Jump to content
conrad2k

Stacking: Noise Reduction, Optimum number of images

Recommended Posts

How many images would be considered ideal to the best effect in stacking images for noise reduction?  In the tutorial, you seemed to add 20+ images.  Could you get a comparable effect with fewer images?

Share this post


Link to post
Share on other sites

Thank you! I'll play around. I am curious whether there will be differences between how it handles noise reduction on a Bayer pattern sensor (e.g., a NEF) and the Fujifilm X-Trans sensor.

Share this post


Link to post
Share on other sites

Conrad2k, your question caused me to experiment a while. A very general answer is: there is no optimum. An optimum assumes that there is a number of photos from which on the result gets worse. This is not the case. Every additional photo improves the noise reduction. But the contribution per photo gets smaller and smaller, and the calculation time gets higher and higher.

 

While adding pictures to the stack, my observations are (observed with 100%zoom):

- the first 4 pictures significantly contribute to the noise reduction improvement, wow!

- the 5.-8. picture can be considered as polishment, they raise the noise reduction of the picture to 90% of the overall  result. 

- the 9. and 10. picture only slightly contribute to improvements, i guess it is now about 98% of what you can reach.

- the 11. and 12. picture brings quite invisible improvements, its more the light than the noise that changes slightly

- the 13. and 14. picture are the last pictures where I remark any visible changes at all at 100% zoom

 

For more pictures, I have to zoom in:

- the 15.-18. picture bring, visible with 170% zoom, still slight improvements on pixel level 

 

Finally it is a trade of between quality, computational power, hard disk space and time. If you want maximum quality and you have a zoom of 150%, you see still noticeable noise reduction improvements between 14  and 18 pictures. 

 

Another perspective: 

- at the zoom level "overall picture" (cmd 0): the 11.-18. pictures do not bring remarkable improvements

- at the zoom level "100%": the 14.-18. pictures do not bring remarkable improvements

 

As you see,

- 10 pictures provide super results for the overall picture zoom,

- 14 pictures provides super results even for 100% zoom, and

- 18 pictures are for purists. 

 

Here the examples for 1, 4, 10, 14 and 18 pictures. Click on them to see them full size (each ca. 220kb)

post-23077-0-35962800-1450422511_thumb.jpg post-23077-0-96155800-1450422513_thumb.jpg post-23077-0-07791600-1450422515_thumb.jpg post-23077-0-32306400-1450422517_thumb.jpg post-23077-0-25668500-1450422519_thumb.jpg

 

In any case, thanks to the Affinity Photo development team for this powerful feature. The noise reduction is impressive.

 

Best regards

BlauerClaus

Share this post


Link to post
Share on other sites

I wonder if you could create an even more efficient stacking algorithm.

I faced the problem that without a tripod 10+ fotos in a boost weren't 100% alligned. There was a tree in the foreground and it may have moved a bit as well.

Couldn't the algorithm detect big areas that are consistent trough out the stack (the tree) and only recalculate the small areas (noisy pixels)?
Well the tree would be more noisy in that case but the balance between "exclude big areas" and "recalculate everything" could be adjusted through a slider?


 

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×