CCD vs DSLR

[LDUNN1]

RikM - did your 2.5hrs on the M27 via the 1000d also include the time for the darks, or where they extra time - or did you not take darks?

[RikM]

RikM - did your 2.5hrs on the M27 via the 1000d also include the time for the darks, or where they extra time - or did you not take darks?

Good point no, the 2.5hr is integration time only. Add another 1.5 hrs for Darks and about 20min faffing about for flats and bias. The SX camera has no darks (didn't need them ), but does have flats and bias. Since it was a mono shot through an astronomic UHC filter for L, only one set of flats but as you say, they are quick and easy.

[LDUNN1]

........Thanks Rik - I think that gives a better comparison, I'll give you the 20mins faffing on flats & bias - you have to do similar with the mono CCD anyway. So the comparison is really 4hrs vs 45mins in your example........

[LDUNN1]

Rik, one other thing, did you shoot both of your images at the same f-ratio? if so, what was it?

.....which scope would be also interesting to know - I assume it was the same setup for both your DSLR & CCD.

For info, my image (link posted above) was at a slowwwww f9.

[RikM]

Yes, same scope each time. 150P Newtonian so f/5 for me, nice and fast

Can't remember the sub lengths but I think it was something like 70x 2min for the DSLR. The SX was 9x 5min.

[LDUNN1]

.......I recall in my early pre-guiding days, about a year ago, doing something like 80x 90 sec subs with a Canon 5d2.......which knocks out 25mg RAW files.....80x 25mg starts gobbling up harddrive space real fast! I found the saving in harddrive expenditure quickly paid for the guiding equipment as I jumped to 300s subs ;-0

[LDUNN1]

What is also interesting to me, is comparing your CCD 45min M27 @ f5 vs my CCD 1hr 40secs M27 @ f9 .......mind you, mine also includes the time for RGB - L channel was only 40 mins (4x300s + 2x600s)........so maybe more alike than I first thought.

[nytecam]

Not a very good comparison as it was only a test to see if the new (old second hand) camera worked, but here you go:

Thanks for that - indeed a great comparison eg SX image is 1/3rd the exposure of DSLR but goes as deep and sharper to boot With our fickle weather we need all the photons we can collect quickly and CCDs do the business. However it's easy to gloat but some need their DSLR for family shoots too

[RikM]

What is also interesting to me, is comparing your CCD 45min M27 @ f5 vs my CCD 1hr 40secs M27 @ f9 .......mind you, mine also includes the time for RGB - L channel was only 40 mins (4x300s + 2x600s)........so maybe more alike than I first thought.

Remember I was shooting through a UHC filter which knocks the brightness back a fair bit compared to a traditional L image, so you can't read too much into the comparison of f/9 vs f/5. My DSLR image was using just a Skywatcher LPR filter rather than the UHC.

[LDUNN1]

ahhhh - good point. I was shooting from a light polluted city without LPF!!! ;-)

[laser_jock99]

So another AP is hooked on megapixel CCDs via the DSLR and what do you do when posting DSO results on SGL - you resample to 1200 x 800p eg <1Mp which dedicated CCDs do in the first place where their larger pixels are better at collecting photons in the first place.

CCDs are not automatically H-alpha hobbled straight-out-the-box thus unfiltered and unfettered are much 'faster' resulting in briefer exposures than DSLRs for the same/similar result. You can have too many [tiny] pixels in AP. CCDs the way to go!

I think what some us really want is CCD sensitivty on a large area sensor- at DSLR pricing. To me the beauty of working with DSLR's is the large sensor, meaning you can make efficent use of the light gathering power and large field of view of many scopes. Seems a bit pointless having a 1/2" square chip sitting in the middle of a 3" focusser? Some scopes are now desigined ground up to give a flat field for a full framed 35mm CCD chip. So we just need a sensibly priced TEC cooled 35mm size chip based camera to fit in our 3" focussers. Surely this is possible for less than £20K?

[LDUNN1]

I agree!

I have wondered what performance a dedicated astro cam manufacturer could extract if they got their hands on a supply of DSLR sensors. Would it be about the same as a modded DSLR? Or would it be much better if they threw all their tricks at a DSLR sensor with the single direction of maximising astro long exposure imaging.

Mind you, I also wonder why the DSLR manufactureres have not got into active cooling of their sensors - they put so much effort into eeking out small performance gains in terms of low light / hi iso - & they rate of improvement seems to be tappering off in recent years, & we all know how much improvement you can get with just set point cooling - or shooting astro on a cold winters night vs a hot summers one, that having built in active cooling on a DSLR sensor seems a logical thing for them to introduce to steal a march on their competitors. It would probably need a custom function to turn in on & off mind, due to the increased drain on battery life, &/or have it kick in for hi iso &/or long exposure shots.

[MikeWilson]

I agree!

I have wondered what performance a dedicated astro cam manufacturer could extract if they got their hands on a supply of DSLR sensors. Would it be about the same as a modded DSLR? Or would it be much better if they threw all their tricks at a DSLR sensor with the single direction of maximising astro long exposure imaging.

Thats the QHY8L, which has a Nikon inside :-)

On the other hand, isn't that also what Canon are doing with their new 60Da?

All the best,

Mike

Sent from my mobile using TapaTalk (so please excuse bad grammar & spelling!) :-)

[nytecam]

I have wondered what performance a dedicated astro cam manufacturer could extract if they got their hands on a supply of DSLR sensors.

There are already have - try http://www.centralds.../products-x.htm

Mind you, I also wonder why the DSLR manufactureres have not got into active cooling of their sensors - they put so much effort into eeking out small performance gains in terms of low light / hi iso etc

Only a tiny, tiny fraction of the millons of pics taken with DSLRs are astro so effectively the DSLR is an incredible lo-cost bargain for the technology. Only Canon seemed aware that regular IR blocking filter killed H-alpha @ 652nm and made briefly the 20a. Coolers need energy and battery life would suffer - there's no viable market.

What would be interesting is user controlled DSLR pixel binning where the binned signal was added rather than averaged as currently - that would 'speed-up' the sensor, I think ! Current HD-video is average binned unfortunately

Astro CCD makers use sensors, including DSLR sensors with larger pixels, which are appropreate to the needs of AP. Higher pixel counts still sell DSLR to punters but aren't the answer to AP. And the dedicated Hubble Palette/ LRGB APs hereabouts need unfiltered monochrome sensors for max speed for effective photon collection

[LDUNN1]

Mike, when I wrote that the QHY8L did cross my mind.....I have only read of two people having these (you being one of them) & both sold.......curious!

The Canon 60a (& the 20a before it) are not really what I had in mind - these are basically modded DSLR's.

Nytecam - now that is a new link to me! Many thanks, although, that is not what I had in mind - the models shown in that link look like modded DSLR's mostly, & even the one that looks like a dedicated astro CCD describes itself as a modded 600D. I was thinking more if the likes of Atik or SX started with just the Sensor, & maybe chip set, & built their camera around that, what might be achieveable. Still, your link is new & interesting to me.

I agree that coolers would take a chunk of battery power from the camera, but if it was a Custom feature it could be turned on/off as necessary to boost dark noise suppression when conditions became challenging - hi iso, hi ambient temps, long exposures etc........Liveview & video eats battery life, that didn't stop 'em doing it!

If Canon were that smart, their stock IR filters would cut in more aggressively after 656nm, not lazily way before, attenuating the sensors response to Ha & near IR light......

Not so long ago, I would have agreed with you ref larger pixels being more appropriate to the needs of AP & associated comments on higher pixel counts, however, my mind has been changed over the past year on this subject. One of the more sensitive sensors out there is the Sony ICX-694 - & that has 4.5 micron photosites. The QE graph for the 694 shows it to be about a third more sensitve again to the ICX-285. Both the 285 & 694 have exceptionally good noise control too. Compare that to Kodak's 11002 sensor which has much larger 9 micron photosites with noise not as well controlled as the Sony chips. My own DSLR has 6.5 micron photosites, & yet the ICX-694 offers greater sensitivity & less noise despite it's much smaller photosites

The high pixel counts are also due to the physical size of the sensors too, not just the size of the individual photosites. As laser-Jock pointed out, why have a physically small sensor when you have a scope with a larger focuser that is able to offer a large flat, evenly illuminated field much greater than most of these small sensors........other than cost of course!

[Merlin66]

Lawrence,

the filter removed from the Canon DSLR to enhance the red response is not an IR filter. It's a colour balance filter which works with the QE curve of the chip and the De-Bayer algorithm that Canon use to give " pleasing" images.

The front Ant-alias/ dust filter (in the >400D) is actually the UV-IR cut filter (400nm and 700nm)

Also with DSLR OSC remember what you see is not what you recorded!!

If you zoom in on your RAW image you'll never see RGGB pixels....they are already "modified" by the de-bayer process........

[dph1nm]

What would be interesting is user controlled DSLR pixel binning where the binned signal was added rather than averaged as currently - that would 'speed-up' the sensor, I think !

There is no real difference between adding and averaging as far as signal-to-noise is concerned. If you mean real hardware binning of colour ccds (so the three colours are binned independently) then there are/were Kodak chips which can do this.

NigelM

[LDUNN1]

I'm reasonably familiar with Canon SLR's, I've owned & been using the EOS system since the days of the film based Eos 620 shortly after the Eos system was introduced. I currently have a 20d, modded 40d & a 5d2. It has always been my understanding that the Canon's do have a filter that blocks IR. Canon's own website backs this up as it gives information about the 60Da they state: " In a standard DSLR, an infrared (IR) filter limits the amount of Ha light that passes through to the camera sensor" source:-

http://www.canon.co.uk/About_Us/Press_Centre/Press_Releases/Consumer_News/Cameras_Accessories/Canon_unveils_the_EOS_60Da.aspx

.....they call it a 'low pass filter' not an IR blocking filter, but it's effect is the same.

I'd love to be able to remove the bayer filter off a DSLR too - to give a hi-res monochrome, but I believe this is nigh on impossible as the bayer filter is 'printed' onto the microlens filter infront of the sensor & you can't just 'remove' it.

[Merlin66]

Hmmm

According to the spectroscopic transmission curves (see attached) the filter in question is definately a colour correction filter and the remaining front filter is a UV-IR cut.

[laser_jock99]

Only Canon seemed aware that regular IR blocking filter killed H-alpha @ 652nm and made briefly the 20a.

Not strictly true there was a series of Fujifilm IR sensitive DSLR's ending with the IS Pro model (I own one) which were marketed as an astro/forensic cameras.

http://www.digitalca...dslr--13935.htm

I managed to get one brand new for £350 when Fuji decided to discontinue!

EDIT- I'm sure this camera too would benefit from being run at 30 deg C below ambient. Maybe I should rig up a Heath Robinson TEC to it as well !!

[LDUNN1]

....."low pass filter", "IR filter" or "Colour correction filter" - call it what you will, they all amount to the same thing - it is attenuating Infra red & near infra red wavelengths, & when the filter is removed, or replaced with a more agressive cut off filter cutting in after the Ha wavelength, then you get a significant boost in sensitivity to Ha wavelength light, which is what we all seek.

I agree with you that if you zoom into a RAW you don't get to see individual pixels of RGGB. My understanding is that the groups of 2x2 RGGB photosites are used to calculate the 'average' (my poor terminology) colour of any group of 4 RGGB photosites......so for example here is a 2x2 block of 4 groups of RGGB photo sites:-

RGRG

GBGB

RGRG

GBGB

the top left corner 4 can be used together, the top right 4, the bottom left & the bottom right - but also the middle 'BGGR' 4 can be calculated. I read an articule some time ago that explained that effectively the cameras algorithm can use every group of 4 to calculate the individual pixel colour that it ends up saving. Now, I don't know if this is true or not, but it would explain why when you do zoom in to a raw file, you don't see RGGB pixels together - the camera is not saving those, it's saving the combined output of all the 2x2's as a calculated single colour. It also explains why you get less than a full 4x resolution improvement between a mono & a OSC sensor that is otherwise the same - the OSC information is used to interpolate the seemingly 'missing' resolution information. Hope I explained that well enough.

[Merlin66]

The only point I make is the colour filter affects the red responce and the other filter the IR responce.

"" it is attenuating Infra red & near infra red wavelengths"" This is not correct.

Not one filter does both.

Craig Stark has a good section on his web site about differnt types of De-bayer algorithms and how the RAW image is actually built up....

[cgarry]

Its a good a camera, you need to learn the tricks of using it to take flats as its not quiet as straight forward as others (need to have a delay in-between shots).

Hi Earl,

Can you point me towards some more info on this? I have a QHY8L and I never quite got the flats working.

Cheers,

Chris

[LDUNN1]

Merlin, I really think we are in general agreement here & are talking about the same thing & have teh same understanding, but maybe our terminology is different.

There is more than 1 filter infront of the Canon sensors, typically, they have one which is a anti-alias filter whcih may also be used for the 'ultra sonic' anti-dust 'shake' filter.

Then they have an 'IR cut filter' which is two filters that are sandwiched together (Canon call these two 'Phaser-layer IN absorption' & Low-pass) - but their combined job, & they come out as one piece together as they are sandwiched togther, is to attenuate IR - but it is not a 'cliff edge' type filter - its effect is more gentle, a rolling hill that starts rolling down rather early which has a gradient between 550nm to 700nm (or there abouts) - it is blocking IR wavelengths above 700nm close to 100% (blocking), but around the 650nm mark it is blocking around 70% - hence why I said the IR filter is attenuating the IR & Near IR wavelengths of light.

Mods involve replacing this sandwiched pair with a more aggressive 'cliff edge' IR blocker - the Baader one for example starts attenuating at around 680nm & is 100% blocking by around 700nm - a much steeper gradient & crucially, it starts cutting in after the Ha wavelength & Sll.

I didn't mention the UV end, but these filters are deisgn to pass 'visible light', so they also block at the UV end of the spectrum too, but it is the IR & Near IR end of the spectrum that gets use Astro folk all excited as that is were the source for much DSO light falls.

I think we are singing from the same song sheet......

[Merlin66]

Looking at the graph for Filter #1 (message #44 above) the anti-alias gives a clean UV-IR cut off.

I question whether it's necessary to add an additional UV-IR cut replacement....