Jupiter: DBK or Canon 600D?

[MarkS]

Which is better for planetary imaging, the DMK (colour version) or the Canon 600D?

Alternatively, has anyone come across any data for the read noise of Philips SPC 880/900 webcams or DMK cameras?

It would be interesting to compare them to the Canon 600D in HD video mode.  In one of its video modes, the Canon allows the 1920x1080 central portion of the sensor to be selected for creating HD videos - very tempting for capturing Jupiter and its moons.

Otherwise I'm going have to do my own experiments  

First I need to illuminate a test target to the same surface brightness as Jupiter.  So how bright is that.  Any ideas?

Mark

[The Thing]

APT has a planetary function that will use the HD video mode if available to capture sequences of images. With the 1000D it uses Liveview to capture JPEGs. Not tried it in anger as the image scale is small - I haven't the bits to put a barlow in the image train with the DSLR.

[JonH]

I came across a really clever bit of free Russian software that lets you create a video file from any Canon DSLR that has live view. Weather it has a video mode or not!
I tried it out on the moon at the weekend and worked great. But unfortunatly my Barlow is rubbish and couldn't get a good focus.

[The Thing]

I've got that as well.

My recent experience with planetary imaging tells me frame rate is most important as the seeing is the biggest factor in capturing some sharp frames, rapid short exposures are what freeze those elusive moments of good seeing for the software to identify and use.

Spc900 frame compression doesn't really make a difference, I've been using 25GPS.

[MarkS]

Just done some quick "back of envelope" calculations:

The moon is in direct sunlight so using the "Sunny 16" rule, at F/16 it will be correctly exposed if the the shutter speed matches the ISO.
So F/16 ISO1600 1/1600s should work for the moon
Jupiter is 5x further from the sun than the moon is, so it's surface brightness will be 25x dimmer
so F/16 ISO1600 1/64s should work for Jupiter (25x slower shutter)

But typically we use F/32 for Jupiter to bring out maximum detail
Hence:
F/32 ISO1600 1/16s should also correctly expose Jupiter.

On the Canon, a "correct" exposure is probably around 2/3 the maximum ADU of 16000 i.e. 10000
Using the known gain of 0.15 e/ADU at ISO1600 this equates to around 1500 electrons/pixel

But at 25 frames per second, exposure time is probably 1/50s instead of 1/16s so we only get 500 electrons/pixel.  The photon shot noise is the square root of this i.e.  around 22 electrons.

The darker parts of Jupiter might have only 25% of the light intensity which would then give a photon shot noise of 10 electrons.  This is much larger than the 3 electron read noise of the Canon.  But a webcam might easily have a read noise approaching or exceeding 10 electrons so it becomes a significant factor.

Well worth doing a few tests to validate the above ...

Mark

[MarkS]

So I gave this a try last night - a cloudy night experiment!  

My calculations suggested that the surface brightness of Jupiter is about the same as a laptop screen with brightness dimmed down to the level where it's comfortable to use under artificial light at night.  So I displayed an image of Jupiter on it and in turn pointed the SPC900, the DBK 21 (USB) and the Canon 600D at this target using the same lens at F/22.

The SPC900 would only run at a max of 15 frames per second so I also ran the DBK at 15FPS but the Canon ran at 25fps.  There was very little to choose between the image quality of the SPC900 and the DBK 21.  The SPC900 seemed to give more accurate colour - I just couldn't get the colour looking good on the DBK 21.  The noise on the DBK 21 looked much more random and easier to remove - I would describe it as "speckley".  In contrast, the noise on the SPC was very "lumpy".  However, I'm now beginning to doubt if the DBK is going to offer me much (if any) improvement over the webcam.

The Canon was quite disappointing.  Although the LiveView on the LCD looked much cleaner than either the SPC or the DBK, once a MOV HD video file was created, lots of horrible noise and artifacts appeared.  However, a single 1/50sec single exposure "photo" looked very clean.  So I think the MOV file is being over compressed - a 10sec MOV file (1920x1080x25fps) was roughly the same storage size as the 10sec (640x480x15fps) files coming from the SPC and DBK.  Unfortunately the Canon gives no control over the bit rate used for video.  I might try playing with the Magic Lantern "hack" for the Canon because this allows control over both the frame rate and the bit rate.  It also displays histograms during recording - very useful.

Mark

[RobertM]

Very interesting.  The frame rate is important for capturing a large number of frames but what about the single frame exposures for each - were they comparable ?

I'm missing having a planetary camera now

[MarkS]

I'm missing having a planetary camera now

I thought you bought a job lot of SPC900 webcams a couple of years ago?

[MarkS]

Here are some images from the test - all taken under identical conditions:

One frame from the SPC900 at 15fps:


Note the annoying red hot pixel - does anyone want a 2nd hand SPC900  

One frame from the DBK 21 at 15fps:



Here is one frame (cropped) from the Canon 600d HD MOV file at 25fps:
http://www.markshelley.co.uk/webdisk/jupexpt_canon600d_oneframe.jpg

Here is a single 1/30s exposure (cropped) from the Canon 600d:
http://www.markshelley.co.uk/webdisk/jupexpt_canon600d_singleexp.jpg

One puzzle is that the frame from the MOV file is a slightly different scale to the single exposure.  This tells me that most of the reviews of the camera are wrong when they say the 3x digital zoom takes the central 1920x1080 section of the sensor.  It isn't - it is doing some kind of digital interpolation - it really is a 3x digital zoom!

Note the hot pixels in both the movie frame and the single exposure. This sensor clearly has a big variability in pixels.

In conclusion I would say the DBK 21 is coming out the best.  The Canon 600D comes out the worst but I think it would be very suitable for a brighter object (the moon?) where the larger number of pixels would be useful.

Mark

[Rocket Pooch]

Makes me smile your 1 frame from the DMK is better than most professional images pre 1995

I hope you tickled you scope under the chin and said "who's a wittle C11, good boy".

[MarkS]

Just to clarify - these are all images of a computer screen through a lens.  No telescopes were used (or harmed) in the making of these test images ...

[JonH]

 
I'm pretty sure my wallet just gave me a nervous look!

[JohnP]

Mark the detail in all those Jupiter images is amazing... Can't wait to see a complete fully processed image with any of the camera's... John

[Fay]

did you use the Bananascope for those Mark?

[MarkS]

Mark the detail in all those Jupiter images is amazing...

Well it just goes to show what a difference the wobbly atmosphere makes!  Remember these are images of a laptop screen from across the hall.  

The test was to compare the 3 cameras under controlled conditions.

did you use the Bananascope for those Mark?

No -I used a Nikon 300mm lens stopped down to F/22.