Canon 450D Noise Profile

Post by **weixing** » Fri Jan 29, 2010 11:11 pm

Hi,
Nothing to do, so took sometime to do a quick noise profile for my Canon 450D DSLR. Posted the result here for those who might be interested.

Anyway, I took all the shots at night in my room, so the temperature might be slightly higher than outdoor. Also, the DSLR is fully rested before took all the test shots. Below are the procedures:
1. Took 60 shots of a specific ISO with 60s exposure time at 10s interval to simulate normal shooting of light frames.
2. Took another 25 shots of the same specific ISO with 60s exposure time at 10s interval to simulate shooting of dark frames.
3. Median the 25 dark frames together to get the master dark frame of that ISO.
4. Measure the thermal noise on every of the 60 light frames before and after master dark frame subtraction.
5. Median stack the 60 light frames together to simulate normal stacking without dark frame subtraction. Measure the result.
6. Median stack the 60 light frames after master dark frame subtraction together to simulate normal stacking with dark frame subtraction. Measure the result.

Below are the graph of the result:
1. Noise profile before noise reduction (60 x 60s exposure)

2. Noise profile after noise reduction (60 x 60s exposure)

3. Noise profile after stacking (60 x 60s exposure)

Have a nice day.

Tachyon · Post by **Tachyon** » Sat Jan 30, 2010 5:43 am

Hi WX! Very free har? Maybe you can do the profile for my cameras... heh.

By the way, how did you measure the thermal noise values?

If I read your charts correctly:

1. With each successive shot, the noise level increases (probably due to internal heat?)

2. Dark frame subtraction eliminates the 'noise creep' but does not help in ISO-related noise.

3. Recommend to stack after noise reduction.

Am I interpreting it correctly?

Post by **weixing** » Sat Jan 30, 2010 2:51 pm

Hi,

Tachyon wrote:Hi WX! Very free har? Maybe you can do the profile for my cameras... heh.

I'm quite free sometime... but not that free leh... ha ha ha

Tachyon wrote: By the way, how did you measure the thermal noise values?

Use the dark frame lor... since there is no light, whatever is there are bias, thermal noise and random noise. All the frames I took are dark frames... I just use the first 60 shots to simulate the noise in my "light frames" and the last 25 shots as dark frames. Also, my frame are not thermal noise, it's the combination of all the noise. It's doesn't matter in this case, because all my "light frames" in my shots had the same exposure time, so I don't need to create thermal frames. Anyway, since I still had all the RAW files, I'll create a thermal noise chart, but I expect the shape of the chart will be very similar.

Tachyon wrote: 1. With each successive shot, the noise level increases (probably due to internal heat?)

Yes... as the heat built up in the camera, the noise increase.

Tachyon wrote: 2. Dark frame subtraction eliminates the 'noise creep' but does not help in ISO-related noise.

Dark frame subtraction eliminates the noise generate by the dark current (thermal noise), but not the random noise. The noise in high ISO seem higher because high ISO is done by amplify the signal which also amplify the random noise.

Tachyon wrote: 3. Recommend to stack after noise reduction.

Yes, it's show that stacking does reduce the random noise.

Anyway, from this experiment, it's does show me that the dark frame noise reduction does work.

Have a nice day.