Not much new to report on the camera front this time, as I had employed the 2005 annular eclipse as a "dry run" for this one. As ever, I used the faithful 600mm mirror lens plus x2 convertor but made an improved filter out of cannabalised eclipse specs as there were doubts that the previous one would absorb enough of the (potentially damaging) infra-red radiation. This time, I added a "collar" to the filter so it would fit over the barrel of the mirror lens, instead of into it, making mounting it easier. The new filter worked well, but was much denser than the old one so just I used it to track the sun between exposures during the partial phase, briefly swapping it for the old one for each exposure. I allowed the camera to make its own decision as to shutter speed during the partial phase ("very short" was the answer!) but used manual mode for totality, as I had in 2001. I started at 1/1000 sec and went up to 2sec, then down again to 1/250 where I stopped to await the end of totality. The final prominences were taken at 1/500 and 1/1000 before I dashed off three frames of the Diamond Ring, also at 1/1000. This sequence turned out to be about right again, I'm glad to say. The accuracy of focus is actually better on the slides than the scans so I think the slight fuzziness is more down to the resolving power of the scanner - I'll still be trying to improve things next time though!
We had new the digital camera running in "movie" mode throughout totality, and this turned out surprisingly well. I set the camera to take frames at 800x600 and 15fps, with the intention of cropping and re-coding the video afterwards to generate as small a file size as could reasonably be achieved so it could be downloaded from the website (but only via broadband, even then!). The raw (Quicktime format) file size was 211MBytes but after cutting out some "slack" before totality, cropping to 280x200 and re-coding in AVI format using the RAD Video Tools application with the Indeo® video 5.04 codec (set at a compression factor of 65) the final size was an amazing 14.2MBytes: just 6.7% of the original!! I also did an "audio only" conversion, in 8-bit mono: this came out at 2Mbytes.
Fuji's attempt at scanning the slides was fairly abysmal so I again had to do it myself (a long job!). I scanned at 1200dpi and chose an archive size of 1280x1024 for the shots of totality: this seemed to be the best balance between image detail and file size. These were saved as BMP's, and JPG's and GIF's derived from them as required. Strangely, the Fuji scan contained one image (4th contact) that had not been included as a slide: I was able to rescue this and create a useable image.
When centering the images it was very noticeable that the shadow of the moon was considerably bigger than the disc of the sun (as was actually the case). The motion of the moon was also apparent (shown by the corona appearing to move slightly between successive images) but I decided not to allow for this when centering as it would only affect composites. I hoped the corona composite would not be greatly affected as the frames were taken close together in time and the relevant parts of the images did not overlap much anyway. The animation contains almost all the images of totality, plus a few "fakes" in the sequence leading up to 4th contact (derived by reversing those on the way in!) to make it look right.
The "enhanced corona" image was produced by a slight variation of a method published by Gerald L. Pellett in 'Sky and Telescope' magazine in January 1998. The original technique took each original image destined for the composite and applied the Photoshop "Radial Blur" filter to it. This processed image was then subtracted from the original to produced a very low contrast image which, almost magically, contained within it all the detail of the original but none of the over-exposure. The difference images were then added together so the resultant contained all the detail of the full corona, in the same way that adding the originals would, but with none of the high-brightness areas. This image was finally combined with the longest exposure original (by a multiplication process, in fact) to transfer the full detail onto a picture of the full corona, before pasting in a central black disc for aesthetic reasons. I could not follow this process exactly, as I do not have Photoshop, so instead of the radial blur filter I simply slightly rotated the images - this is, in essence, what the filter does anyway. I then wrote programs to perform the subtraction, addition and multiplication operations (in BASIC, would you believe!) as the image-processing programs I use could not do this exactly the same as Photoshop. Complicated certainly, but I think the result was well worth the effort - I hope you agree!
Back