Account types in Windows

On a properly designed system, users should have permission to do all that they need to in everyday life without having to resort to using special (admin) rights. The reason for this is that admin rights allow the user to perform actions such as make system-wide changes, install programs and access other people's accounts.

If a user was doing something questionable such as browsing with scripting permitted, a hostile user could take over the computer (yes, really. Think about it) and they would have the same rights as that user. If those were admin rights, the whole system would be at risk.

Unfortunately, Microsoft is only just getting around to implementing this (even though Linux and UNIX users have had this by default for decades) and in Windows, it is not the situation by default.

However, you can do something about this before whatever comes after Vista by creating an account that has admin rights - you can call it 'root' if you want to - and then changing the privileges of all of the other accounts to just a normal, limited user account. There is no secret about the account names that you use because any user can see them all.

Use the limited account for everything unless it won't let you. It is then up to Microsoft to sort out its security.

And the password itself...

There is no point in choosing an admin account password that is easy to guess so you might as well make it hard to remember as well.

Choose something with at least 8 characters - 12 is better - and don't limit yourself to just letters of the alphabet. A password such as qjd:5g92.wmg will survive dictionary attacks and is long enough to need to be shoulder-surfed three times. If you never write it down or send it using a protocol that passes such information in the clear (such as telnet, ftp or http) or use a program that uses weak keys such as Internet Explorer then you should be able to use it for years.

Shoulder surfing?...

Shoulder surfing is simply looking over someone's shoulder while they type in a password (or PIN).

People can remember four characters at a time so if your password is eight characters long, they will have to look at it twice - the first time to remember the first four characters and the second time to remember the last four. If you have nine characters, there could be a lot of confusion (possibly giving the game away that their is someone trying to break into an account if unsuccessful logins are logged). If you have twelve characters for them to remember they will have to observe you three times.

Using characters that require modifier keys to make them work is another trick because then, they have to try to remember which keys you were pressing - was it an 'f' or an 'F' and so on.

So, more characters from a pool of more characters. Now, where does that put chip and PIN?

Infra Red - 'Super Black and White'

As long as we don't want to start taking pictures of anything too out of the ordinary, or expecting too much from our cameras, that is.

The skill of taking good black and white photographs has apparently been superseded by the ability to force an image to lose all of its colour saturation information and we can therefore create a perfect black and white image just with the click of a mouse.

Like so...

There is little contrast between the sky and the clouds and the whole thing looks very grey.

You can modify the weighting of each layer so that you might end up with a bit more of, say the red layer to make your blue sky darker or, say, the green layer to make your leaves a bit lighter but here, like so many pictures, the image was taken with colour reproduction in mind and therefore the skill of taking a good black and white image was never written in.

You cannot write it in afterwards - no matter how much you paid for you image processing program. Garbage in, garbage out.

The best you can do when mixing layers is to get just one of them - mixing in others will only reduce any effect (which is all right if your image will permit it).

Clearly, taking a colour image and trying to make a black and white image from it is not guaranteed to work by any means.

So, just what is a black and white image and, for that matter, what is a colour image?

We naturally look for these images when we take photographs because these are what we have been looking at since we were born. This, however, also represents a problem when we take black and white photographs because we need to visualise a scene differently when we take such photographs.

Of course, without any colour information to have any contrast with, a good black and white photograph has good luminosity contrast although that does not mean that you can zap the contrast up on any old image to make it look good.

The image on the right is the same scene as the colour one above but with a Wratten 87 filter (although it was necessarily taken a few seconds later so that I could hold the filter in place).

The camera is nothing special - it is a Samsung Digimax A400. With images like these, the light that is allowed through the filter (which looks black in normal light levels) lands on the sensor in the camera. At those long wavelengths, from around 750nm down to the sensor's lower limit of around 900nm, the red, green and blue filters have little meaning and the light reacts fairly evenly with them all (although we always end up with the magenta tint because one of the dyes in green layer isn't quite the same at these wavelengths. Note that for a dye to be green, either it has to be a single molecule that has two electronically separate sections so that one can absorb red and the other absorb blue; or, you use a mixture of two dyes, one that absorbs red and the other that absorbs blue. With the latter, you have more control over the absorption).

Unfortunately, so the story goes, some people in Japan discovered that if you use an IR filter that tends to pass towards the 900nm part of the usable IR spectrum, thin layers of clothes disappear (or at least partly). And therefore the camera manufacturers use sensors that have the IR below around 700nm filtered out.

So, we have our useful 700nm to 900nm band filtered by an IR layer so that naughty people cannot remove layers of clothing and the residual absorption of any colours from visible light filtration (such as the red absorbing dye in the green cells) and the image on the right is what we get.

Also, you might note that leaves reflect a lot of infra red light. This is because they absorb red and blue for the photosynthesis process but don't need to absorb IR or green. This is why they look green to us and if you use colour IR film, they look magenta (IR and G shifted up one colour so they become R and B).

Shots like these cannot be derived from colour images that haven't, at some stage, started off with IR.

Digital-Image Equivalent to Retinal Contrast Effect

In issue 247, I showed you how to make an image with extremes of density with a void in the range could be modified so that you can see more of each of the two extremes - in that case, a bright sky and a dark corn field.

However, you can get these extremes in an image where, instead of just trying to see more of the detail at each of the extremes, you actually want to preserve the impression of density change where the border between the two exists.

This type of effect is usually seen where you have a shadow and want to see what is in the shadow as well as elsewhere but preserve the fact (although implicitly) that the difference in light levels between the highlight and the shadow is still quite high.

In order to do this, we will need to expand the extremes as before but give the impression that we are extending them beyond the density range of the image, not within it. Just like in the image on the right.

Of course, we cannot do this so let's look at the image and also look at how the eye does it...

All of this will be done on the Gimp so you can do it for free on any operating system from Windows, BSDs and Linux to the Mac. Also, being the Gimp, you should be able to do this on Photoshop if you have gone and bought yourself a copy.

For those not familiar with the Gimp, the main toolbox is in the top left, the layers dialogue box is in the bottom left and if you double-click on any of the tools, you will get a context sensitive toolbox options dialogue box open up for you.

You can see in this painting (if you look closely, the lower layer - made from potato chips - is in the shape of the north of Scotland with north pointing to the left so that the east coast is at the top - you can work out where the rocks and the pole in the middle is and go and find out what it is if you want - this was painted in 1983) that on the inside of the shadows the surface is darker towards the edge. This gives the impression that the shadow is a lot darker than it really is.

So, how do we get our image to do this?

Let's build a mask and take it from there.

So, duplicate the existing image so that we can mess around with it and still have something at the end of the process.

To do this, right-click on the image in the Layers dialogue and click on 'Duplicate Layer' in the menu.

You can now select a level at which the image undergoes a transition from black to white and you can do this either by dragging the marker around or by scrolling the number at the left.

You can also do this at the right if you want (or instead of - which would make more sense but who is to judge about uses you will have for this tool in the future?)

Click on a black area of the image and all of that colour will be selected.

The radius to choose depends upon how big your image is, how big it is going to end up.

Ideally, this value will be no bigger than the smallest 'blob' of darkness (or light) on your image.

Next, click on the foreground colour and change it to a value of '7f7f7f' (or 127 for each of the RGB values - you might want to change this or experiment with it depending on your photograph), then clicking on 'OK'.

Finally, right-click on the upper image in the Layers dialogue box and select 'New Layer...' and then make sure the 'layer Fill Type' is set to 'Transparency' before clicking on 'OK'.

In the 'Layers' dialogue box, the new, transparent layer should be selected.

This function will draw a line around the selected parts of the image using the border of the selection as the vector.

Selecting the area, changing the brush and changing the colour have all been for this function.

Note that instead of this, you could use any of: 'Pencil'; 'Paintbrush'; 'Eraser'; 'Airbrush'; 'Ink'; 'Clone'; 'Convolve'; 'Smudge'; or, 'Dodge/Burn' so this is quite a versatile and powerful tool.

Click on 'Stroke'.

This is half of what we want with this new mask.

Instead of just pasting it into the current layer (it might well be in the wrong position any way), we right-click on the pasted layer in the 'Layers' dialogue and then select 'New Layer...'.

Instead of giving us the new layer dialogue box, the pasted layer gets its own layer which we can now position using the move tool.

Click on the other layer (clicking once on the text will select it without doing anything to it) and then, with the image zoomed in (maybe 200%), move that layer around (to select that layer, move the mouse over the image until the cursor changes from the normal mouse to one with an arrow and the 'NSEW' arrow cross) until its sharp edge matches the black and white mask below it.

Next, click on the eye icon that you previously hid in the layers dialogue box - this will bring back the other layer.

We now have all of the bits in place to change the image.

Next, change the mode to 'Subtract'. This will subtract the darkness from the shadow and subtract the lightness from the lit areas. Note that the resulting image will display some of the detail in the shadow but not the highlights - this is because when we made the black and white image using the 'Threshold' tool, we selected a threshold that was in the shadow part of the image density range.

In order to make this image usable, we need to reduce this layer's effect. This is done by changing the opacity. For this type of effect, something between 10% and 25% will do - much more than this and the final image density range will be squashed up too much.

Next, click on the eye icon to turn on the next layer up (the second one down - this is the highlight-side blurred border). We want to add this to our image so select 'Addition' as the mode and an appropriate level of opacity (remember that we used a 50% ink to start with) would be around 25% to 50% (ish) - this depends on your image.

Finally, bring in the top layer (the layer that blurs the border just inside the shadow). We need this to darken the border so that the local contrast is increased. So, select 'Subtract' as the mode and, as we are already in shadows and close to the limit of density, virtually any value will do 0% to 100%.

When you have finished it, save the layered file as an .xcf.gz file (you do not have to be on a UNIX-like system to do this as the Gimp has .gz compression built into it).

Next, press [Ctrl][D] to duplicate the image and then flatten it by right-clicking on one of the layers in the 'Layers' dialogue box and then clicking 'Flatten'.

Now, right-click on the image and select 'Layers', 'Colours', 'Levels' and drag the image density markers to the ends of the image density or click on 'Auto'. You can treat the image like any other image by tweaking its saturation, curves and so on.

Another point is that you can also use the mask to divide up your original image so that you can process each part differently - density range, colour correction ...

Note that the area in the shadow is lit by the diffuse blue light from the blue part of the sky whereas the areas in the sunlight are lit with a light that has had that blue taken out of it (it goes to make up the blue of the rest of the sky for other observers).

False colours from the GIMP

Images with hidden detail

Good black and white images, such as the glass-backed watch below, have a good range of densities and are nice and clear. However, in some imaging applications, you need to be able to see some of the details in what would otherwise be interpreted as areas of fairly uniform grey.

Importance of hue sensitivity:

You can see this already with analogue colour TV signals. The colour information is transmitted as two signals on a single carrier using a technique called Double Sideband Suppressed Carrier Modulation (or DSB) - if it was transmitted on two different carriers, you would get interference patterns in the output colour signal. This all sounds a bit complicated but essentially, if you modulate a carrier with a signal, you get the carrier along with an upper sideband and a frequency-inverted lower sideband. This is normal AM (Amplitude Modulation) and you can decode it with a non-linear circuit such as a diode in a crystal set. However, you are transmitting a lot more power than you need to with AM. You can get rid of the carrier by filtering it out and this leaves the upper and lower sidebands (known as 'DSB'). You can remove half of this again by filtering out the lower sideband thus leaving a single sideband (SSB).

To decode a suppressed carrier modulated signal, you need to inject the carrier back into the signal again and threat it like a normal AM signal - just like a crystal set with carrier injection. An SSB signal and DSB signal will reproduce the original signals quite well. However, it is when things start to go wrong that it gets interesting.

With an SSB signal, if your decoding carrier drifts off-frequency, the decoded signal either has its frequencies added to or subtracted from (depending on whether the shift was higher or lower than the original) with negative frequencies wrapped around 0Hz if it was lower than. Note that the frequencies are added to/subtracted from and not multiplied, so music (where a key shift is a multiplication of frequencies) will sound increasingly horrendous but voices just sound increasingly odd. With a small shift (say 4Hz) a sound system can be prevented from feeding back and when this technology was first implemented, some groups used it with feedback as an effect - Pink Floyd at the end of 'Echoes' for example.

With a DSB signal, if the decoding carrier merely goes out of phase, the signal is cancelled (completely at 90 degrees). Therefore it is important for a DSB signal to be decoded with the correct phase - not just the correct frequency. However, this can be used in an interesting way. If you send one signal down a carrier, you can send another signal down a carrier of the same frequency but phase-shifted by 90 degrees. Once you have removed your carriers, you have a composite signal that you can broadcast.

At the other end, you pick up the signal but have to know the phase of the signal to decode it properly. This is done at the beginning of every scan line with a 'colour burst signal'. This lets the decoder know the phase of the signal and it is able to decode it properly - using the normal decode-carrier for one signal and a decode-carrier that is phase-shifted by 90 degrees for the other signal (the carrier being 90 degrees out for one signal ensures that it is cancelled out. That's what I call clever - especially as this was all designed so that black and white receivers could still pick up the signal and process it correctly for a black and white set.). So, what can go wrong now? The signal can have an effective path length change usually caused by things moving and reflecting the signal before it gets to the television set. This can cause a phase shift that will change the hue.

• In the USA's NTSC system (called, rather cynically, 'Never The Same Colour') the colour vector information is the same on each line so if you get a phase-shift-induced hue change, people end up with green faces (hence 'never the same colour').
• In the UK's PAL system (called, rather cynically, Pale And Lurid [it actually stands for phase alternate line]), the polarity of alternate lines for one of the axes is inverted before it is transmitted. In the receiver, it is inverted back and added to the previous line's signal for that axis using a 'bucket-brigade' device to delay the signal by 64µs. In this way, the colour information is spread over two lines but the black and white information is on each line - you can get away with this because the colour receptors in your eyes are not as densely packed as the black and white sensors. The end result is that it looks the same but if you get a similar phase shift, the two rotations cancel each other out so that the hue stays the same - the only loss is that the saturation is reduced (hence, 'pale and lurid').

One thing you can do - should you ever get the chance - is to look at the colour of the beam in a UV-Vis spectrophotometer. Put a mirror (or something shiny like a spatula) in the beam so that it is reflected upwards. Next, make sure that the dial is somewhere between 700nm and 420nm. The light levels are low (they have to be for it to work linearly) so you should not damage your eyes unless you have some obscure and previously unknown medical condition because your eyes will already have seen light at well above these intensities during normal, everyday living. Now, look into the beam and turn the dial until you see the same hue as sodium light (the yellow streetlights that make any colour just a shade of yellow). You see these lights all of the time so you will surprise yourself as to how accurately you can remember their colour. When you look at the dial, you will see that it is on (or very close to) 589.3mn (5892.9 angstrom units). If your memory of hue is that good then your immediate perception of it when next to something to compare it with is even more important.

So, hue is important as our eyes are very sensitive to changes in it. The next question is how do we add it to a monochrome image such as the one above?

Creating a false-colour gradient

In the Gimp, you could HSV-decompose an image and put the V layer into the H layer then flood-fill the S layer with white and when you recompose, you get a false-colour image. However, although this works, it is not particularly flexible or dynamic - there is no way of optimising it in real time (you can mess around with the gamma and so on for the H layer but you have to do several steps to see the result).

As always (or at least most times), there is more than one way to do it and with the Gimp, this is fairly easy.

You can use an existing gradient (skip to 'Using and adjusting the gradient') but you can make it more interesting.

Click on 'New gradient' and...

There are several important areas to this dialogue box:

• without trying to sound too obvious, the title (currently 'Untitled') gives it an identity that you can use;
• the gradient area itself shows what the gradient looks like;
• the area just under the gradient that comprises of a mid-grey bar with three, upwardly pointing arrows in it; and,
• the 'Instant update' checkbox.

Make sure that the checkbox is checked.

Next, give it a title - I'm just going to call it 'temp' - then hit the 'save' button.

You will see the gradient appear in the Gradients list box in the Gradients dialogue box. Now, - as that is the current gradient - you can click on it in the list box if it isn't already for some reason - any changes you make to the gradient will appear automatically in the gradient in the Gimp toolbox.

Right-click on the bar at the bottom and select 'Split Segment Uniformly...'.

On the next dialogue box, you select the number of segments to split it into - here, we need three - so move the slider to 3 and click on 'Split'.

Let's click on the middle section.

Right-click on the highlighted segment of the gradient. Select 'Load Left Colour From' and then a colour. Here, we are going to use the red so just click on that. There are a number of other options but we'll come to them later.

Let's repeat the process with the other side. So, right-click on the highlighted segment of the gradient and select 'Load Right Colour From' and then blue.

We've got the end colours the same but remember that we wanted the colours of the rainbow between. We could do this by dividing up the red-blue segment and then assigning left/right colours to them all but there is an easier way.

• RGB - this is what it is now. If the first colour is red and the last one blue, the level of red decreases as the level of blue increases as the colour is interpolated. The green never gets a look-in;
• HSV (anti-clockwise hue) - this takes the starting hue and the finishing hew and as it interpolates, it takes the hue, saturation and value - this time taken in an anti-clockwise so between red and blue, there are all of the natural hues; and,
• HSV (clockwise hue) - does the same as 'HSV (anti-clockwise hue)' above but in the opposite direction so, between a saturated red and a saturated blue, you get a saturated magenta (unless you swapped over the blue and red in the gradient).

Select HSV (anti-clockwise hue).

Now, we can concentrate on the areas to the left and right of our central rainbow.

Your gradient should now look like this...

Now, we are ready to use the gradient.

Using and adjusting the gradient

Right-click on the image and select 'Filters', 'Colours', 'Map', 'Gradient Map' and the current gradient will be substituted for the greyscale.

1. make the change to your gradient;
2. give your image the focus - say by clicking on the border;
3. press [Ctrl][Z] to undo the last change (returning it to monochrome); and then,
4. apply the mapping again which will automatically use the updated version of the gradient without you having to click on any 'save' button.
   

Online Drawing

There are limits to what you can do legally on a web browser and if your job is about trying to find other things to do (don't tell me about it, I'm a journalist), you'll spend a lot of time on one.

This site is marginally more productive than a Tetris site and you'll know that somebody you have never met will like your drawing efforts. Probably.

Of course, a laptop touchpad is a great leveller and it really doesn't matter if you have decades of experience of oil painting or you are eight years old - they will look the same (use a tablet instead).

Try it out for yourself by clicking on the link here http://www.sketchswap.com/.

Lithium Ion Batteries

With many types of rechargeable batteries, only partially discharging a battery, recharging it fully, then repeating this behaviour will create a 'memory' - the battery will only discharge to that level after a while.

In order to prevent a memory, charge the battery fully and then use it until it has run down fully before you recharge it - using it only ever in a full charge/discharge cycle. However, it doesn't take an experienced laptop user to notice that this type of use is more like mobile phone use that laptop use.

One thing that is worth remembering though is that Lithium Ion (or 'LioN') batteries don't suffer from memories so partial charge-cycling is not an issue.

Fortunately, the rate at which they degrade is to some extent dependent upon temperature. So, whilst your current, dead battery is no longer fit for use, this is what you need to do with your next battery:

At 25C, a laptop battery will lose around 20% of its charge per annum with this rising to around 35% at 40C - a normal operating temperature for a Windows machine (try BSD or Linux for a cooler machine).

If you are going to have your laptop switched off, or run it off the mains for a protracted period of time, take the battery out and store it somewhere cool.

Ideally, you should have it at around 40% charge and keep it in the refrigerator (sealed in a plastic bag so that moisture cannot get to it) although you must not let it freeze. At around 0C, it loses only two per cent per annum at a 40 per cent charge.

However, when you take it out of the refrigerator, let it get back to room temperature before you take it out of the bag so that you don't get moisture condensing on it.

So, to make it last a long time, keep it in the refrigerator in a plastic bag at 40% a charge and only use it for as short a time as possible. That way, it will last as long as it can.

The next question is why don't the manufacturers make it clear that that is what you need to do and tell you how long your battery will last under various sets of conditions you are likely to encounter?