Violet noise removal?

A forum for exporter development discussion.
rerdavies
Posts: 31
Joined: Fri May 04, 2007 4:38 am

Re: Violet noise removal?

Post by rerdavies » Sun Jun 03, 2007 3:30 am

IanC wrote:Is anyone working on adding noise removal to violet?
Can't see any reason not to. The current UI won't get any more cluttered because of it.

re progress bars on noise removal. I didn't add it because I didn't think they were that slow (unlike the bloom effect, that really is slow). If you think they are slow, lemme know, and I'll add progress indicators. (Pure laziness, not hard to do).

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Sun Jun 03, 2007 3:52 am

@ rerdavies:
Did you read through the rest of the thread?

IanC
Posts: 147
Joined: Thu Aug 24, 2006 4:46 am

Post by IanC » Sun Jun 03, 2007 4:04 am

Kram1032 wrote:so, ICA is a "REAL" Noise Filter, instead of a hmmm....
a what?
A "unreal" one? noooo....
Actually, it's for separating signals. Like hearing two people talk at the same time, it is designed to give you two outputs, one for each person talking.

There are some fairly complex ways it's used for images though, more complex than I thought. I've a few simpler ideas for it though.
OOG searches for pixels with too much difference from it's surrounding neighbours.
Then, it takes the average of all the neighbours, and fills the noise pixel with that colour.
Hey, that's my algorithm :p back on the old forums I posted that up with a program. Someone else did it before me too, I think.

FFT might work but it might blur and distort things too much.

User avatar
zsouthboy
Posts: 1395
Joined: Fri Oct 13, 2006 5:12 am

Post by zsouthboy » Sun Jun 03, 2007 4:46 am

Yeah OOG is a simple box-filter - like the way hot pixel reduction works on many DSLRs.

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Sun Jun 03, 2007 8:00 pm

I spoke with Arne about it, asking, if it's a good description.
But then, he said, it is a good description on how it WILL work, not on how it works NOW...
at the moment, it indeed only blurs a little amount arround Fireflies.
It wont remove dark spots. If there is many black in your scene, it even could happen, that the dark spaces get BIGGER :shock:

User avatar
manitwo
Posts: 1029
Joined: Wed Jul 05, 2006 4:50 am
Location: Tirol - Austria

Post by manitwo » Sun Jun 03, 2007 11:34 pm

Kram1032 wrote:... that the dark spaces get BIGGER :shock:
definatly a case for scully and mulder :shock: :lol:

User avatar
arneoog
Indigo 100
Posts: 504
Joined: Sun Jun 25, 2006 2:19 am
Contact:

Post by arneoog » Sun Jun 03, 2007 11:50 pm

:lol:



..it is only a firefly filter..
..for images that has rendered for ages, but still can't get rid of them.

:)

rerdavies
Posts: 31
Joined: Fri May 04, 2007 4:38 am

Post by rerdavies » Tue Jun 05, 2007 12:07 pm

Kram1032 wrote:@ rerdavies:
Did you read through the rest of the thread?
Thinks so. What I meant to say is, if you want add, modify, or extend existing noise filters, there's plenty of space for new buttons, UI, &c. Did I miss something?

I haven't really done much with sources since posting 1.03. Barring a SVN commit (don't have rights), I'd be happy to host any changes or updates that anyone cares to make.

I had some glossy-eyed theories about adding support for iris shapes in the new bloom filter, but then i found why optics is *frickin' hard*!! (Much harder than audio stuff, and 3D stuff, and graphics stuff. which is my usual bread and butter). Still trying to figure out a way to compute the diffraction patterns in something less than hours or days.... :-(

User avatar
zsouthboy
Posts: 1395
Joined: Fri Oct 13, 2006 5:12 am

Post by zsouthboy » Tue Jun 05, 2007 2:57 pm

Try calcing the diffrac patterns for a much smaller subset of wavelengths, and try fewer points of sampling with more interpolation?

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Wed Jun 06, 2007 6:19 am

Where I used it for the sun, in my amethyst scene, it wasn't calculated that slow...
The glare was the thing, my computer computed for half an hour, or so...
diffraction limited bloom was done in about ten minutes. :)

User avatar
manitwo
Posts: 1029
Joined: Wed Jul 05, 2006 4:50 am
Location: Tirol - Austria

Post by manitwo » Wed Jun 06, 2007 6:28 am

:shock: half an hour?
did you render it with your old gameboy?
for glare max. 2min by me.
diffr. limited bloom is even faster - around 20sec.
... and i do not own a superfast monster-computer nor use small-size igi's.

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Wed Jun 06, 2007 7:41 am

yes, the bloom is really fast. I guess, my mind blowed up the time^^
but glare gets extremely slow with a high number of arms (argh, I know, these aren't called arms, but I everytime forget the correct name :? It's more logical for me, to call them arms): I used 100 arms for the pic. It needs quite long, then ;)

User avatar
dougal2
Developer
Posts: 2532
Joined: Wed Nov 15, 2006 8:17 am
Location: South London

Post by dougal2 » Wed Jun 06, 2007 8:43 am

well, for every glare "arm" ( ;) ) Violet has to do a directional gaussian blur, rotate the image, and accumulate the result.

for 100 arms that's no small number of calculations.

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Wed Jun 06, 2007 8:46 am

yes.
I just wanted to say, that there is NO problem with the speed of dif-lim-bloom ;)

IanC
Posts: 147
Joined: Thu Aug 24, 2006 4:46 am

Post by IanC » Tue Jun 19, 2007 10:26 am

Well, I don't know if I'll get around to this, but I'll have a tinker over the summer. I'm off on holiday, then hopefully have a scholarship working with robots (woo!) over the summer.

Anything anyone would like to see in violet?

Post Reply
31 posts

Who is online

Users browsing this forum: No registered users and 19 guests