Why are landscape photos "subpar" on Canon EOS 750D with Sigma 17-50/2.8 lens?

Question

I currently own a Canon EOS 750D (Rebel T6i), which I mostly use with a Sigma 17-50/2.8 and a Hoya circular polarizing filter. However, I can't help but feel disappointed when in looking at my pictures. Fine details are often noisy and with landscape images they just don't have that relaxing feeling when you're looking at them. And I feel like there isn't a nice wide angle that costs you a kidney.

Either the 10-18 which in my mind will always be a budget lens, and the 10-22 which is old by today's terms and doesn't really offer that much more than the 10-18. Of course most answers you're going to get is to get a full frame, and I'm open to that but that does impact the weight and size of my gear quite a bit.

Any people here that recognize what I'm feeling? What's your point of view on this? Why is it I'm getting these what I consider to be sub-par results?

Shot with the aforementioned combo at ISO 400, 17mm f/10 1/250sec. This image isn't as bad, but I wouldn't consider it to be printing quality. Check the top ridges of the mountains and the sky. Also, the trees in the distance, although that is a bit pixel peeping.

Also shot with the aforementioned combo at ISO 100, 25mm f/9.0 1/40sec. especially notice the trees how grainy they look. The darker trees on the right side of the image, I consider to be a problem area. I have noticed there is a very slight bit of motion induced softness in the image when checking the mountain ridges, which begs me to question the lens' IS.

No sharpening or noise reduction was done. Only a little bit of tweaking done in Lightroom: +25 on the contrast, -40 on the highlights and +30 on the shadows. I've done more than that in the past and still had better results with it.

I always shoot full sized raw 6000x4000. When limiting myself to the summer image, sharpening was set to 70, radius to 2, details to 25 and the masking to 80. No noise reduction was applied just to be extra sure it wasn't that.

Answer 1

Both of your examples are shot at apertures significantly narrower than your camera's diffraction limited aperture.

The EOS Rebel T6i/750D has a 22.3 x14.9 mm sensor with 6,000 x 4,000 pixels for a resolution of 24 MP and a pixel pitch of 3.72 µm. This figures out to a DLA of f/6.0

The answer linked above states the following:

With a digital sensor the DLA is the aperture at which the size of the circle of confusion becomes larger than the sensor pixels and begins to visibly affect image sharpness at the pixel level. Diffraction at the DLA is barely visible when viewed at 100% (1 pixel = 1 pixel) on a display. As sensor pixel density increases, each pixel gets smaller and the DLA gets wider.

DLA does not mean that narrower apertures should not be used. It is where image sharpness begins to be compromised for increased DOF. Higher resolution sensors generally continue to deliver more detail well beyond the DLA than lower resolution sensors until the "Diffraction Cutoff Frequency" is reached (a much narrower aperture). The progression from sharp to soft is not an abrupt one.

Further down it says:

So what happens once you select an aperture beyond the DLA? Diffraction begins to negatively affect the sharpness at the absolute point of focus. In exchange the narrower aperture increases the depth of field that is in nominal focus. There are techniques that allow you to maximize depth of field using the widest aperture possible. Learning how to calculate hyper-focal distance (or carrying a chart for each focal length you use) allows you to place the point of focus as close to the camera as possible while allowing for everything beyond that point all the way to infinity to remain acceptably in focus. At close distances and wide apertures the depth of field is about equally in front of and behind the point of focus. As the subject distance increases and/or the aperture narrows, a larger and larger percentage of the DOF is behind the point of focus. Here is a link to a DOF calculator you can use to illustrate this.

Instead of shooting at f/9 or f/11, try opening up to f/6.3 or so, and focus on an object at or near the calculated hyperfocal distance using the DoF calculator. If you intend to display the images at greater than 8x10 inches (or pixel peep at greater enlargement ratios than the equivalent of 8x10" - when looking at a 24 MP image at 100% on a 23" HD monitor with a pixel pitch of 96 ppi, that's like looking at a portion of a 60x40 inch print!), then click the show advanced button and enter your intended display size and viewing distance.

Beyond that, practice good, basic technique for shooting landscapes:

• Use the lowest native ISO possible for your camera (usually ISO 100 for recent models).
• Use a sturdy tripod with a rock solid head to allow longer shutter times required by narrower apertures and low ISO.
• Use graduated neutral density filters to avoid blowing out the sky or underexposing the non-sky areas.
• Use mirror lockup and remote (wired, infrared, WiFi/Bluetooth, etc.) shutter release to reduce camera shake when actuating the shutter.
• As much as possible, shoot when there is no or very little wind. Not only does wind move objects within the scene you are capturing, but it also can move the tripod enough to affect image sharpness.
• Wait for the right light for your location and subject. You can do everything above, but if the light is not right, you won't get the photo you want. Weather, time of year/month/day, etc. all have an effect on the light illuminating your scene.

Bear in mind that most of the amazing landscape photos you see at sites such as 500px and flickr have extensive post processing applied that tends to sharpen the results compared to how the image first looks straight out of the camera. In some cases advanced techniques such as focus stacking and highly detailed lens correction are being applied.

Answer 2

Fine details are often noisy...

In your first sample image, the fine details in the tree branches near the skyline are probably hitting the resolution limits of your sensor. The anti-aliasing filter of your camera is likely making them look fuzzier than necessary.

For the second image, I don't see much detail in the shadows of the trees. It might help to increase exposure to better capture shadow detail. However, as small details near the size of pixels, they resemble texture and noise.

Options to consider:

• Use a longer focal length and stitch. – Small details that are near the size of pixels appear more like texture or noise than real detail. A longer lens is needed to increase the size of the details you wish to capture. Getting a sharper, more-expensive wide-angle lens won't solve the underlying problem.

• Adjust your raw-processing settings. – Noise reduction is necessary to counter the effects of using a camera with an anti-aliasing filter and Bayer color filter array, as well as the settings you are using (highlights, shadows, contrast, sharpening, etc), which increase the appearance of noise.

• Try a camera that does not have an anti-aliasing filter. Here is an image that demonstrates the effects of an anti-aliasing filter. The same EF 40/2.8 STM lens was used on two different bodies with the same exposure settings, f/2.8, 1/80, ISO 800.

  

• Use a lower ISO, lower shutter speed, and f/8. – Although diffraction limit calculators may state that f/8 is past the diffraction limit of your camera, practical limits result in many lens-camera combinations being optimally sharp past the nominal diffraction limit. Factors include lens sharpness, sensor resolution, Bayer color filter array, anti-aliasing filter, and gaps between sensels.

  To find the sweet spot for a particular setup, you need to test it. Many lens-camera combinations are sharpest at F5.6-11. Depending on your subject and objectives, it may be reasonable to trade-off some diffraction for increased depth of field or overall sharpness. You can read more about diffraction at Cambridge in Colour: Lens Diffraction and Photography, which has a couple of diffraction calculators.

... they just don't have that relaxing feeling when you're looking at them.

• Your images look fine at web display resolutions, so you're probably doing quite a bit of pixel peeping, which is an inherently stressful activity.

• If you're looking for an immersive experience, some options to consider:

  • Crop.

    

  • Create stereoscopic images.

  • Stitch multiple images into a panorama.

  • Display multiple images side-by-side in a triptych.

  • Use a fish-eye lens.

  • Use an anamorphic lens.

  • Use a dedicated panoramic camera.

  • Get closer to photograph details of interest.

Answer 3

However I can't help but feel disappointed when in looking at my pictures. Fine details are often noisy and with landscape images they just don't have that relaxing feeling when you're looking at them.

Please post an image or two along with (1) either the EXIF data, (2) or the most important settings (aperture, shutter speed, ISO). Please also say whether you use a tripod. Also, are you using the image stabilizer? You can have a number of problems:

• Out of focus
• Too high ISO
• Camera shake

...although the noise suggests it might be too high ISO.

Full frame cameras are known for their better high ISO performance (lower noise than crop at the same ISO). However, typical cheap full frame zooms are f/4, and you already have f/2.8, so that makes up for most of the difference. Are you using the lens at f/2.8?

Of course most answers you're going to get is to get a full frame, and I'm open to that but that does impact the weight and size of my gear quite a bit.

Canon EOS RP paired with 17-40L is actually quite lightweight and small in size. I wouldn't worry about anything except the higher cost. 17-40L is about the cheapest L lens you can find, and EOS RP is about the cheapest full frame mirrorless you can find. Ok, 6D mark II would in some cases be very slightly cheaper, but at the cost of a higher weight and outdated technology (DSLR, not mirrorless).

After getting the 17-40L and mounting it to RP, I compared the image quality to 10-18 mounted on 2000D. The results are:

• Image quality is much better!
• Autofocus actually works in the 17-40L + RP whereas 10-18 + 2000D missed quite often

(Sample image)

Yes, what you're seeing is noise indeed. Please use ISO 100 and increase the exposure time to something like 1/60 sec. Also, you can experiment with various noise reduction algorithms. Some are in-camera settings for noise reduction, others are done in post-processing, where it helps if you shoot RAW.

(Second sample image)

Now I see far less noise, so ISO 100 clearly helps. What you are seeing in the trees may be lack of resolution. If you expect top-notch results from your equipment, you have to have top-notch equipment. I'd say an upgrade to full frame could be worth it for you, but of course it costs a lot.

Don't buy the 10-18 if you expect better resolution! I don't think it will improve the resolution of your image.

But I'm also saying that you may be expecting too much from your equipment. To me, the second picture looks just okay. What you are doing here is sometimes referred to as "pixel peeping".

What lens to get, then?

Let's compare the Sigma 17-50 and Canon 17-40L on crop cameras: comparison -- as you can see, just by updating your lens to a high-quality lens won't obtain you any better picture quality.

What if you update to a very high-resolution full frame camera and Canon 17-40L: comparison -- now you can observe that the 17-40L + 5Ds R combo has an extraordinarily high image quality. Note it has more megapixels, too, and you can actually make use of these megapixels with the lens.

So I'd say the question is not "what lens to get" but rather "which full frame camera and what lens to get".