Software - - Filter Forge - Page 4

Stroker · #46 05-29-2006, 12:14 PM

You'll have to pardon me. Wisdom teeth + codine + vicodin = brain not work so good. All I really remember is seeing the word border.

When I get my higher brain functions functioning a little better, I'll see about putting together an NPR thingie.

Steve Conway · #47 05-29-2006, 12:32 PM

Craig, all I did was sign up and the XP "save file" window popped up and I downloaded the pgm...(beta version, I assume). Now, I am also reading something about a trial time limit. Am I suddenly going to lose the program after a certain length of time unless I submit filters? If so I'll pass. I did not sign up as a beta tester.

B.T.W. the setup file is 24.7 megs. This sound like the file you folks got when you downloaded?

Steve

Quote:

Originally Posted by Kraellin

steve,

currently, the way things work, you HAVE to sign up for the beta AND be approved before you can even download it. and yes, it comes with ALL currently approved filters and textures. i dont know exactly how many there are now, but it's a fair number.

and, from what i gather, once it's out of beta, this will still work this way. the filters can NOT be used without the program. so, you get everything when you sign up for beta or buy the released version. and, there is a 'check the site for new filters' function. all nice and slick.

i've checked the file sizes too. the biggest filter size i've got so far is only 14 kb. that's pretty small. and the biggest one i've seen of the pre-made ones is only 35 kb. that means you could save roughly 40 THOUSAND before ever using a gig of harddrive space.

craig

Kraellin · #48 05-29-2006, 12:42 PM

steve,

i'm surprised you could download it without signing up. maybe they changed that. i do know about the time limit. it's 30 days. and yes, if you dont sign up it will expire. when you sign up for beta and get approved they send you a registration code so the time limit wont expire. you simply pull down one of the menu items and enter the registration code there and you're good to go.

i dont know how many they are accepting for beta either. vladimir could answer that. but i wouldnt wait. the list of folks is growing and you've got a good chance of getting a very nice program for free once it's released.

i think what they must be doing now is allowing folks to download the beta version but they dont have to sign up and they can get a discount or a free version by submitting 3 approved filters within the time limit. and the 2nd way is by signing up for beta, helping out with testing, submitting filters and having possibly a better chance of being rewarded the full version for free. at least that's the way i see it currently based on what i've seen and what you've said.

craig

Steve Conway · #49 05-29-2006, 03:15 PM

I did sign up, just not for beta testing.

This is all just a pastime thing for me and having to make filters or having to write reports on software is not something I care to get involved in. Makes it too much like a job. So I guess I'll pass.

Steve

Quote:

Originally Posted by Kraellin

steve,

i'm surprised you could download it without signing up. maybe they changed that. i do know about the time limit. it's 30 days. and yes, if you dont sign up it will expire. when you sign up for beta and get approved they send you a registration code so the time limit wont expire. you simply pull down one of the menu items and enter the registration code there and you're good to go.

i dont know how many they are accepting for beta either. vladimir could answer that. but i wouldnt wait. the list of folks is growing and you've got a good chance of getting a very nice program for free once it's released.

i think what they must be doing now is allowing folks to download the beta version but they dont have to sign up and they can get a discount or a free version by submitting 3 approved filters within the time limit. and the 2nd way is by signing up for beta, helping out with testing, submitting filters and having possibly a better chance of being rewarded the full version for free. at least that's the way i see it currently based on what i've seen and what you've said.

craig

filterforge · #50 05-30-2006, 08:53 AM

Quote:

Originally Posted by byRo

Here's a (lot) more complicated sketch effort

byRo, have you finished this sketch filter? It looks good, do you plan to submit it?

byRo · #51 05-30-2006, 09:20 AM

Had to do a rush translation (Portuguese -> English

) yesterday for my daughter.

I've got a lot more ideas to try out, hope to finish it tonight.

Rô

byRo · #52 05-30-2006, 07:38 PM

OK, I uploaded the sketch filter to the site.

In the atachaments, Default settings and two of the presets.

Rô

Cameraken · #53 06-06-2006, 09:56 AM

This is fun. Isn’t it?

Rô
I’ve been playing with your convolution filter (lots of uses for this as FF can’t provide this component)

I think I may have found an error in the code

In the blend component at the bottom. Please see my picture. I have marked what I think is wrong in Red and the correction in Blue.

Could you please take a look at this and see what you think?

I think I may attempt a 7x7 Kernel.

Has anyone got any good painterly kernels?

Ken.

byRo · #54 06-06-2006, 03:46 PM

Ken, I think you're right. (actually, I made two errors not one

)

I don't have a copy of the filter here, and I just discovered that I can't download from the site, 'cuz I've already done that.

As I remember the version before this one had the connection that you corrected. However, then I remembered that the central difference component should have double the weight of the others. I was trying to do that, but looking again, I think I got it all wrong

.

7 x 7?

More than the 48 offsets, the big problem would be in the weights. Just trying to do 3x3 with simple weights [1 2 1] I got into trouble

. Imagine 7 of them!
(or, maybe, you were just kidding?)

Quote:

Originally Posted by Ken

Has anyone got any good painterly kernels?

Not a kernel, but I do have a filter in the "unfinished" section using distorted tiling which may lead to something interesting.

Rô

Cameraken · #55 06-06-2006, 03:59 PM

Hi Rô.

I have been reading this thread
http://www.retouchpro.com/forums/sho...ghlight=kernel

but I am still a little confused about how to do the conversion to FF

Chris Russ posted the laplacian filter as
-1
-1 4 -1
-1
also here
http://www.homepages.inf.ed.ac.uk/rbf/HIPR2/log.htm#3
and the Sobel is here.
http://www.homepages.inf.ed.ac.uk/rbf/HIPR2/sobel.htm
and the Roberts Cross
http://www.homepages.inf.ed.ac.uk/rbf/HIPR2/roberts.htm
(Hypermedia Image Processing Reference from the University of Edinburgh is an excellent source of info.)

I still can’t figure the logic with your filter. The pixels are compared with a difference blend mode and then blended with screen? I know there is no Add and Subtract (yet) and so difference should be OK for a subtract {result := abs(a-b)} but why use screen? Does this bring it back to a Plus? Screen is the opposite of Multiply. {(a,b) = 1 - (1-a) * (1-b)}

So how would I make a Laplacian Filter? Would it be difference mode to compare and then normal mode to blend?
And what would be the best way to increase the effect if the numbers in the kernel are bigger than 1?

So, if I wish to apply 0,1.0.1.1.1.0,1,0
To pixel values 164,188,164,178,201,197,174,168,181
I would add the surrounding pixels (932) and divide by 5 (186).
So I would change 201 to 186.

So how would I apply this in FF ?
Also how would I apply 0,-1.0.-1.-1.-1.0,-1,0 ?

Attached is a 5x5 Filter. I give up with 7x7. I don’t have a big enough monitor

Ken.

byRo · #56 06-06-2006, 07:05 PM

Ken,
First, thanks for spotting the error. I have corrected it.

Vladimir had cleared my queue so that I could submit the "Adaptive Tiling" filter and now I'll have to wait until that gets through the process.

As to the workings....
There are three differences in each direction with the middle having double the weight of the others. As per your link the weighted sums of the differences in each direction will be "Gx" and "Gy".

The screen formula 1 - (1-a) * (1-b), can be rearranged to a + b - ab.
With a bit of sleight of hand (or just assuming that a and b are small - well sort of ) that reduces to just a + b, an accepted alternative.

Laplacian
Your matrix got a bit messed up in the post (extra spaces get eliminated) that would be:
00 -1 00
-1 +4 -1
00 -1 00

See atachment: Invert the original and set up 4 offsets, one in each direction.
As the differences are now signed (not absolute) then the subtraction can be done blending 50% of one with 50% of the other inverted - note that now mid-grey is our zero and the values can be positive and negative.
To sum them up we would need a Linear Light blending mode. Lacking this maybe Hard Light won't be too far off.

Got to run now...try and reply some more later, OK?

Rô

byRo · #57 06-06-2006, 09:02 PM

Ken, I had a look at your 5 x 5.

I noticed that you left all the blenders at their default value of 50%.
I changed them all to 100%, and also changed the summing points from difference to screen.

I don't have the faintest idea what this thing is actually calculating, but I just loved the result.

Rô

Steve Conway · #58 06-07-2006, 09:07 AM

F.Y.I. FF users. Found out there is a built-in device in this software, at least in the beta version, to detect what you do with, and what type of filters you make with it. Not sure if there is anything else it's peeking at?

Would like to try this, as everyone seems pleased with the results, but am a little leary of this bit of info.

Steve

byRo · #59 06-07-2006, 09:18 AM

Steve, I saw this too - but I also saw somewhere that you can turn it off.

Doesn't worry me too much though.
Vladimir and the guys are serious folk and they are only interested in feedback about their filter.
(or they fooled me real good!)

Rô

Cameraken · #60 06-07-2006, 02:55 PM

Hi Steve.

Here is the information about the info that FF use.
http://www.filterforge.com/more/usagestats.html

However it can be switched off in the preferences if you wish. There is also a message which asks if it’s OK to send the info before it is sent.

FF is worth trying. Some wonderful effects can be achieved. Just look what Vladimir had to say about Byro’s filter.

Quote:

Vladimir Golovin
Administrator
This one is absolutely brilliant from the technical standpoint. A month ago I wouldn't believe that this is made in Filter Forge:

http://www.filterforge.com/filters/471.html
Posted: June 7, 2006 4:14 am

Ken