Don't be afraid of CCD
Webcams and other digital cameras (IV)
At the time of the electronic integration, small webcams at a few hundreds euros and digital astronomy cameras have seduced a lot of advanced amateurs yet used to work with devices of another quality. Why such a passion ?
Honour to pioneers, the webcam was invented in 1993 in England, at the Cambridge University Computer Science department.
In 1994, Jeff Schwartz and Dan Wong then students at San Francisco State University (SFU) did the same discovery and developed the "fogcam".
Pros and cons
First of all, webcams are cheap and display a reasonnable resolution between 320x240 pixels and more than 3 Mpixels depending on models. They support images in VGA or full HD format and video formats AVI, some WMN or MOV. Their price increases with their performances (40-150 €).
Able to record between 5 and 60 fps depending on the resolution an their performances, individual images can display a very good quality, an excellent color balance, contrast, clearness and a sharp image on models like Philips ToUcam or la Logitech Pro 9000.
In view of their low profile and lightness it is also very easy to fix them at the eyepiece of a scope using a simple adapter or to build oneself an adapter with second-hand parts, as explain French-speaking fans on Astrocam Yahoo usergroup.
However, technically speaking the sensitivity of the CCD chip drops quite rapidly in blue light but offer a good efficiency up to the near infrared.
If webcams benefit of a low price, a light weight and are simple to use, they require a direct connexion by USB to a computer on the observation site.
Digital astronomy cameras
Digital astronomy cameras (ImagingSource, iNova, etc) also request a link to a computer but they are more flexible and performing. More expensive (200-700 €), they are also a bit heavier (300 g in average) than webcams and a bit more compact that classic CCD cameras.
In addition to their excellent image quality astronomy cameras sold by ImagingSource for example are equipped with a USB 2.0 port, Firewire (IEEE1394) or Gigabit Ethernet (GigE). The high-rate connexions are required because these cameras can reach a 1.2 Mpixels resolution and a rate of 60 fps.
These cameras, including the USB model from 'iNova displayed at right support most image formats RAW, BMP, JPEG, PNG, FITS et TIFF as video formats AVI and SER.
Camcorders, DSLRs and digicams in video mode
Conversely, HD camcorders (e.g. Canon HF200) like DSLR with video capabilities (e.g. Nikon D7000) are autonomous, versatiles but rather heavy (300-1200 g) and mid-end and high-end models, even without lens, are till more expensive than digital astronomy cameras.
In general these systems support the AVCHD format (MPEG-4 Ethernet) and sometimes MOV while dificams (compacts) usually support AVCHD Lite and Motion JPEG (M-JPEG) in low resolution.
Exposure times are generally ranging between 1/10000th to 60 minutes for an ImagingSource camera, from 1/8000th to Bulb for DSLR and from 1/2000th to 1/2 of sec for camcorders. Usually this range is never used at full because the Moon for example supports exposures times between 1/500th and 1/10th of sec. However, for planets we can go up to 1/10th of sec or less.
What ever the camera used, the CCD or CMOS sensor being usualy not cooled, astronomy images recorded in low light conditions appear grainy, all the more at summer when the outside temperature increase the thermal noise.
The termal noise can however not be noticeable at the speed of 30 images per second, maximum exposure time of most off-the-shelf cameras, as the brain integrates successive images rendering the grainy effect much less obvious.
These systems give excellent results if there is enough light and if we know their limits. In this regard, in planetary imaging, the exposure time is often intantaneous and thermal noise, even if it is low on some models, does not always permit to get images of quality (see examples in links page 5).
For deep sky imaging or for any low-light celestial object showing many details, conditions worsen because the exposure time can reach tens of minutes to reveal all its extent or structure. The noise becomes so visible that the substraction of a dark frame is practically mandatory as explained previously.
However, there is an alternative to control and reduce this thermal noise.
To reduce thermal noise in images the unique solution is extracting from the video the best individual images (frames) and to stack (add) them to increase the definition and thus the quality of the resulting image. In this case we create a film with the objective to generate a high-resolution image.
This procedure reduces the noise (residual noise is inversely proportional to the square root of the number of framed averaged) and averages out images shifts due to seeing. This is still truer in color.
Indeed, as for an LRGB composite, it is not important that images extracted from RGB channels are a bit blur or shifted (the lesser the better of course hence the use of reference points in each frame to combine) because this is first of all the luminance that will give its contrast to the result, RGB images reducing only the electronic noise and averaging the seeing.
In working at 25 fps, in two minutes we can record a HD 720p sequence containing 3000 individual frames. The video being directly available in digital format, if needed we can easily extract the best frames of the sequence then stacking them to increase the signal-to-noise ratio and the image dynamic (range of colours and details) as we very well see on documents displayed below (with a DSLR and a webcam).
To see : Saturn before and after processing
Canon EOS 450D on Celestron NexStar 5 SE XLT
To download : HDRinstant, software by HDRlog
Among imaging software supporting the stacking of thousands of images, name Registax from Cor Berrevoets, IRIS from Christian Buil, Avistack from Michael Theusner, Astrostack and HDRinstant from HDRlog.
Thanks to these software, it is quite easy to convert filmed sequences of an object in a single picture resulting of the stacking of several hundreds to some thousands of individual frames extracted from the best sequences.
In other words, if we take a small scope of 5" of aperture showing a theoretical resolution of 1.1" (Rayleigh limit), and reaching with difficulties a photographic resolution of 10" in eyepiece projection, the stacking of 1500 frames made under good conditions will offer a photographic resolution near 0.5", so 22 times higher than a raw image !
Duration and size of video recordings
For technical reasons, due to the file size and the low transfer rate between the camera and the computer (few cameras have a Firewire interface at 50 or 100 MB/s), the recording is usually made at rates between 5-10 fps, rate limiting the size of files to some tens of megabytes.
Indeed one must known that for a resolution of 640x480 pixels and 24-bit depth per frame, each image is 0.92 MB. Recording a 10 seconds AVI film at 10 frames/sec (thus a 100 frames film) will request a space disk of 92 MB. Avoid also using a too high image compression what should lose image quality and prevent any later optimization. All these parameters and many others (focus, gain, luminosity, etc) can be set up via the software driving the camera.
For DSLR and compacts with video capabilities it is a bit simpler and settings are usually limited to the selection of the format and resolution, the other settings being set automatically (white balance, sensitivity, light, etc)
At last, if you work with an analog camera, you can digitize the film using a video digitizer or "frame grabber". Matrox among other manufacturers provides various performing interfaces (1750$ for Matrox Radient eCL). Now your film can be read by any good image processing software and you can apply to it the entire range of image-enhancement techniques to improve its quality and even convert it in other formats.
By way of conclusion, if the digital camera or the still camera and the scope allow to record the image, it is still the image processing work after the shot (postprocessing) that will reveal the amateur know-how. Documents display here are very nice examples.