SCNR with video for question

[ngc1535]

Hi,

My previous post on SCNR didn't get a response... but perhaps this one is easier again with a video.
The Average Neutral protection doesn't appear to always take the average of the red and blue channels when this is smaller than the current green value.

Please see this video:

https://youtu.be/9o1aWd7MsVc

Perhaps there is a truncation or the algorithm as changed? (or of course I am not fundamentally understanding it)

Thanks in advance.
-adam

[Alejandro Tombolini]

Hi Adam,

I don't know how the tool works but it seems to happen only in that pixel, did you notice something similar in other bright green pixels on the same image?
I tested in an image and could not reproduce it in any pixel.

Saludos, Alejandro.

[ngc1535]

Thanks ALejandro,

No... I couldn't find another pixel... but I didn't search too hard. I was just surprised.
Earlier today I got another response that indicated the issue could be due to the fact the images were represented as 16bits.
So it could be that the fact I brought it old data (before a time I was using PI) could be the source of the issue. I am not certain.
It doesn't entirely make sense to me that some pixels the math works and some it does (even a single case... computers are deterministic afterall).

But I now feel more comfortable since it appears it could be a bit space/rounding issue.

-adam