll consider trying them. I wish I could afford TS lenses, but that's for later on.

I wasn't intending on being condescending and or rude, I was just seeking clarification about how you said you had it, tested it, and loved it and then "just got it" the other and are having buyers remorse.  Do keep in mind regarding distortion... this lens is designed as a 17-40 on a full frame camera... 17mm on any camera will have some level of distortion... Same as the 17-55 in some regards on the 17 end, as well as lets say the tokina lens and the 16-35 I or II... The full frame will show the distortion more than crops, but it's still there.  Shooting parallel will help diminish the distortion however if you are wanting no distortion, post production will be a must.  Some lenses to keep in consideration that will have fast(er) speeds and keep distortion in check (20mm 2.8, 24mm 1.4, 24mm 2.8, 35mm 1.4, they are all within the range of the 17-40, should keep distortion better in check, and 2.8 or faster) It's a great outdoors/travel/walk-around lens, but indoors, you will need to have a fast prime if you really want to shoot handhold.  I do architecture (one of my specialties) and real estate photos... I shoot with the 10-20 and 17-40... But then again I shoot low apertures, tripod always, and they are static.  Sometimes if I cant shoot tripod I still use low ISO but throw in off camera flash or strobe.  Light will always be an issue indoors so either use a faster lens or use a flash (ideally either strobe or off camera flash).  The 7D has a great commander feature if you can pick up some 580's or 430's... scatter them around the scene out of view from the camera and you never have to worry about lack of light. 

I never said it was foolish not getting the 17-55... I said I liked the 17-40 personally, however you need to really and fully know it's capabilities and limitations to get the most out of it.  By going in a shoot fully educated on your gears limits will allow you to forward think so you know how to counter the limitations and push the limits to get awesome shots.  Also remember the 17-40 can be used by the 7d and 5d whereas the 17-55 can only be used on crop cameras.  Regarding your expectations of high ISO, check out the link i posted earlier... it'll give you a good idea of what this camera can do not only against itself but competitors. 

[dtaylor]

However, the results I got were similar to the review test (here) 12800 ISO. It was pretty bad.

Then you underexposed by at least 2 stops giving you the same noise as a 12800 shot. I consider 3200 quite usable on the 7D. 6400 might make a decent 8x10 in a pinch. But 12800 is not usable IMHO.

It looked like this, though honestly I'd complain if it was equal to the 3200 too.

You wouldn't complain at all if someone took the full 3200 test image and printed an 8x10. Pixel peeping vs. prints. On top of that it's possible to get better noise and fine detail out of ISO 3200 using RAW. In their test they choose an area of relatively fine detail yet used JPEG and left in camera NR at its default setting. Naturally this smeared detail. This would be handled far better by Adobe Camera RAW than the camera's JPEG engine.

[awinphoto]

I do think I didn't have the shutter slow enough. It just shocked me to see that much noise.

Yes, high ISO noise is not too pleasant... however the test does show characteristics of the noise vs other cameras... 5d, being a bigger sensor, only besting the 7D by 1 stop.  I understand we want to limit noise but then again we also have to understand how far we've come... "back in the day" with film, ISO 1000 was pretty much unprintable bigger than 5x7 and even with that, it was a grainy mess... (Keep in mind that was with over 40-50 year technology and development of film, that was the best they can do for high ISO film)  It was a nice special effect but that was all it was, a special effect.  Now the same film probably has ISO 6400 characteristics on the 7D.  When looking at that perspective, I find shots very usable up to ISO 2000.. maybe 3200 if I like a shot and want to take the time to do the appropriate post.  As others said, High ISO requires perfect exposure.  To be honest, as your school will tell you when you go there, you NEED to use a separate light meter to meter the scene appropriately.  Also Auto ISO, it is a pain to get that sucker to over expose when you want it to, it will tend to try to underexpose it, so by having that external meter, you can set it to full manual everything and ensure proper exposure. 

[dtaylor]

It's a very blanket statement to say that you shouldn't use such narrow apertures on crop cameras, it depends on the actual landscape, focal length and to a lesser degree the actual lens used.

You said that diffraction hurt your 7D images more than your 5D2 images. But diffraction is the same for a given FoV and DoF regardless of format. If you see more diffraction in your 7D shots then you're stopping down too far.

Also, hyperfocal distance isn't always appropriate to use. While I started paying more attention to hyperfocal distance when I got the 7D (and still do), if there is a cliff smack bang in front the lens about half a mile away, it's pretty pointless trying to get infinity focus at the expense of nearby detail.

I'm not sure what you're getting at here. In this specific scenario a half mile away is effectively infinity for WA glass. But in the general scenario that your near to far range isn't very large, you don't need to stop down as much any way. Unless of course we're talking about macro ranges. But the same rule still applies.

I also do some micro-landscapes, where hyperfocal focusing is completely useless and DoF is even more critical, plus there is generally more fine detail which is affected by softness. Also, f/11 on a 50mm lens wasn't always narrow enough to get sufficient DoF, even using hyperfocal focusing.

That's all fine and good, but none of it violates what I've stated above.

Additionally, there are few modern lenses that have detailed enough distance scales to accurately judge the focal distance selected.

LiveView is a very useful tool for this.

You could argue that I could have chosen different subjects to make better use of the abilities of the 7D and I did for a while to experiment and try new things, but for rocky seascapes, the wide open landscape is the best option, which requires a lot of DoF, with a sharp detailed foreground.

Rocky seascapes are my specialty. When using my 7D I rarely stop down more than f/8. f/11 is the limit. This is with my camera on the ground and desired DoF ranging from the sand 1 ft in front of the lens to the horizon.

Since getting the 5D, I find that I tend to frame differently, tending to go for a wider FoV, so essentially using the same focal length as on the 7D. As I am restricted to the same standing points (due to the high tidal range and inability to swim with gear on), the DoF is actually greater on the 5D than the 7D.

But this wouldn't be true if you had a wider lens on the 7D.

I did once see some quoted DLA figures for different cameras and for the 7D I believe it was wider than f/5.6, so it doesn't take long to start seeing the effects of diffraction when shooting landscapes, if that is the case.

Diffraction limits extinction resolution at f/8 on the 7D, and f/12 on the 5D mkII. Note that for the same FoV, DoF on the 7D at f/8 roughly equals DoF on the 5D2 at f/12.

In practice, for critically reviewed 24" landscape prints, I obtain acceptable fine detail and sharpness at f/11 on the 7D. I can see the diffraction effects while pixel peeping the RAW file, but they do not impact the final print. f/16 is more narrow than I care to go. On FF I find f/16 usable and f/22 unusable. But then on FF I need to stop down that extra stop to match the 7D's DoF.

[dtaylor]

As others said, High ISO requires perfect exposure.  To be honest, as your school will tell you when you go there, you NEED to use a separate light meter to meter the scene appropriately.

The built in spot meter is pretty effective once you learn how to use it.

Also Auto ISO, it is a pain to get that sucker to over expose when you want it to, it will tend to try to underexpose it, so by having that external meter, you can set it to full manual everything and ensure proper exposure.

Auto ISO is worthless. Should have mentioned that earlier.

[awinphoto]

The built in spot meter is pretty effective once you learn how to use it.

Also Auto ISO, it is a pain to get that sucker to over expose when you want it to, it will tend to try to underexpose it, so by having that external meter, you can set it to full manual everything and ensure proper exposure.

Auto ISO is worthless. Should have mentioned that earlier.

The spot meter is ok for people who understand what the reflective spot meter is doing.... For someone who doesn't can easily under expose or over expose depending on the situation... The spot meter is still not as accurate in overall readings and effectiveness than an incident meter.  Reflective meters measures 18% gray and reflect only what it sees reflected whereas incident meters read the actual quantity/quality of light entering a scene, hence more accurate overall.  Sekonic just had a really interesting webinar regarding metering systems with cameras spot meters and sekonics incident meters. 

[neuroanatomist]

Yes, high ISO noise is not too pleasant... however the test does show characteristics of the noise vs other cameras... 5d, being a bigger sensor, only besting the 7D by 1 stop.

One comment about the ISO noise tests you commonly see - they tend to under-represent the noise you would see in real-world situations.  Total light gathered is the primary determinant of noise, in an inverse proportion (more total light, less noise, which is why a FF sensor has less noise).  However, the duration of the exposure (integration time) also matters - shorter exposure, less noise.  You can see that most easily with very long exposures, where dark frame subtraction is used to cancel out some of the noise. 

Typical ISO noise tests are performed with constant lighting, so when you compare noise at ISO 100 with noise at ISO 3200, if the ISO 100 shot was taken with a 1/60 s exposure, the ISO 3200 shot was taken with 1/2000 s exposure - that short exposure somewhat mitigates the increased noise at higher ISOs.  Importantly, that's not representative of the normal situations when high ISO is used, i.e. we don't shoot at ISO 1600 if there's plenty of light, but rather we bump up the ISO when light is limiting, and that situation is not duplicated in most testing scenarios.

[dtaylor]

However, the duration of the exposure (integration time) also matters - shorter exposure, less noise.

For all intents and purposes this only matters once you get into exposures that are too long to handhold. For low ISOs you're talking about multiple seconds. For high ISOs you're talking about a half second or more. There's no real difference between, say, 1/500s and 1/30s at ISO 3200.

Imaging Resource ISO tests illustrate this and also address your underlying complaint. Their fastest exposure in the 7D series, for example, is 1/83. That's for the 12800 11 lux test.

[awinphoto]

shorter exposure, less noise.  You can see that most easily with very long exposures, where dark frame subtraction is used to cancel out some of the noise. 

From what I understand, back with the early digitals, yes, long exposures could lead to noise, as well as film... however, for example, this last 4th of july, I shot with my 7D in a scenic area overlooking a casino that launches fireworks every year, with my 70-200 on a locked down tripod... ISO 100 with 8-30 second exposures... No noise... I would have to wait til this evening to upload 100% clips of the shot, they are on another computer, but I can upload an overall (combined shots with a base shot plus a handful of fireworks blended in... The base shot was 10 seconds of the building.)  Long exposure noise is pretty much a moot point anymore...