Google Wave: First Impressions

I am lucky enough to have been sent an invite to Google Wave. Here are my initial impressions.

Move over Silverlight/Flash

The interface is amazing to say it is not flash or silverlight. It is slick and easy to use in both Firefox/Linux and Chrome/windows. It does not struggle at all on my netbook and also it fits just fine into the 1024x600 world without tweaks. I really see this as stealing some of the attraction of more complex browser plugging type interfaces like Silverlight or Flash. I clearly is possible to make a really clean, smooth operating interface just with AJAX/HTML. I can only assume it is the power of the GWT (Googles Java->Javascript based web interface toolkit) which has permitted this level of quality.

Yes - there are some bugs in the interface but wow - this is in a new league.

The flip side to this slick interface is that it does not seem to do very much which is actually useful. I suspect this could say more about me than about the application its self. I lost interest in Facebook and ever got into MySpace. Wave definitely has that feel about it; it seems to be about talking a lot and saying very little.

I have tried to figure out how I would run a webinar or presentation from it and failed to be excited. I also suspect it is a long way from being a SharePoint killer yet as it does not seem to support the document handling required. Maybe if Google link up Google Docs with Wave then they will have a killer application indeed?

In Conclusion

The very first thoughts are that the technology of Wave could mark a water shed. If browser based interfaces can be this good, then there might be a time when everything but hard core gaming is browser/html based.

The second thought is that Wave is underwhelming - maybe. It is possible that as multi-site collaboration tool it will be very good - kind of a merge of email and im. I am not sure yet. We will see when more users are on the system.

Rest assured, I'll be blogging the minute I have more thoughts on this!

Landscape Photo Enhancement - Tricks With The GIMP

Ordinary pictures taken at a normal time of day with commodity digital cameras never seem to capture landscapes as we humans see them. One reason is gamma, here is how The GIMP can help.

The gamma of a picture is the brightness shown in relation to the brightness picked up by the camera (simply put). Humans tend to compensate for poor light and misty conditions when they look at things by taking into account the context of the view. When an image is taken out of context, we loose the ability to compensate in this way. Part of this compensation can be re-created by changing the gamma of an image in an editor like The GIMP. You can then go even further to make an image look more intense then it would even to the naked eye.

Step:1 Take a not very exciting looking image

The above image is taken with 3 times zoom from an aeroplane window. The canyon is being seen through a lot of air and so looks misty and washes out in the mid day sun. To start to fix this we can load it into The GIMP and open up the 'curves' window.

The curves windows (above) shows the amount of pixels in the picture which are at each brightness starting from black (left) to white (right). As we can see, the picture only really as large numbers of pixels in the middle, there is a big section at the right and left of the graph with no pixels. The first step it to grab the ends of the line through the window (bottom left to top right) and drag them across the top and bottom so that they line up with the low and upper bound of the area with a reasonable number of pixels (see below).

This immediately improves the image; however, we can make it even more striking. By clicking on the sloping line its self (rather than the ends) we create a drag handle. Moving this drag handle bends the gamma line. If we do this at the top and the bottom of the image we can make a curved S shaped gamma line. The idea is to get the maximum slop in the areas with the most pixels and a more gradual slop where there are less pixels (the graph is lower). This bit requires careful hand adjustment for each and every picture. If you end up in a mess - hit the 'reset' button!

Now the picture should look really intense. However, the process can make the noise in the image stand out. Most cameras make pictures which are larger than one needs for the web. Also,if a digital camera says it is 12 mega-pixel that is not actually 12 true colour pixels, it is 12 million pixels in total. Some will be green, some red and some blue (usually less blue because the eye is less sensitive to blue). This means that for many pictures, the effective resolution of the image is actually 3/4 or 2/3 the dimensions of the reported resolution. This is one of the reasons the image from a commodity camera can look noise when viewed a full resoltion.

We can take advantage of all this by performing a very gentle blurring. By using a selective Gaussian blur with radius 1 or 1.5 and a cut off of 24, we get rid of a lot of noise but very little true detail. We can then use a sinusoidal function to reduce the size of the image. Here I a shrinking my 2592px image to a 1600 px one for upload to the blog:

After all that a tiny little bit of unsharp-mask can help, but not always. Too much and the image will look noisy and fake.

Here is the finished product. It does not 100% real, but it does look very striking. Actually, the marble canyon does not look real anyhow!

Also notice how the image is bluer at the top than the bottom. This is due to riley scattering. It is possible to correct for this as well, but very difficult to do so and get an image which looks authentic. I might have a go at creating a blog post on this another time.

Out of curiosity, here sis the curves window for the new image; we can see how much more spread out the pixel density is across the brightness range.