Stephen Burch's Birding & Dragonfly Website

Home | Trip Reports | UK index | Oxon pics | UK pics | Dragonflies | Other Nature | Links

Computer processing of digital camera pictures

This page gives information about how I process my digital camera pictures, with the emphasis now on those from my Canon EOS 350D DSLR. For further info about my photographic equipment, click here.

DLSR: Raw format

Until March 2007, I just used jpeg format for all my pics, including those from the DSLR. However, I then took some comparison shots using the jpeg + raw mode, and a careful inspection showed that the raw pics were slightly better both in terms of sharpness (on very fine detail at high magnification) and also colour. These differences are relatively minor, and difficult to see unless you look hard!

Nevertheless, since then I have become a convert to raw, despite the extra stages of processing needed. With raw, you can also vary the white balance after taking the shot, which can be useful in some cases.

The main down-side of raw format is the extra space the files need - about 3x that for a fine jpeg. So 1 Gb memory cards only hold a bit over 100 raw images, compared with >300 jpegs. A 2 Gb card which holds over 200 raw images is then better for my usual half day local trips, but for anything longer (like a holiday), some other backup storage will be needed - such as the portable hard drive described in my equipment section.

Overall 'work flow'

So, for those interested, my 'workflow' is as follows:

Importing raw images into Photoshop Elements 3.0

To read raw format images, Photoshop Elements needs the Camera Raw plug-in, which can be downloaded from the Adobe website. However, the latest version is no longer compatible with the older versions of PhotoShop Elements (the latest is version 6.0). Typical Adobe tactics to make people pay up for expensive upgrades! Fortunately, Camera Raw 3.6 which I downloaded some time ago is compatible with both my camera and PSE 3.0.

But even something seemingly as simple as reading a raw file is by no means straightfoward, and there are several options/parameters available - the meanings of some of them I do not fully understand yet! The values I normally use are shown below:

Some comments on the above:

• I often try changing the White Balance Temperature. Increasing the Temperature can sometimes give a warmer feel to the pic (mainly the greens).
  

• I think changing the Tint can sometimes be useful to get sky colours looking more natural (less red).
  

• For raw images that have some slightly saturated areas, reducing the Brightness parameter can result in an imported image completely free of saturation. This only works for marginal cases, though. If the raw image is too saturated, you have had it!
  

• I set all the Detail parameters to zero, on the grounds that sharpening and noise reduction is best done later (see below). Note that recently when I forgot to reset Sharpness to zero, I was concerned to find all my images were noisier than before. It took me a while to realise why! I even suspected a camera fault!
  

• I currently stick with 8-bit images, as PSE 3 only has very limited support for 16-bit files, which are huge!

NEAT IMAGE for noise suppression

I first came across mention of the NEAT IMAGE software for noise suppression on Richard Bedford's amazing website, but I now find it is fairly widely known amongst keen bird photographers. Having used this software for sometime now, I find it an essential part of my processing sequence (see above). This software uses very advanced image processing routines to suppress noise, crucially with little effect on the actual wanted image detail. This can allow use of higher ISO settings to get increased shutter speed, if needed.

There is a free download version which is pretty useful, but is restricted in the options available for saving files. There are various other options which require payment, with better facilities in terms of input/output file formats, but the 'Home edition", which I've purchased isn't very expensive and allows saving the output in the loss-less tiff format - worthwhile if the initial images are in RAW format (see above).

My approach is to apply the noise reduction first, then move on to the 'normal' PhotoShop operations, as described below. Recently, I went through the process of generating my own noise profiles for ISO 400, which was asomewhat involved process - and probably not significantly better than using the ones downloadable from the NeatImage website. But at least I now have a noise profile for every shutter speed, not just a selection.

With my EOS350D on ISO400 or less, I find NeatImage works very well and I find it best used at about the 50-55% level (just below the default 60%). Random noise is then almost completely removed from the processed image, without any appreciable loss of sharpness.

For ISO 800, it is still effective, and I have recently found it worth applying it at about the 70% level. However, it doesn't have much effect on the "chroma noise" (small coloured spots), which appear at ISO800 and above, most apparent in darker image areas. These spots can of course be removed manually using the PhotoShop clone stamp tool (see below), but this can be somewhat labourious!

Hence, keeping to ISO400 and below seems the best policy, but if light levels aren't great, then ISO800 will still produce pretty good results. Best to avoid ISO1600, though, unless absolutely essential.

The NEAT IMAGE website can be reached by clicking here.

Incidently, NeatImage is also highly effective on digiscope jpeg pics as well.

PhotoShop Elements (2.0 & 3.0) : Tips for digiscopers & DLSR

This section contains some tips for Photoshop Elements Versions 2.0 & 3.0, initially written based on my experiences with digiscope pics, but I now find much of this is still relevant for DSLR pics, and I have added a few notes about any notable differences. I've now upgraded to version 3.0, as it can cope with raw format DSLR images (see above), but I find most of the other options below are the same as in version 2.0.

PhotoShop Elements (PSE) can be a tremendously powerful tool for improving digiscope & DSLR pictures. Here is a list of the features and functions I find useful. Generally, having opened the original image, I go through the following sequence:

1. Crop to size
This is the first step. As I am usually aiming eventually for a standard size for commercial prints, or putting on my website, I stick to a standard 1.5:1 (or 6:4) aspect ratio. In PSE, you can easily achieve this by entering 15 cm in the width and 10 cm in the height for landscape format, or vice-versa for portrait format.

This results in a rectangular box which can be dragged out to the right size, and centred over the subject. If the picture is tilted, then the box can be rotated as well to line things up.

Just double click inside the box to crop, when you've got the right area of the picture.

2. Enhance/Adjust Lighting
It is important to ensure when taking the initial picture that none of the areas (pixels) in it are too bright (saturated). To make sure of this, I usually use a manual adjustment of -0.7 or -0.3 to the camera's automatic metering value. This often makes the raw picture a little dark, but this doesn't matter much, as it can easily be rectified using PSE! Note that for DSLR flight shots, the reverse exposure compensation is usually useful (i.e. +0.3 to +0.7, to avoid the subject being too dark against the bright background.)

Levels

The main option I usually use for contrast enhancement is found in Enhance | Adjust Brightness/contrast | Levels (v2.0) or Enhance | Adjust Lighting | Levels (V3.0). This brings up a dialog box as shown here, which displays a histogram of the levels in the input image. You can pickup and move the sliders below the histogram. The left slider should be moved to near the left (lower) edge of the histogram (the point at which the histogram drops to near zero) - as shown here. Similarly, the right hand slider should be moved to close to the upper end of the histogram - again as shown here. For some pictures, especially those of bright white birds, or silhouttes, it is worth playing with the central slider - moving it to the left introduces a non-linear transformation which lightens the darker areas of the picture without saturating the brightest parts. Also try the shadows/highlights option below, if you have V3.0 or later.

Shadows/Highlights

Shadows/Highlights is another useful tool, which is a valuable addition to version 3 of PSE (not available in PSE 2.0). I use Shadows/Highlights most of the time for DSLR pics, in preference to levels. Lighten Shadows does what is says - makes the darkest parts brighter (i.e. the opposite of Darken highlights), but both can be used to good effect together for example on a dark bird on a light background Darken highlights allows you to make the brightest parts of the image darker - something which is impossible with Levels above. This is particularly useful for slightly over-exposed parts of the image (e.g. sky, or a bright background), provided they are not saturated.

Example of Shadows/Highlights

Adjust Hue/Saturation
Sometimes, the Adjust Hue/Saturation option in PSE can be useful to add colour to an otherwise 'washed' out image, maybe taken in very overcast conditions. However, care is needed with this, and increasing the saturation too much can produce very garish, un-natural looking results.

3. Filtering - Unsharp Masking
Virtually every digiscope shot I take is not quite as sharp as it might be! If it is very fuzzy there is little or nothing that can be done to improve things, but if it's just a little unsharp, then there can be real benefits by using the unsharp masking (USM) feature in PSE. This is much better and more powerful than the simpler "sharpen" options, but needs more experience to use effectively.

To find the unsharp mask, goto Filter | Sharpen | Unsharp Mask. Prior to entering this option, it can be useful to zoom in on a key area of the picture, such as the bird's face/eye. Then the unsharp masking dialog box shows a little box containing the key area.

Note that the settings needed for DSLR pics are generally different from the lower quality digiscope shots, as indicated below.

It also shows the values last used for the three parameters Amount, Radius and Threshold.

The Amount is essentially a measure of how much filtering is being applied. For digiscope pics, values in the range 200 - 400 are worth trying - the higher values for the fuzzier pictures. For any half reasonable DLSR pics, these values are way too high, and instead something in the range 125-175 seems better.

The Radius is the distance over which the filter acts. For digiscope pics, I find values between about 1.5 and 2.5 best. For pictures that are already quite sharp, use values at the lower end of this range. For the more blurred shots, use the higher values. For DSLR pics, which are generally sharper, values in the range 1.2-1.4 are more typical.

The Threshold specifies, for each pixel, the minimum contrast difference for applying the filter. If the pixel's brightness is less than the threshold different from its neighbours, the filtering is not applied. This can be used to prevent the filter being applied to largely uniform areas of the picture (e.g. sky etc). I generally use a value of about 5-10, which seems to work well and prevents uniform areas being filled with "noise" after filtering. Strangely, this value differs a lot from that given on Andy Bright's website, where he recommends 0.1-0.3! It is possible that these are for the full Photoshop not Elements, or that I am missing something!

Often I use the "undo" button and "redo" buttons to check the effect obtained, and adjust the parameters accordingly. With a bit of experience, its quite quick to get reasonable results. But note that for some shots it seems to work better on than others.

As stated above, often DSLR pics can also benefit from unsharp masking, but generally the Amount needed is somewhat less (typically in the range 120-175) and the Radius should be less (say 1.2 - 1.4).

Unsharp masking should always be done last, before saving the image to file.

4. Save as
Lastly, I use the save as feature, not just save, to make sure I can always go back to the original image from the camera, if I decide later I'm not happy with the enhancements made. I generally save the images in jpeg format again, but with the maximum "quality" of 12. For DSLR raw images, I save as loss-less tiff as well, to avoid any loss of image quality.

For website pics, I also use Image | Resize to reduce the size to a constant value (e.g. width of 500 or 700) and a lower quality of 7-8.

(Extreme!) example

'Before' image of the Slimbride Little Crake- straight from the camera	'After' image - following all the PSE options described above. Still not a brilliant picture, but a big improvement on the raw image (which was badly affected by part of the hide window getting in the way of the 'scope lens!)

More advanced options

Clone Stamp
This can be useful for removing unwanted features from a picture, and if used with care can be very effective.

Having selected the clone stamp tool, the basic technique is to first select the area you wish to copy by holding down the Alt key and simultaneously clicking the left mouse over the required spot. This area should be close to the feature you want to remove.

Then release the Alt key and go to the feature you want to remove, and hold down the left mouse button while moving the cursor across the feature. It will disappear in front of your eyes!

You will need to experiment with the size of the brush - to match it roughly to the size of the feature to be removed. Hardness seems best left on default.

It is relatively easy to remove isolated bits of branches, twigs etc from the background using this technique. A more challenging example is shown below!

'Before' image of the same Slimbride Little Crake- straight from the camera - note the wire fence in front of the bird!	'After' image - following all the PSE options described above, including clone stamp to remove the wires - even across the bird!

Removing colour fringing
Even with good quality optics, pictures can sometimes show evidence of colour fringing around edges where there is a very large change in brightness (e.g. the edges of a bird taken against a very bright background). This unwanted fringing can be removed as follows:

1. First use the lasso tool to select just the region of the picture affected by the fringing. This is important because it prevents the colour being removed from the whole picture - which may well correctly contain areas of this colour elsewhere. (I have not seen this step mentioned in other accounts of this process).

2. Then goto Enhance | Adjust Colour | Adjust Hue/Saturation. This brings up the Hue/Saturation dialog box illiustrated opposite.

3. Decide which colour you wish to suppress. I usually find it is blue, but I have seen others mention magenta. Either way, just select this colour from the drop-down list called Edit.

4. A useful even cleverer trick is to first select a colour from the Edit list. Then click, using the dropper, on the actual colour in the image you want to remove. The colour automatically adjusts to the one you have selected! (Thanks to Mike Flemming for pointing this out to me).

5. To remove the offending colour, move the Saturation slider bar all the way to the left, and check how it looks.

6. Try selecting and removing a different colour, if this hasn't had the desired effect.

Other sites
Andy Bright has a useful page on Photoshop for digiscopers, covering much of the above ground, but in a bit more detail in places.

Have you found this new page useful? Got any comments/further tips?

Why not get in touch by e-mail?

Home | Trip Reports | Oxon pics | UK pics | Damsels and Dragons | Other Nature | Links