Northern Lights and a request...

[Rovastar]

Hi all,

Adding to the positive plethora of presets at moment here are another 4 presets as part of a series entitled Northern Lights at:

www.milkdrop.co.uk/NorthernLights.zip

The original Northern Lights is one of my favourites that I have created at the recently.

But (and there always is a but) I am not totally happy with these and therefore a request to all you preset writers out there.

Why are these presets not symmetrical? Much more noticeable on the ‘Bud’ mods (sorry if they are slow I just wanted to try something a bit different with these), in the original it is difficult to see and does matter too much.

I believe that it is to do with this line:

rot=pow(-ang,3)/10-sin(ang)

Can it be fixed? I realise that this may be challenging (well I am stumped) but can anyone out there help? Krash, Zylot, unchained, anyone…..

If not sit back and enjoy (that’s if you like them).

Rovastar

[Krash]

The reason it's lopsided is because of that line you mentioned. You see, any pixels below the midline have an ang value which is negative, so pow(-ang,3) yields a positive value. (negative of a negative is positive, to the 3rd power is still positive).
Any pixels above the midline will yield a negative value. When this is combined with a sin of the angle, it ends up making the wierd lopsided effect you've got here.

I did a bunch of fiddling to try and generate values which would overcome this - they worked, but because of the way the preset works (its basically a fractal), the subtle changes multiplied, and the preset didn't look the same. And not in a good way. I ended up losing the nice rounded effect on the centre area (and its fractal copies), and it was replaced with strange jagged bits.

Unfortunately, I couldn't duplicate the effect you've already got, and have it symmetrical.

- Krash

[Rovastar]

Thanks for trying Krash but I feel that the problem is not with pow(-ang,3) but with the -sin(ang). I have played a little more with this myself and if you have a mulitpler of 1.2 like this

rot=pow(-ang,3)/10-1.2*sin(ang)

It is nearly correct but why 1.2 relevant I do not know.

Have a look at that and see if you can perfect it more.

Rovastar

[unchained]

I was running into a lot of those types of problems myself, they can be a pain in the ass when you have something SO CLOSE to the effect you're trying to get.

MANY thanks again to Krash for the excellent "krash course" in polar geometry he gave me a while back. I'll take a look at this batch in a minute...even if I can't fix them, they're bound to give me some ideas. Milkdrop's one of those great sciences where the failures are often more interesting as the successes and lead to new and better ideas. :)

--
unchained

[unchained]

No luck

pow(-ang,3)/10 and -sin(ang)

are both lop-sided all by themselves, and any way I try to fix either one blows the whole effect. multiplying by 1.2 (or 1.18, somewhere in there) or by sqrt(ang) fixes it to a large extent. If it were one of my own presets, i'd do something like

pow(ang,3)/10-sin(ang)*(1.18+.1*mid_att)

and let it "wiggle" a bit :)

It's a damn cool effect either way. I wouldn't sweat it.

--
unchained

[Zylot]

That stinks, however:
I like the effects very much.. I think despite the fact that they ain't perfect they are so damned good it doesn't matter.

Kudos to Northern Lights.

[Rovastar]

First of thanks people for trying to sort this out for me. It just proves what a good community that we have got on the milkdrop forums. :)

but is is still a little frustrating that they are so close to being just right and just miss out...maybe I am too much of a perfectionist....but at least you lot seem to like them anyway :)

I think at I will settle on a 1.19 multipler unless someone comes up with a better equation.

Anyway if I can return the favour with your presets then do not hesitate to add a post to the forum and I will see what I can do.

Cheers,

Rovastar