Displacement Activities

David Evans, and his wife Topsy, kindly lent me a couple of hours of their time, and their Huon Pine Go set. These are some of the photos I took at the Himeji Gardens in Adelaide.

I was a little late for Autumn this year, but I hope to have another attempt in a few months, when Spring comes around.

A friend introduced me to the work of Corinne Vionnet recently – which is where I first got the idea that I could use Frame Averaging to achieve the long exposure images I mentioned in another post – and it turns out I am currently working with someone else who has played with this technique of adding images together. Mark Wendell has created a great collection of images, using python and Imagemagick to average out a series of photographs. Where Corinne takes hundred of images from the web, Mark has had an idea of the final image he wanted to create and then taken photographs to generate this. Both of these techniques intrigue me a lot, and I have begun experimenting already.

I wrote a little script that walks through my reference collections. I point it at a certain directory ( like: animals, water, food, people, metals … ) and it looks for all possible image files. What it finds, it copies to a temp directory. I don’t want to be doing any image operations on my reference collection. Once I have my collection of images, I run a second script that takes all the files, puts them into a list, loops over the list dividing each image by the total number of images, saving the result out in a second temp directory. Once this first loop is complete it does a second process, where it get’s all the divided images, and adds them back together to create the final image. Here are some of the results:

[Show as slideshow]

For the image processing I’m using Shake, a discontinued compositing software from Apple. Mainly because it offers great command line options, 32 bit support, and it’s what I know. I’m simply using python to assemble a series of sub processes which send commands to shake. This seems to be working well. I have worked out several actions for resizing, panning and centering the images. I need to take advantage of 32bit colour and remove the srgb colour space, THEN do the math, but that shouldn’t be hard. It should hopefully allow me to get more contrast in the final images.

Photography is fascinating because it captures an instant of time that would normally slip away, and preserves it for us to appreciate more fully. This ‘instant’ is often a 60th of a second, but I have long been a fan of the art work by Michael Wesely who increases this ‘instant’ up to 3 years. Which gives a new view all together of our world. Some of his work can be seen here.

Working as Lighting TD, I am fascinated in what happens to the illumination of a scene and subject over this time.  You would really have to think of the resulting photograph as even more of a canvas, being very deliberate about your exposure and the subject of the photograph. You can’t just judge your composition with what you have in front of you right now, but with everything that will be in view of the camera over the entire exposure. The trajectory of the sun, how the shadows will move across the image, the way an object will look with indirect lighting, expected changes in the landscape, the flow of motion.

I have started to have a play with these things myself, though not with the continuous capture like Michael achieves in his images. I’m curious mainly with the passage of light through an entire day, and I’ll initially try and capture that digitally. I have a few things to work out; the first of which is how far down I can stop my camera, and what length of exposure I can get from that. From there, I will be able to work out how many photos I will need over a given period, to give me the smoothest result. If the shutter is really fast, then I will need hundred’s of thousands of images. But if I can get it really slow, I may only need hundreds.

I plan to stitch the images with a technique called Frame Averaging which can be as simple as dividing a series of images by the total amount, and then adding them back together to generate a new image. So if I was to take a photo every 30 seconds (at a 60th of a second), for an hour, I would get 120 images. Then divide each image by 120 (total amount of images taken), and adding all these new images back together (for now ignoring all gamma and colour space issues) to create a new image. The problem here is that if we are only taking a snapshot of a 60th of a second, the edges of shadows will still be too defined. Possibly fanning across the image. So it’s important to stop the camera down as much as possible. It would be great to have the shutter open for 29 seconds in the 30 second period.

Enough prattle. Here are some images from my initial tests. These were 4 second exposures taken over 3 hour periods, at 30 second intervals. The resulting images seem to be moving toward grey, and don’t have the highlights I’m expecting. I think this is because I’m not stripping the srgb colour space before doing the conversion, in my frame addition script. I’ll address this soon enough.

[Show as slideshow]

Game development is something that has interested me for quite a few years. Seeing all the amazing indie games that are being released, I can’t help but be intrigued. A while back I tried out XNA, the Microsoft SDK for community games, but my lack of understanding when it comes to Object Orientated Programming, coupled with having to learn C#, got in the way. I lost interest. Six months back I had a look at the iPhone SDk.  But this ended for the same reasons as XNA, just with Objective C being the language I struggled with.

Now, I’m going to have another attempt. This time using Python to ease me into the Object Orientated world. I use Python often in my work, but in a very linear fashion. Do this, now do this, this, this, and this, that to all those, not that one, and do these to them, done. Boring, but functional. Turns out there are quite a few options out there, for Python game development. I have settled on pyGame for 2d games, and Panda3d for games that require a bit more depth.

After a few weeks just getting everything talking ( everything being pygame, panda3d, python, osx ), a couple of weeks following tutorials, and another week or two of evenings tinkering, I made more progress than before in a much shorter time. And I have a basic game running! It has all the cool things: user input, sprites that are objects, extraneous debug info, simple vector math driving some rubbish physics, collision with screen bounds … and a rocket ship.

Here is a screen grab, showing it in action. I’ll post more, when I have more to say.

In the interest of getting some content on here, this is a small script I put together to help make isolating objects in Maya a lot faster and more intuitive.

Very simply, it runs the isolate command on whatever viewport you have active. Normally you have to go to the ‘show’ menu, choose ‘isolate selected’ and then ‘view selected’. And you have to do this for each pane you want to run it. Tedious. But with the help of this script -  assigned to a hotkey – you just work away as normal, and hit whatever hotkey you assign, to toggle the isolation of whatever items you want to focus on, in any modeling pane. It also works on sub components like faces and verts.

Here is the script. Information on running it, is in the file. Should be fairly straight forward.

isolateToggle.0.0.2.mel

Yea … so I now have a website. You should check it out.