I had this dilemma: what to use to image the whole disk of the Sun or the Moon? Make full disk pictures with my DSLR – 60 frames tops – or record many little videos with my webcam and put the mosaic pieces together? On one hand it is how much time I am willing to invest and on the other, what output quality I am expecting.
SCOPE: I used my 150/750 on the motorized EQ3.
DSLR: it is a Canon 1100D (also known as the EOS Rebel T3), unmodified hardware but enhanced by Magic Lantern (thanks guys :) ). The sensor is a 12.20 Megapixels CMOS APS-C (pixel size = 5.2 μm), capable of outputting 14 bits deep raw (cr2).
WEBCAM: it is a Scopium webcam, 640×480 pixels 8 bits sensor, max framerate is 30 fps, video is recorded in raw mode (pixel size = 5.6 μm).
I started with the DSLR when the Moon was 30 degrees high.
DSLR: I recorded 30 raw frames with the DSLR in prime focus. The frames needed to be converted to TIFF16 before being fed to AutoStakkert!2 using Canon’s Digital Photo Professional. After AS!2 did its job – stacking optimized for surface, no drizzle – I ran the resulting TIFF16 through Registax6‘s wavelets to get the most pleasing result. So the steps taken: set up the scope and camera (low iso, fitting exposure time, intervalometer), sit back, convert raw=>tiff16 (takes a lot of time, but I can sit back), stacking in AS!2, wavelet in R6. Ready. Since it was there, I kept the color data.
WEBCAM: I made a dozen videos, each about 800 frames in prime focus. I processed these videos in Registax 6. To get a pleasing result the individual mosaic pieces must be well overlapping, need to be somewhat cropped and then there is the manual fitting process. Note that this is a longer process and the longer the process the more chances are there for the athmosphere to change, thus substantially increasing the time needed to flatten the final mosaic. So the steps taken: set up the scope and the camera hooked to my laptop, framecount x (positioning the scope to get the right framing, sit back), framecount x (stacking and wavelet in R6), putting together the pieces. Ready. Didn’t really care about the colors.
Didn’t measure the times needed for each, but lets make an educated guess here.
TimeDSLR = (setting up the scope and the camera) + framecount x ( short exposure + intervalometer delay (set to 10 seconds) + (<1 sec needed for processing and saving the large raw file) ) + framecoutn x (converting cr2 to tiff16, at least 5 seconds) + (stacking with AS!2,a wild guess 1 min / 5 frames, lazy to test now :P ) + (wavelets in Registax6, done in way under 5 minutes) + (some clicking around) + (taking the gear apart and getting inside) = (setting up the scope and the camera) + 30×11 s + 30×6 s + (30/5)x60 s + 5×60 sec + (some clicking around) + (taking the gear apart and getting inside) = (stuff) + 30 * (11+6+60/5) + 5*60 = (stuff) + 677 seconds = (stuff) + ~12 minutes. Most of this time I can sit back. So together with carying the gear, setting it up and the disassembling it, let’s say the output is ready in 20 minutes, the (stuff) part of the function being about 8 minutes. This is an educated guess based on my habits, I will probably never measure the process.
Timewebcam = (setting up the scope and the camera) + (setting up the laptop – mine has no battery, have to carry the cable, though startup is superfast due to the SSD) + piececount x (positioning the scope to get the right framing, about 30 seconds + recording, about 1 minute + saving, about 1 minute due to AMCap being more than rudimentary) + piececount x (processing in R6, let’s assume 2 minutes) + piececount x (putting the piece to the right place, cropping, offsetting and making it fit by applying touchups at the borders, let’s say 1 minute) + (some clicking around) + (taking the gear apart and getting inside) = (stuff, 8 mins) + (setting up the laptop, 3 mins) + piececount x 2,5 mins + piececount x 2 mins + piececount x 1 min = (stuff)+3+12×2 + 6 + 12×2 + 12×1 = 8 + 3 + 30 + 24 + 12 = 8 + 69 minutes = let’s say 1 hour and 17 minutes.
Can’t say there is an order of magnitude difference between 20 and 77, however getting a final picture with a DSLR takes well under 1 hour, while getting one by mosaicing the webcam frames takes well over an hour.
TimeDSLR = 26% Timewebcam
What about quality?
Well, on one hand there is the 30 fps promise of the webcam, though only 8 bits depth. The higher the framerate and the lower the exposure time of one individual frame, the sharper the outcome. On the other hand the DSLR has 14 bits, and 1/125 would in fact correspond to about 8 fps. Does the higher fps and the higher number of frames (ie. better s/n ratio) overcome the superior APS-C sensor? The pixel sizes are roughly the same, the DSLR being the winner with its marginally smaller pixels. Long story short: no. I couldn’t find significant differences between the two outcomes. Basically they are the same picture but one is made the easy way and the other the much harder way.
The winner is: the DSLR
The DSLR does the job in less time without the huss and fuss of staring at the screen, positioning the scope, minding which region to record during the imaging phase and then mosaicing the outcome. My setup proved there is no gain in using an unaided webcam instead of a DSLR when imaging the whole disk. Obviously, with the webcam+Barlow lenses one could record finer details on the surface than with a DSLR which has no crop mode and – face it – a DSLR is not the same with a Barlow lense. But mosaicing a whole surface with a setup which has say 1/4 or 1/8 or 1/16 of the field of view of the webcam config detailed above might turn the night into a nightmare. The outcome may or may not be a giant gigapixel size poster, but I have no intention to make such a surface map – it’s just not fun enough, it is only the iteration of a nerve consuming process. And it would be like taking a portrait picture showing each cell of the skin…