Welcome, Anonymous (Profile, Private Messages)
Having trouble signing into the forums? Please [ logout ] and log back in.  Log inLog in 
SEARCH:     » Advanced Search

Shutterstock Photographer Forum Forum Index : Critique / Tips / Tricks :
Multiple Exposures to eliminate noise
 Page Previous  1, 2, 3  Next
 
Post new topic    Reply to topic
Author Message
ajancso


Joined: 18 May 2009
Posts: 1891
Location: Right Behind You
Post Posted: Sat Aug 20, 2011 9:56 am     Reply with quote

This is a 100% crop of one image. Clearly a lot of noise.

1 image 100 percent forum.jpg
 Description:
 Filesize:  140.79 KB
 Viewed:  1093 Time(s)

1 image 100 percent forum.jpg


Last edited by ajancso on Sat Aug 20, 2011 10:09 am; edited 1 time in total
ajancso


Joined: 18 May 2009
Posts: 1891
Location: Right Behind You
Post Posted: Sat Aug 20, 2011 10:01 am     Reply with quote

This is the same crop with 3 images stacked and no noise reduction software. You can already see a significant improvement.

3 images 100 percent forum.jpg
 Description:
 Filesize:  162.86 KB
 Viewed:  1086 Time(s)

3 images 100 percent forum.jpg


Last edited by ajancso on Sat Aug 20, 2011 10:10 am; edited 1 time in total
ajancso


Joined: 18 May 2009
Posts: 1891
Location: Right Behind You
Post Posted: Sat Aug 20, 2011 10:04 am     Reply with quote

This one is 10 stacked images at 200%.

Even at 200% it looks reasonable.

10 images 200 percent forum.jpg
 Description:
 Filesize:  182.18 KB
 Viewed:  1082 Time(s)

10 images 200 percent forum.jpg
ajancso


Joined: 18 May 2009
Posts: 1891
Location: Right Behind You
Post Posted: Sat Aug 20, 2011 10:06 am     Reply with quote

Finally the same 100% crop with Nik denoise:

First a single image plus denoise and then the stacked 10 images plus denoise.

1 image 100 percent nik denoise forum.jpg
 Description:
 Filesize:  163.68 KB
 Viewed:  1075 Time(s)

1 image 100 percent nik denoise forum.jpg
10 images 100 percent plus nik denoise forum.jpg
 Description:
 Filesize:  125.42 KB
 Viewed:  1075 Time(s)

10 images 100 percent plus nik denoise forum.jpg
ajancso


Joined: 18 May 2009
Posts: 1891
Location: Right Behind You
Post Posted: Sat Aug 20, 2011 10:08 am     Reply with quote

Clearly, stacking images really gets rid of a significant amount of noise. Stacking plus de-noise software results in a very good quality image, even at very high ISO.
northumbria


Joined: 26 Mar 2011
Posts: 494
Location: The Netherlands (Europe)
Post Posted: Sat Aug 20, 2011 2:17 pm     Reply with quote

Very clear example. Thank you. Good to know higher ISO is not always hopeless.
vanhart


Joined: 21 Jul 2006
Posts: 2380
Location: Houston
Post Posted: Sat Aug 20, 2011 3:15 pm     Reply with quote

Actually you would be enhancing the signal to noise ratio. The signal keeps adding up while the noise is random.

Now the problem comes that (if I remember right from my college days) the signal to noise ration will get better by the square root of N. Where N is the number of images. The consequence is that if the noise levels are really nigh, you will have to take an enormous number of images.

Would really love to see your experiment.... Suggest that you show the results of averaging 1, 4, 9, 16, 25 images. You will never see any difference between 14 and 15 images.
diro


Joined: 10 Sep 2009
Posts: 153
Post Posted: Sat Aug 20, 2011 3:18 pm     Reply with quote

Great examples. Thank you for sharing. I will have to try this next time, maybe on star fields, but I'm curious if I could the earth's rotation would prevent me from stacking the images properly. I guess I will have to find out.
bsluijk


Joined: 08 Jul 2011
Posts: 123
Location: Netherlands
Post Posted: Sat Aug 20, 2011 3:56 pm     Reply with quote

ajancso wrote:
bsluijk, thank you for the experiment.
Looks like this method is worth a try. The difference between the single image and the 10 images is tremendous.
The only thing I don't think you did correctly is the opacity level for each layer.
...
I would be interested if using this method you would see further improvement in your example.

Thanks for sharing your data - you did the ultimate experiment, including the effect of stacking on your image quality. It clearly proves the benefits of this method.

To follow up on your remark: yes, I incorrectly applied the same opacity to all 10 layers. I re-did my experiment and it pays to do this correctly, because the noise is further reduced.

For the nerds interested in this (I was!) I've also included the standard deviation (a measure correlating to the noise level, indicated by SD ) in these graphs.

Noise2.jpg
 Description:
Second test, now with correct stacking algorithm as mentioned by ajancso. SD = Standard Deviation, a measure for the noise
 Filesize:  171.09 KB
 Viewed:  1028 Time(s)

Noise2.jpg
khaledalkassem


Joined: 27 May 2011
Posts: 79
Post Posted: Sun Aug 21, 2011 6:08 am     Reply with quote

LIKE!

Here's my try:

10 images shot at ISO 12800.

No noise reduction applied.

before.jpg
 Description:
before- 100% crop
 Filesize:  155.17 KB
 Viewed:  989 Time(s)

before.jpg
after.jpg
 Description:
after- 100% crop
 Filesize:  113.02 KB
 Viewed:  989 Time(s)

after.jpg


Last edited by khaledalkassem on Sun Aug 21, 2011 6:27 am; edited 1 time in total
northumbria


Joined: 26 Mar 2011
Posts: 494
Location: The Netherlands (Europe)
Post Posted: Sun Aug 21, 2011 6:25 am     Reply with quote

khaledalkassem wrote:
10 images shot at ISO 12800.

Amazing!

So if applied at night outdoors, I would have to get 10 subsequent images without any movement.
Maybe the best way is to use continuous shooting, this will flipper the mirror but if e.g. there are clouds, they won't freeze for me, so the intervals should not be to long. Or is this foolish?
khaledalkassem


Joined: 27 May 2011
Posts: 79
Post Posted: Sun Aug 21, 2011 6:34 am     Reply with quote

northumbria wrote:
khaledalkassem wrote:
10 images shot at ISO 12800.

Amazing!

So if applied at night outdoors, I would have to get 10 subsequent images without any movement.
Maybe the best way is to use continuous shooting, this will flipper the mirror but if e.g. there are clouds, they won't freeze for me, so the intervals should not be to long. Or is this foolish?


Continuous shooting of 10 images outside at very high ISO like this shouldn't take much long, so I assume the clouds would stay in place in that time. Be sure to put on Manual mode so that all of your images have the same settings. And take a lot more than just 10 successive photos, then choose 10 out of them to be on the safe side.
khaledalkassem


Joined: 27 May 2011
Posts: 79
Post Posted: Sun Aug 21, 2011 9:42 am     Reply with quote

Another experiment: Comparison of stacked ISO 12800 with ISO 100

ISO 12800 single.jpg
 Description:
ISO 12800 single
 Filesize:  164.42 KB
 Viewed:  962 Time(s)

ISO 12800 single.jpg
ISO 12800 stacked.jpg
 Description:
ISO 12800 stacked
 Filesize:  121.85 KB
 Viewed:  962 Time(s)

ISO 12800 stacked.jpg
ISO 100 single.jpg
 Description:
ISO 100 single
 Filesize:  115.76 KB
 Viewed:  962 Time(s)

ISO 100 single.jpg
semmickphoto


Joined: 12 Feb 2012
Posts: 6518
Location: Stuck between a shutter and a hard place
Post Posted: Mon Mar 05, 2012 2:45 pm     Reply with quote

It looks like the photos with noise are sharp and the ones after stacking are soft.
geoffwnz


Joined: 10 Feb 2012
Posts: 174
Location: Wellington, New Zealand
Post Posted: Mon Mar 05, 2012 3:18 pm     Reply with quote

semmickphoto wrote:
It looks like the photos with noise are sharp and the ones after stacking are soft.

Possibly due to small movements of the camera from shutter and/or mirror meaning each shot could be just slightly moved.
 
Display posts from previous:   
Post new topic   Reply to topic Page 2 of 3 All times are GMT - 5 Hours
Page Previous  1, 2, 3  Next

 

Contact Us
Terms of Service
Privacy Policy
Tell a Friend!
Submit Referer Program
Shutterbuzz



Shutterstock Forum for Contributors