Current Status:

Third year has begun and my enthusiasm proportionally waned. It's hard to quantify exactly what is is about an education at UCE that can be so draining. As yet i have not been assigned a tutor to guide this project but this might turn out to be a good thing. As long as i work.
The first version of the program to reconstruct images is now being tested, i hesitate to say that i have only directed it's development rather than written it but this is mainly because of time constraints. It's written in an age old language, GLPro but it it does the job kind of. I gathered 24Hrs of HDV footage at 1440x1080 so i could gauge the look of images relative to the number of images input but as yet have only managed to process 65,800 of the 2.1 million frames that make up the HDV video file - this is because of a peculiar limitation of Quicktime and MPEG Streamclip to open files over 6.5Hrs or 300Gb in size. By now it's obvious that i have avoided any talk of the content or subject matter that may be contained in these static timelapses. This is for a mixture of reasons, im still developing the idea conceptually in my mind and im of the unfortunate persuasion to not really believe in myself needing to be involved in the authorship of any media. In essence im not really interested in subject matter layer but the process's below that. This is something that needs far more exploration and a dash of theoretical justification to keep tutors from telling their has to be a concept whiles they fail to see it just not in the form of expressing myself with pictures of peoples pets.

Technical Demands

From a purely technical standpoint the creation of each image will be demanding. To obtain a high volume of shots over a period of days will require the camera to be operating 24/7, so tempted as i am to do it with my DSLR it's mirror mechanism would almost defiantly fail. My research has lead me to the Canon G7, mainly because it can interface with a iStopMotion and it's whopping 10 megapixel jpg's. Neither of these are the ideal but are the best for getting the job done in my time-frame. Canon's own software has a time-lapse function but is limited to a measly 1000 exposures and short of using the SDK to create a personal version the choice is between iStopMotion or GBTimelapse

As i envision the final images being printed 30x40 on SuperGloss i would love to be able to work in Raw and get the most out of each image but with the possibility of upto 500,000 source images this is impossible. That said SuperFine Jpg's are almost as good and at 10Mp will certainly scale. Of course their is a possibility that they will need some treatment as small sensors are notoriously noisy and i have a plan! Either batch de-noise, massively time/processor intensive or de-noise each segment used but de-noising software requires bigger area than a few pixels square (tiles) to calculate from my program for cutting them up would need to be modified to provide bigger tiles for this stage.

Looking back over this text it makes little sense to anyone unfamiliar with the concept. I will gather my thoughts and write with clarity the idea soon. In the mean time i'm going to source images at 25fps via HD video camera, even at 25fps it will take 28hrs to capture enough frames to create a final image at 1920x1080. Hopefully all will become clear.....