ISO 1600 not as sensitive as hoped?

[Keith_M42]

I posted this on the G series section, but thought maybe someone here might have an insight or thought...

I just purchased a Digital Rebel 300D. I'm an astrophotographer and I was eagerly anticipating what that 1600 ISO setting could capture vrs the 400 ISO of my Canon G2. A camera 4 times more sensitive is the equivalent of a scope 4 times bigger, and a lot cheaper than a bigger scope. Not only that, but the Rebel has a reputation for low long exposure noise. I expected a 4 fold increase in sensitivity. It just didn't happen!! At 1600 ISO and 30 seconds exposure (twice the exposure time of my G2) I got a LESS image of the faint objects I'm after than I do with my G2 at 400 ISO with only 15 seconds exposure!? To say the least it was a surprise and great disapointment. I don't know why this is the case?

I'm spectulating, but I wonder if Canon has simply raised the threshold above which photons accumulate or record into the pixels, and that's why it has the low noise reputation, and partly why I'm seeing this surprising result?

For the techies who might be able to shed some light (no pun intended, but appropo) on this, so far I have only used the jpg mode with the 300D, but that's all I ever use with the G2. Another tech point is that due to a difference in the optical path an equivalent image on the 300D is spread over 2.6 times more pixels than on the G2. If the pixel size is the same, that would account for that amount of light reduction per pixel, but not all of the 8 fold difference that 4x more sensitivity and 2x longer exposure should give. What is the area of 1mp of G2 vrs 2.6mp of the 300D?

Not all is a loss here. The 300D can take longer than 30 second exposures, and the results of 2 to 5min exposures look like they'll be worth every penny I paid, - way beyond anything my G2 can possibly get. But that cost me another $500 for an equatorial wedge, and a whole new learning curve for getting the scope to track accurately for that kind of time.

I wonder how this observation translates to standard low light point and shoot type pics? Given the same F ratio, Is ISO 1600 vrs 400 ISO a legitimate comparison then? I haven't had a chance to check this out.

Keith

[dn7elson]

Light received is a function of sensitivity (ISO setting) and lens opening. If you are using the "kit" lens with the DRebel, you do not have a fast lens to start with as it is an 18-55 f3.5/5.6 and at full zoom is f5.6.

If you were to use the 50 f1.8 lens in its place you would allow much more light to reach the sensor over the same exposure time.

The Canon G2 is an f2.0/2.5 lens. This supports your findings, but has nothing to do with the DRebel itself, but rather, lens selection. Hope this helps.

[Keith_M42]

I wish differences in the camera lens F stops were the answer. Sigh. Although I could set up the Rebel to take a picture through the telescope's eyepiece, focussed with the camera's lens, - like I must do with the G2 - but the big advantage of an SLR is that I can dispense with all that glass and let the scope itself be the camera lens. I take off the 300D's lens and remove the telescopes eyepiece and mount the camera directly to the scope so that the focus point of the mirror falls directly onto the camera's chip. The F stop for either setup, G2 or 300D, effectively starts with the fixed F ratio of the scope. The 300D is about as wide open as it can get, limited only by the scope.

Keith

[dn7elson]

I wish differences in the camera lens F stops were the answer. Sigh. Although I could set up the Rebel to take a picture through the telescope's eyepiece, focussed with the camera's lens, - like I must do with the G2 - but the big advantage of an SLR is that I can dispense with all that glass and let the scope itself be the camera lens. I take off the 300D's lens and remove the telescopes eyepiece and mount the camera directly to the scope so that the focus point of the mirror falls directly onto the camera's chip. The F stop for either setup, G2 or 300D, effectively starts with the fixed F ratio of the scope. The 300D is about as wide open as it can get, limited only by the scope.

Keith

I assume that the adaptor you use to mate the DRebel to the scope allows the scope image to be focused on the sensor...otherwise any number of issues could arise.

[dtrayers]

The 300D has a 22.7 x 15.1 mm CMOS sensor at 6.1 MPixels, and the G2 has a 7.2 x 5.3 mm CCD sensor at 4.1 Mpixels. I'll leave it to you to do the math. There's a lot more surface area on the 300D, and the pixel size is larger.

What is the focal length and diameter of the scope you're using? I have a smaller Mak on an equitorial without a clock drive and I tried some shots of the moon when I had a G3. Even 10 second exposures had motion blur. I would think that at 30 seconds a lot of the fainter objects are just smearing due to the apparent motion.

You can get a basic EQ mount with a drive for about $100US from Orion Telescopes (http://www.telescope.com).

[Guillermo Freige]

Keith:
ISO 1600 in the 300d is a true 1600, but canon digicams are actually underrated. ISO 400 is more like ISO 640-800 in the G2 (as ISO 50 is more like ISO 80-100), so the actual difference is between 1 and 1 1/3 stops, not 2 stops. My S50 at ISO 50 and the Drebel at ISO 100, using the same aperture, using the same shutter speed, have similar exposures.

[nikee]

take a shot without lens (but cap on the cam) at ISO 1600, 300 seconds (bulb mode), note the ambient temperature and the serial number of your 300d. then cut a 200x200 pixel part from the center of the resulting image.

compare this to these samples:

http://www.sternwarte-brennerpass.de/10d/darkvergleich/darkvergleich.html

if you think that your 300d has really a lot hot pixels comparing these samples with yours, canon will maybe replace the CMOS sensor.

you may send your sample with all infos to the guy operating the website above.

regards
nicola

[Guillermo Freige]

take a shot without lens (but cap on the cam) at ISO 1600, 300 seconds (bulb mode), note the ambient temperature and the serial number of your 300d. then cut a 200x200 pixel part from the center of the resulting image.

compare this to these samples:

http://www.sternwarte-brennerpass.de/10d/darkvergleich/darkvergleich.html

if you think that your 300d has really a lot hot pixels comparing these samples with yours, canon will maybe replace the CMOS sensor.

you may send your sample with all infos to the guy operating the website above.

regards
nicola

I don't think it will work in the 300d. Apparently the camera shutdown the CMOS when no light is captured, to keep noise levels low. This doesn´t happen with the 10d. I've read some complaints of astrophotographers because the 300d inhability to take a dark frame reference picture because of this fact.

[nikee]

I don't think it will work in the 300d. Apparently the camera shutdown the CMOS when no light is captured, to keep noise levels low. This doesn´t happen with the 10d. I've read some complaints of astrophotographers because the 300d inhability to take a dark frame reference picture because of this fact.

I would find this very strange - but it would confirm the crippled nature of the 300D once again... :(

regards
nicola

[Laziferous]

I don't know if the Drebel has it, but on the 10D you can turn on ISO expansion, and pump it up to 3200. Ends up looking like an ISO400 shot from a G-series, in regards to noise (actually, maybe a tad better). Again, not sure the Drebel has that option or not, but it's worth checking out to see.

[dtrayers]

I don't think it will work in the 300d. Apparently the camera shutdown the CMOS when no light is captured, to keep noise levels low. This doesn´t happen with the 10d. I've read some complaints of astrophotographers because the 300d inhability to take a dark frame reference picture because of this fact.

I would find this very strange - but it would confirm the crippled nature of the 300D once again... :(

regards
nicola

I don't think this is the case. I just shot two exposures at 30 seconds, one at ISO 100 and the other at 1600. The 1600 shot clearly has a lot of hot pixels. The most telling is that when converted with BB to 100% quality JPG, the 100 ISO image is 777KB while the 1600 is 3,689KB. There is a lot more noise in the 1600. If the CMOS was shut down, then I would expect the images to be similar size.

Here are the full size images. Feel free to download them. Out of respect to our dialup friends I didn't post the images.

ISO100
http://home.comcast.net/~dtrayers/photos/crw_3907.jpg

ISO1600
http://home.comcast.net/~dtrayers/photos/crw_3908.jpg

[nikee]

I'll upload them in a little while and post links.

oops - it's not ISO 1600 but ISO 800 those shots on this website are.

http://www.sternwarte-brennerpass.de/10d/darkvergleich/darkvergleich.html

so you are not too sad when comparing with others.

regards
nicola

[Guillermo Freige]

Oops. Sorry, my mistake. It happens when you talk without testing the fact by yourself :(
Laz, the Drebel has no ISO expansion, but you can simulate it shooting RAW at 1600 and just pumping the exposure using a +1 exp.compensation during conversion (or a +2 for an ISO 6400 shot, but noise is awful).

[timmyquest]

I don't know if the Drebel has it, but on the 10D you can turn on ISO expansion, and pump it up to 3200. Ends up looking like an ISO400 shot from a G-series, in regards to noise (actually, maybe a tad better). Again, not sure the Drebel has that option or not, but it's worth checking out to see.

Nope ;-)

[Keith_M42]

Thank you for the discussion above. This is starting to make sense. Doing the math on the sensor size from the info above and number of pixels per sensor, the pixels of the 300D are an incredible SIX times (almost) larger each than those of my G2. The light gathered by my (8" SCT) scope is spread out over six times the area on the 300D for the same number of pixels of the G2. So the light is 6 times dimmer on any square millimeter of sensor. Not only that, but due to the difference in how I use the G2 vrs the 300D, I further spread the light from a specific subject over 2.6 times more pixels of the 300D. Multiplying the 2 factors together says that the intensity of the light falling on a given area of the 300D is about 1/16th of the same area of G2. This is close to matching the results I’m getting. What I’m finding is that I have to set the 300D at 1600 ISO (4 times more sensitive) and for about 60 seconds (4 times longer) to gather about the same image.

Ah well,…. Believe it or not, I’m still happy. I’ve been able to take 2-5 minute pictures with Wonderful results. Beyond anything my G2 can get. In addition, because I can put the camera at the prime focus of the telescope, the images are clearer and more contrasty. Also the images are nearly flat or in focus edge to center to edge. All that said, I appreciate much more what the G2 can do.

Keith

[robertwgross]

Modern electronics... isn't it wonderful?

---Bob Gross---