Generic noise samples

suggest a way to improve Neat Video
Post Reply
Lugarimo
Posts: 114
Joined: Mon Feb 09, 2009 2:51 pm

Generic noise samples

Post by Lugarimo »

I haven't upgraded to new 4.6 yet but I have a suggestion to take the generic noise samples to another level if you haven't already done so.

Make a huge library of generic noise samples (perhaps let users submit them as long as they're at least 90% quality) and when a user tries to select a noise sample in a video where there's no suitable frame, NV can compare that selection with the best fit from the library and point him to that one instead.
Now he has a very similar noise sample with 90% quality vs. the 30% he could get in his video with no flat frames.

As someone who frequently had to photoshop frames to make them suitable to select a noise sample from, this would save me a lot of trouble.
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

Making a huge library would make it more difficult for end-users to work with. Also, those profiles would not be generic in the sense of the word we use with these new generic profiles we added in 4.6. Those are created in such a way as to work in the cases where you do not have an area to build a regular profile, regardless of the device that captured the video.

You still have to try several of those generic profiles to find one that works best in that clip (reduces enough noise and preserves most details). If the number of those profiles was hundreds, then choosing the right one would take ages. With just 7 profiles, that is much easier.

Vlad
Lugarimo
Posts: 114
Joined: Mon Feb 09, 2009 2:51 pm

Post by Lugarimo »

You misunderstood, I'm suggesting to make the process automated. No need for a user to browse the library manually.

He selects an area on the (unsuitable) frame and NV compares his selection to the best match in the noise profile library. So instead of having a 30x50 profile, he now has a 256x256 profile with 90% rating.
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

That would not be better than using the originally selected area.

Vlad
Lugarimo
Posts: 114
Joined: Mon Feb 09, 2009 2:51 pm

Post by Lugarimo »

Why not?
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

A small area is not representative sample of a larger area in such a situation. If it was, we would not need a larger area to begin with. There may be many larger areas that would "match" the small one, but those larger areas themselves may be very different in terms of the lower freq noise present there and not present in the small one.
Lugarimo
Posts: 114
Joined: Mon Feb 09, 2009 2:51 pm

Post by Lugarimo »

In most cases, inability to select a larger area is due to the fact that the video is already a small resolution. Many times I'm lucky with a 320x240 video because there's a scene with a flat area covering the entire frame so right away I get 256x240 noise profile but a lot of times I can't.
So very low freq noise shouldn't be an issue here.

Isn't it a better idea for a user to have a huge library auto-picked for him that will be close to the noise he's looking for versus a few generic profiles that will maybe SOMEWHAT resemble the noise?
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

> So very low freq noise shouldn't be an issue here.

Then the profile built using the small area selected by user will already be good for that purpose. Matching any other profiles will only add complexity but will not add accuracy in this situation.
Post Reply