EOS Digital Rebel - Astrophotography problem - Darkframe

[vwestervelt]

Just purchased the New Digital Rebel. I've been involved in astrophotography for quite some time, I was anxious to try this camera with my setup.

For those that know, I have a Celestron Nextstar GPS telescope with Faststar lens assembly that drops the 2000+mm f10 down to an F2 (you'll notice the spherical abberation in the included photo...it's normal for this lens, the desired image area is an area in focus that is easily recognizable).

Problem....in Digital astrophotography long exposure (the included shot is 30sec @ ISO 1600) it's normal to take 2 images, one of the target, and a blackframe (lens cover on) to capture noise generated by blooming and digitally subtract the noise from the target image. This increases contrast and greatly enhances the image.

The problem is, I can't get this camera to generate a suitable darkframe. No matter how long I expose w/the lens cover on, the image is flat black. I've been an electronics engineer for 17 years, it's hard for me to believe the ccd array won't generate any noise in a 5 minute exposure (although I admit, for the raw frame that I included, noise generation in the image is as low for any ccd I've ever seen in this application from a commercial camera not designed for astrophotography...I'm very pleased with that!). It's likely that software cancels noise when it thinks there is no image.

Without the Darkframe, I'm unable to functionally use this camera for this application as I can get much better end results from other ccd imagers. While I'm VERY excited about the raw images this camera has given me initially....I really need a way to get a dark frame. I'm not an expert in photography, but I've been doing astrophotography w/digital for several years.

If anyone knows a tip or trick that may help in this, I would be very GRATEFUL!

Image : M31 - Andromeda Galaxy (and 2 small neighbors)

Exposure : 30 sec | ISO 1600 | F1.98 - Nextstar GPS w/Fastar

Untouched photo

Thanks in advance I hope!

Vern Westervelt (just realized I can't attach photo...email me for it if you have any ideas...Thanks!

[tony723]

Canon 300D should generate noise for long exposure. I tried it and you may see some sample photos here:

http://www.fotop.net/hktraveller/300D_hotpixel

[vwestervelt]

I've done 5 minutes on bulb setting w/lens cap w/out a spec of noise build up at ISO 1600.

It's got to be the camera processing it out....

is there a way to post images here? or do you have to have your own web page.

Vern

[robertwgross]

vwestervelt wrote:
is there a way to post images here? or do you have to have your own web page.


Yes to both questions.

You post the image link to your own web address.

---Bob Gross---

[PacAce]

vwestervelt wrote:
I've done 5 minutes on bulb setting w/lens cap w/out a spec of noise build up at ISO 1600.

It's got to be the camera processing it out....

is there a way to post images here? or do you have to have your own web page.

Vern

I guess you can consider yourself very lucky that you're not getting any noise in your long exposures. :)

I'm not an expert on this but I read someplace that the Canon cameras does its own internal processing to eliminate the noise that's inherent in long exposures. After the "normal" exposure is made, another "dark" exposure is made to use as a "mask" to cancel out the noise that was generated in the "normal" exposure. This may explain why you're getting a "perfectly" black frame with the lens cap on.

BTW, Canon DSLRs use CMOS sensors, not CCD sensors.

[tony723]

vwestervelt wrote:
I've done 5 minutes on bulb setting w/lens cap w/out a spec of noise build up at ISO 1600.

It's got to be the camera processing it out....

is there a way to post images here? or do you have to have your own web page.

Vern
What is the aperture you are using? You may try F5.6, ISO1600, 1" and I think you should get noise generated.

[PacAce]

tony723 wrote:
vwestervelt wrote:
I've done 5 minutes on bulb setting w/lens cap w/out a spec of noise build up at ISO 1600.

It's got to be the camera processing it out....

is there a way to post images here? or do you have to have your own web page.

Vern
What is the aperture you are using? You may try F5.6, ISO1600, 1" and I think you should get noise generated.


Errr, if the lens cap is on, does it really matter what aperture the camera is set to???

[tony723]

PacAce wrote:
tony723 wrote:
vwestervelt wrote:
I've done 5 minutes on bulb setting w/lens cap w/out a spec of noise build up at ISO 1600.

It's got to be the camera processing it out....

is there a way to post images here? or do you have to have your own web page.

Vern
What is the aperture you are using? You may try F5.6, ISO1600, 1" and I think you should get noise generated.


Errr, if the lens cap is on, does it really matter what aperture the camera is set to???

I am not sure but I did try it using this aperture.

[PacAce]

tony723 wrote:
PacAce wrote:
Errr, if the lens cap is on, does it really matter what aperture the camera is set to???

I am not sure but I did try it using this aperture.


I was being facetious. If the lens cap is on, it really doesn't matter what aperture the camera is set to. :)

[vwestervelt]

I did some experimenting with this last night.

With an exposure as long as 10 minutes w/lens cap on, the only noise that remains are about 5 hotpixels in a red and green variety.

There is not way, electonically speaking, this ccd array won't generate any "blooming" affect over that long of a dark frame exposure.

As soon as there is any light in the frame, you can see low level red & green blooming affect, worse near the area of light, in this case stars.

The CCD array is much more sensitive to the red & green wavelength in extremely low light conditions.

Andromeda galaxy, for example, has to be red & green level adjusted (down) for this galaxy to appear in it's true emission state....(I do this in Photoshop.)

Anyway, I'd still appreciate any ideas to be able to generate a very low noise dark frame. Shooting a short exposure axposure against a black non reflective card maybe?

Vern Westervelt

[vwestervelt]

oh yeah....I don't have a webpage...anyone have a webpage I could post the image I'm referencing?

Vern

[lightandlife]

Vern, I read your post with interest. The noise your camera generates may not be a white (meaning random unbiased) noise. If it were a white noise, such as hissing in an audio tape, it is not possible to record a music or song using an audio tape recorder and subtract the white noise in the blank tape and get a cleaner music signal.

The noise your camera generates may be deterministic, not random. Chaotic, but not random. In this case, a dark frame will contain some signals--depending on the exposure time-- the camera processor generates, which looks like independently and identically distributed noise, but not random.

My thinking is that Canon 10 designer knows it already and would simply erase this chaotic noise (as in the chaos theory) when the cap is on. (In this case, you cannot generate your desired dark frame) It would be a simple matter to detect whether lens cap is on or not.

Some less sophisticated cameras may not have cancelled this quasi noise in the dark frame. Try Nikon or Olympus and see if they can generate suitable dark frames.

[vwestervelt]

I appreciate your comments lightandlife.

I think you're exactly right, it's not random affects the software processing is targeting.

In fact, CCD imaging arrays weren't initially developed for photography....they array's were targeted at developing an "analog" cell memory array. What they found during the development was that "charge" was transferred to each element in 3 manners....1 - Electrically (i.e. whatever circuitry was designed to transfer charge from some trigger source) 2 - Thermal. Thermal energy would also transfer into electron charge in the individual cell, and 3. Light. They found that 1 photon of light would roughly transfer to 1 electron of charge for the cell. Bells went off.....ccd imaging was born.

Every CCD is more sensitive, or receptive, to this transfer at different wavelengths of light. The digital EOS, for example, is more sensitive in red-green than blue.

All things remaining constant....temperature outside, and power supplied to the chip array.....and the noise is VERY repeatable. The problem get's complicated when you add a light source. "Glow" from the light source will transfer angularly to neighboring elements at different emmisions, showing up as red/green/blue noise in a very NON-random manner, completely dependant on the light source, amount of light, etc. which the software programmers will never know because your target is likely never repeat exactly twice.

In the case of stars, stars have many different emissions, reds, yellows etc. The dark frame is mostly to help "compensate" for the noise generation that is random and not processed out by the camera.

The processing done by the camera is mostly theoretical based on known constants. Since the Random event is always your target, they can only really have software that will compensate or process out the knowns...i.e. temperature & electronic noise...and even temperature is probably done at an average...unless there is a sensor in the camera (not aware of that) On a test bench they likely operate the camera at test conditions and calculate based on a worse case scenario......anyway, the light of the target is the unknown in astrophotography. It almost always causes a blooming affect in a ccd array. The dark frame just compensates because although it's caused by different variables, it manafests itself in common patterns in the array.

Some digital cameras have the feature of taking 2 images, one of your target, one dark frame with shudder closed of the same duration (doubling the exposure time you have to wait) , then attempting the subtraction manually. (Nikon coolpix 5700 does this)

The problem again is that this only accounts for the constant variables, not the "glow" from your target. My best results have been from a non-linear dark frame. If it's a 30 second exposure of a bright core galaxy, I'll do 50-100 second dark frame. This greatly increases contrast and reduces the signal-to-noise ratio in the finished image leaves you with about 95% real structure.

Anyway......I wish I could get this dark frame with this camera.....the raw images are very impressive compared to what I've done in the past. Cannon is only about 25% of the way away from not needing a dark frame in this application.

Still open to ideas!

Thanks lightandlife!!

Vern

[tony723]

vwestervelt wrote:
oh yeah....I don't have a webpage...anyone have a webpage I could post the image I'm referencing?

Vern
You may email your photos to me (tony723@yahoo.com) and I can find a place to post it.

[jimsloy]

Vern-

What is your email address? I am, right now, in the market for a telescope to do astrophotography w/ my 10D and I can't quite decide what to get. Without going into detail here and stealing the intended subject of this thread, I'd like to discuss w/ you separately your thoughts and comments on the best equipment to use without breaking the bank.

Thanks,
Jim

[vwestervelt]

jimsloy wrote:
Vern-

What is your email address? I am, right now, in the market for a telescope to do astrophotography w/ my 10D and I can't quite decide what to get. Without going into detail here and stealing the intended subject of this thread, I'd like to discuss w/ you separately your thoughts and comments on the best equipment to use without breaking the bank.

Thanks,
Jim


Jim, I'd be happy to. If astrophotography is your goal, you have a lifetime of choices. Your smartest first step is to talk this through with someone already in it and learn from their mistakes....(I've made many). Otherwise, you can easily end up spending your money 2, 3, 4 or more times.......just ask my wife.


Vern

[vwestervelt]

Ok....Here's the picture I refer too.

It's shrunk about 60% to fit, but you can easily make out what I have referenced.

I have software that will work on some/most of the noise, but a dark frame would work best. Still open for any ideas.

Picture is at

http://www.fotop.net/hktraveller/VernWestervelt

Thank you Tony for allowing the post!!!

Vern Westervelt

[HoldenMan]

My understanding is that the 300d is designed to render a blackframe unnecessary....so maybe that's why you're having trouble

[rhuyps]

HoldenMan wrote:
My understanding is that the 300d is designed to render a blackframe unnecessary....so maybe that's why you're having trouble


Terrible news... I just was deciding to go for the cheap 300D instead of the 10d, to be used for astrophotography, ..... until I discovered the review site which clearly states the issue of darkframe:


http://www.dpreview.com/reviews/canoneos300d/page15.asp

"... Night exposures
The EOS 300D carries out long exposure noise reduction 'on the fly', this means that it unlike other digital cameras it doesn't need to take a second 'dark frame' exposure to subtract from the original shot. As you can see from the samples below the EOS 300D performs extremely well with hardly any visible noise at ISO 100 (despite it being a 30 second exposure) and noise only increasing very slightly at ISO 200. Above this noise does become more of an issue...."

So it seems bad news, I am not sure if the RAW format excludes any processing "on the fly" ....

Regards,

Ronald

[mauba58]

http://astrosurf.com/buil/us/digit/eval.htm

Go here.. but it is in French

Regards

mauba