Normal and Strong v2 Model Discussion

Your image here does have a very low amout of noise, i suggest to lower the denoising to 5 or 10 with the new models and you will get natural results.

And instead of using the old sharpen filter i suggest the new strong one with low settings too, like 5 or 7.

Thanks - that’s what I am doing right now. Even though it has 800 ISO…
I have noticed that I get much better results, when I reduce the denoise effect in Capture One to 0. Seems like the two systems do not like each other very much :wink:

I like the v2 models; in most cases they produce a more natural looking effect. However, sometimes they produce artifacts, even giving a greenish tint to some previously white/pale yellow pixels. In particular, I’m working with images where there are specks of light reflected in someone’s eyes, and I’ve noticed both normal and strong v2 tint the light speck such that it looks unnatural. Sometimes I still prefer the original models’ results, and I like the option to choose between them and the v2 models.

This is a nice improvement. I’ve been editing photos take in 2005 with a Sony camera and what a difference I’m able to make with the improvements using the V2 updates. Keep up the good work!

1 Like

Not quite convinced the V2 (Beta) is any better than the Normal version. If there’s little to none noise, I find it worsenes the picture comparing both options. Not for me at this point…

Strong v2 is OK but Normal v2 has some serious issues tried over human face … Too many artefacts . Compared to old Normal - v2 is much worse !

1 Like

Hi
I strongly suggest you keep both the previous version and the V2 and we, user, get to choose, according to the picture which one to apply.
That means, you don’t remove the previous version.
Because, in many cases, the previous version is doing great, and in some cases, the new one seems to be OK.

You can just rename them.

Thanks

3 Likes

Excellent idea, hopefully they will do exactly that!
It harms noone to keep the old and the new as it is now. Why get rid of it if it’s working fine even for just a handful of users?

I am attaching 2 screenshots of the different “Remove noise” models so that you can analyze the results that in the almost V2 beta have seemed very deficient to me.
It is a very forced photo to evaluate the results.
All the best.


2 Likes

Surprisingly enough, overall, I’m getting worse results with the latest AI (worse yet with the V2) than with the Denoise…
It seems like the AI is not detecting all of the noise/blurr that use to detect, and needs much more time to fine tune it, while the Denoise is doing a faster job… Something’s not working for me with the new update(s)…
Dont’ get me wrong, I’m still satisfied with the products but (to me at least) seems in need of some tuning.

Did you try to lower the denoise slider strength of the V2 models? They should retain more details in smaller slider strength.

I tried most everything I could with all the sliders. I still find better the “old” Normal, but right now, Denoise seems to be making a slightly better job, since (IMHO) the auto-settings on IA changed in the wrong direction…

Thanks for taking time to reply to my comment!

1 Like

I’ve found the v2 version to be much better - it’s all I use. I no longer fear taking higher ISO images, because the noise reduction does not smear or spoil images.

Loving Normal V2. I’ve only used it a handful of times (with no alterations to the sliders) and it is noticeably better than the original/V1!

Yes, keeping both of them will give more control. Also, when trying to restore a picture the new strong gives kind of a craquelure effect, which is too much in many cases I’ve tried.

1 Like

V2 take way to much detail in my opinion.

3 Likes

beta version works much better, it seems to clean up the background well.

Noise normal Beta2 continuously crashes the program, no warning, just shuts down. It doesn’t happen every time but more often than not.

Normal V2 is a LOT better, thanks! Preserves detail and doesn’t over-smooth like V1 did.

the v2 versions are better for de noise, but if there is text in the picture it becomes unreadable. it looks like it is blown up.