There are some counterfeit Canon lenses out there – best not to buy them from a market stall!
If you are thinking of buying a camera, take a look at the Studio Scene Image Comparison Tool on dpReview where you can compare up to four cameras of your choosing at a range of settings. Very revealing, especially on high ISO noise.
We buy nice predictable digital camera bodies and then put variable analogue lenses onto them. It’s quite possible to get a bad one, or at least one whose tolerances are not compatible to the tolerances on the body. The last time I gathered a set of lenses together (for a Nikon D7100), I formed an instinctive feeling about which were good and which were less good – based largely on price paid. When I actually tested them it turned out that my instincts were completely wrong and the cheapest lens (a plastic-bodied kit mid-range zoom) was optically the best.
Now that I have once again gathered a set of lenses together for my new Olympus OM-D OM10 II I have made sure to test them immediately. For convenience, I did this indoors at a range of apertures, and focal lengths and at distances of 1 metre and 2.5 metres.
I used; ISO 200, a tripod, no image stabilisation, JPG and aperture priority. I was mainly interested in centre sharpness. I looked briefly at edge sharpness but the lenses are relatively cheap so my expectations were lower there. Geometric distortions and aberrations are routinely fixed by the camera/lens profiles these days which are applied to JPGs and raws before you ever see them, and can be tweaked if necessary.
The lenses I tested were;
Olympus 14-42mm EZ mid-range kit ‘pancake’ zoom.
Olympus 40-150mm R kit telephoto zoom.
Panasonic 12-60mm mid-range zoom.
… and also an Olympus 9mm body cap fisheye fun lens.
These are all inexpensive plastic-barrelled lenses at the economy end of the range because I seek light-weight and value. I am, after all, going to process the images to within an inch of their lives so ultimate image quality is less important to me. Furthermore my experience is that these can deliver surprisingly good results.
Having taken a hundred or so test images I rapidly compared them side-by-side in groups of four using the Faststone Compare Selected Images feature.
The results were that all lenses were sharp in the centre under all conditions except at f22 where none were sharp due to the expected diffraction effect. The 14-42 was a bit dodgy at 20mm f4 and wider. The 12-60 was a bit dodgy in the corners at 12mm. The ‘fun’ fisheye was fine in the middle. I didn’t bother to look into the corners!
These results were entirely satisfactory to me and confirmed that I should avoid f22 entirely. I bought the 12-60 because I felt that the 14-42 would be somewhat compromised by its pancake design and this proved to be true, but it’s tiny size will be useful if I avoid the widest settings.
COMMON IMAGE PROCESSING FAULTS
I helped out with the last projected image competition and had the opportunity to look at members’ images up close. They exhibited a number of common faults, not all of which were picked up by the judge, so I thought that it would be useful to discuss them;
1. Image Noise. Most-often caused by a high ISO and/or a small camera sensor, this consists of speckles visible in plain areas of an image. It can also be exaggerated by over-processing (see below). Noise in raw images can be treated very effectively by Adobe Camera Raw Noise Reduction to be found under the Detail Tab. The Colour slider is totally effective at removing the coloured speckles. The Luminance slider needs to be adjusted to reduce the remaining speckles without killing all fine detail in the image.
Bear in mind that JPG images will have already been processed for noise but they may also benefit from the above treatment by opening them from within Photoshop Elements>File>Open in Camera Raw.
2. Bad Cloning. Our brains are very good are spotting repeated patterns where they shouldn’t be, caused by inexpert cloning. Remember to make a lot of small strokes from random source points nearby. Don’t use a totally hard or very soft-edged brush (see below).
3. Burnt-Out Highlights. These can ‘shout’ at us, especially if they are prominent on the main subject, such as a white shirt or feathers. In a JPG these are unrecoverable and can only be rescued by cloning-in detail from elsewhere, or from another image. The Burn tool is ineffective in this situation and will only create an un-natural grey patch. If a raw file is available, there is often considerably more detail to be recovered from highlights (and shadows) but it may be necessary to open a second copy of the image exposed for these areas only and then copy the areas over.
One trick that can work in a few situations eg with a shirt, is to select it and use the Levels Output slider to turn it into a black or grey shirt which can be surprisingly effective. It's not so appropriate for white birds!
4. Over-Sharpening. More serious than under-sharpening, this can create halos around edges, increase contrast and has a distinctive, unnatural look. Always sharpen as your last action (apart from any border) and only at the required resolution ie for a PDI , downsize a copy to the projector resolution and only then sharpen. A low resolution image for projection or screen use will require less sharpening than a high resolution image for a print.
The sharpening process is about improving the perception of clarity, not rescuing blurred images. We see a lot of these and they should go straight in the bin - however wonderful they are otherwise.
5. Wonky Horizons and Verticals. Look critically at your image and make sure that any horizon is horizontal. The Straighten Tool couldn’t be simpler to use, just drag it along the horizon. Verticals need not always be vertical but if you want to adjust them, use Image>Transform>Free Transform. Hold the Ctrl key down and drag a corner. Each corner will affect the others so you will need to go back and forth a couple of times. Turn on the grid to help you visualise; View>Grid (if the grid is too busy, you can reduce the number of lines at Edit>Preferences>Guides and Grids).
6. Over Processing. Photoshop is inherently destructive. If the adjustments are small, this doesn’t matter, but large adjustments will eventually become visible as contrasty, ‘crunchy’ images with visible banding. This can be seen in the histogram which starts to show gaps like a comb. Ideally, if you have to make large adjustments you will have a raw file available where they can be made less damagingly. In particular try to avoid lightening and darkening the same area successively eg with multiple adjustment layers. They don’t co-operate to minimise damage.
7. Borders. A single-pixel border is often appropriate, especially when a dark image has no clear edge onscreen. This should be added after the image has been down-sized to the projector resolution. Select the whole image with Select>All and then Edit>Stroke Selection, choosing 1 pixel, the required colour and a Location of Inside. This is best done on a new empty layer so that the border can be removed or altered easily. It should also be excluded from any final sharpening that is done. For printing, a much wider border of perhaps 20 to 40 pixels may be more appropriate on an image of 4,000 or 5,000 pixels width.
8. Bad Edges. When working with selections or brushes, it’s important to work with a suitably soft edge. A fully-hard edge is never appropriate in a photograph. As discussed in the more detailed meeting notes of April 2015, an image of a precise hard edge is actually about 4 pixels wide, but counter-intuitively, this corresponds to a selection feathering of 1 pixel. Furthermore a fully-soft brush is soft across its entire width which would give a muddled result if used for example, for cloning. Unfortunately Photoshop Elements does not give easy access to brush hardness for the clone tool, but it can be adjusted using the keyboard shortcut Shift+[ or ]. There are five strengths available by this shortcut 100%, 75%, 50%, 25% and 0% hard. Different tasks may require different settings for this strength to get the most realistic result but I would avoid both extremes when cloning.
In passing, we had an interesting exchange about down-sizing images. We all know that we must do this for projection, but some members feel that they should do this before printing. There seems to be a perception that downsizing to an exact number of pixels per inch in Photoshop is better than leaving it to the printer driver. I have never done this and have always sent the original full-sized image, with all layers intact, to the printer. I choose the print size in the printer driver and it can be a completely random downsizing percentage. My Epson printers have always produced pin-sharp images of great clarity. I would be interested if anyone can show printed proof that I'm wrong.
Meeting Notes March 2009 to 2018.
1 post • Page 1 of 1