Voicing Opinions

My new Kindle was amazingly cheap and is proving to be really useful. No wonder e-book sales are soaring.

Amazon has sold millions of these devices. This is a 4th generation device I believe. So why on earth does it have two basic usability faults? Can Amazon not employ testers able to detect what I discovered in the first 2 or 3 days?

First is the user guide. A pre-installed document that is the first thing I read. What better device than an e-reader to hold within it a primer! No printed manual needed.

But it falls breathtakingly short in helping a new Kindle owner. It totally omits instruction on highlighting passages, looking up definitions, how to cancel subscriptions, how to share document sections via Facebook, and so on. You are obliged to discover these things by accident or experimentation. What logic is there in failing to finish that document? Any fledgling graphic designer could have done so in few days.

Maybe I have yet to discover something it does, but how will I know?

On a more serious matter, there is no guidance on the management of the battery. Should the first charge be run right down? Should the device always be run right down? Or should it be handled like an iPad and rarely allowed to fully discharge?

The other bug is so basic that I find it staggering that no one spotted it. My Kindle is a non-touch device; you use buttons to make selections. When I select, for example, the 5th book in the book list page, and then return back to the book list, the book correctly moves to the top of the page. But the marker stays on the 5th book. So if I press the selection button to return to my book, I started reading a different book – I have to move the cursor to the top of the screen to return to the book I was reading. This is ridiculously fundamental to have been missed in product testing.

These are simple omissions to correct. But in a 4th generation product they should categorically not be there.

Having just negotiated a street of speed bumps, I thought I would put my thoughts forward on the matter, even if they are blindingly obvious.

Speed bumps, like speed limits, would appear to suitably constrain traffic so as to avoid incidents and accidents. But both introduce problems, and fail to tackle the key issue – the ability to observe the road around you, and drive accordingly. It is called defensive driving. But it is impossible to observe the other traffic and pedestrians if you attention is constantly averted to bumps in the road or your speed gauge.

Both also suffer with the additional problem of permanence. The bumps and speed limits are not correlated with their need. Only at short periods of the day are there pedestrians in presence for which the driving constraints were introduced for. Such as outside schools.

Finally, the bumps hardly do your suspension or tyres much good, even at a slow speed. Nor does the constant adjustment of speed do your fuel economy much good.

Sadly, the best solution is too subtle and too unenforceable. To allow only proficient, thoughtful, street/pedestrian aware drivers on the road in the first place. The good drivers are punished by the bad drivers.

I have been very busy writing a Web based Go game commentary site :

www.learngo.co.uk/GoViewer/ListGames.php

I wrote it using the new HTML5  canvas, and Javascript. Unsurprisingly, all browsers except Internet Explorer support it well. Let me know what you think.

I am not an authority on the technology of cameras, but I believe sufficiently well informed to shed some light on the excessive attention paid to the number of mega pixels a camera has.

Part reason for the timing of this post is that my friend and I have recently bought two very different cameras. Both use CCD sensors (Charge Coupled Device). Mine is a 10 megapizel Panasonic FX37 compact that fits snugly in my pocket. His is a 50 megapizel Hassleblad that does not fit in his pocket. My main camera is a 12 megapixel Canon 5d, that sports a CMOS sensor (Complementary Metal Oxide Semiconductor).

CCD and CMOS sensor types are different – each with a set of pros and cons.

The lens on my 5D, with a sensor pixel count only slightly  larger than the Panasonic, is probably100 times the volume of the lens on the latter. The sheer amount of glass, superior design, and design, allows it to resolve greater detail than the Panasonic. Depending on the lens aperture and focusing distance, it can yield detail that falls short of and exceeds the 12 mega pixel sensor.

The Panasonic lens uniformly fails to resolve to the sensor capacity – most images yield about 3 megapixels of detail. Detail that is smeared across 10 mega pixels. Additionally, because the sensor are so small in compacts that 10 megapixels results in tiny individual pixel sensors. One or two microns across or less. So they receive very little light in many situations – so little that noise manifests itself even at the slowest ISO speed.

The 5D has a large sensor, with much larger individual pixels. So the noise level is much smaller.

Detail resolution and low noise simply make for much better pictures.

However, this is not the whole picture. You might think that each pixel was measuring whatever colour light hit it. This is not so. A filter lies above each  pixel. One in two pixels only detect green light. One in four pixel detects red light, and one in four detect blue light. (We are most sensitive to green, hence double the number of green sensors).

So if you have a nice swathe of red, only a quarter of pixels are registering anything. If you were to photograph a grid of alternating red and blue lines that were very close to the pixel pitch, then some red lines would be missed, and some blue lines would be missed. From a distance, this mix of red and blue – maybe the weave in cloth – would look purple. But if more lines of red pixels were to light up, you would get the wrong colour.

So a thing called an antialising filter also sits in front of the sensor.

This blurs the image from the lens. Honestly!

It spreads light points out so that a single pixel sized light source will be sure to hit all 3 pixel colour types, and thereby get a better colour.

This is why camera images ALWAYS need some kind of sharpening. But this sharpening will not reinstate the lost detail.

It just so happens the Hassleblad has no Antialiasing filter. The computer software can compensate (apparently) for errors that ensue from this problem. But it means that the 50 megapizels from the Hassleblad are naturally sharper than from most cameras. The images are breathtaking.

It still uses Red, Green and Blue pixel filters, so cannot guarantee to get the exactly correct light level, but less detail is lost.

Strangely, the sensor is made by Kodak.

Now, Kodak developed a prototype sensor with red, green, blue and clear sensing pixels.  Half the pixels measured just light – no colour information. This makes sense – we are much more senitive to luminence than colour. But there rae no indications that this has been adopted yet. Maybe the antialisaing filter had to spread so much that too much blurring resulted. A shame, because such a sensor would be much less prone to noise.

If the 10 megapizels on my Panasonic are way too much for the resolving power of the lens, are we being cheated? The answer is yes and no. No, because 10megapizel images come out of the camera. Yes, because it is fattened out to that size. And the reason why? Simply because Joe Public tends to latch onto a quanifiable differentiator when buying a camera. Much much better for the manufacturers to give the pizel size – a better indicator of quality – but 1.2 microns means nothing in comparison to 10 mega pixels. A better 6 megapixel camera will simply not sell. Sad really.