DPReview.com is closing April 10th - Find out more

Canon EOS R3 Baked in Raw Noise Reduction Revisited

Started Jan 21, 2022 | Discussions thread
J A C S
J A C S Forum Pro • Posts: 20,521
Re: Canon EOS R3 Baked in Raw Noise Reduction Revisited

TukTuk wrote:

attempt to summarize

1) was there "signal processing" (which I guess everybody agrees is NR) or not (not amount, just YES or NO) - my "hoi polloi" understanding at this stage is : BClaff says - always is, JACS say - no, not always (again - let us drop injecting adjectives here - just YES or NO) ... that is probably between them two and easy to resolve for as long as they both have the same raw files, right ?

I never said no. In fact, in the other thread, I gave evidence of a very clear measurable effect at ISO 100 which diminishes fast for higher ISOs, even at ISO 200. I even posted plots.

I am not familiar much with the engineering part to be able to tell what part of this is processing, what is not. I can see signs of, I guess, NR along some columns at higher ISOs but it is really small. Not along rows. I do not know if this is really NR. It could be cross-talk or who knows what. There is something for sure but it is barely measurable, forget about visible.

2) can "we" measure that amount ( in situation when we do not have access to raw with/without alleged NR ) without using adjectives but using some proper metric vs a nominal ISO (or some other parameters - sensor temp, etc, etc if Canon account for that) ?

We can make some guess. Mine is about 0.5 stop at ISO 100 which drops faster after that. Again, one number would not describe it fully.

3) if it is present (YES/NO) and regardless of the numeric (vs adjectives) "measurements" (if somebody can measure or estimate) - does it really affect the end users when they (we) are NOT able to directly compare raw conversion/post processing of raw files with vs w/o that NR... ideally w/o resorting to "I do not see it" or "it looks like it is very minimal"... it is not "star eating" - so is there a direct test possible ?

I do not know what type it is. The FT, the auto-correlation, and the histogram do not give me enough clues.  Does it affect the end-result? I guess so, otherwise it would not be there but not necessarily the way it is perceived here. I doubt that anybody can demonstrate a visible softening effect. It might be helping with the color cast in the extreme shadows. If you never dig there, you would not see a difference.

A test - I have one in mind - to take a well exposed file of the test scene, simulate lower exposure without the NR; then with it, etc. That would require some time which I do not have now.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow