YOUR ACCOUNT

Messages 46 - 73 of 73
First | Prev. | 1 2 | Next | Last 
Login or Register to post new topics or replies
Burt
Random Old Guy
Posts: 333
Filters: 12
This may be another lame brained comment but it seems that applying curves to noise - like the Blocks- seperates and might be able to keep particles from overlapping? Maybe I am reading it wrong but this image shows the Blocks component with 100% contrast without a curve and then with one. It seems that by adjusting the curve you affect the distance around an object and avoid overlapping (only managed a quick example as I am stealing a moment at work to do this so some overlap but tweaking the curve changes that) - or am I reading the results wrong? If a curve could be used to define the space around an object to some degree then maybe adding a curve node to the Bomber would be useful?

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
As I said already in the previous page, I would really love that the Bomber component could be upgraded and added a spacing control between the particles and also a possible collision detection (or whatever you want to call it) so it could be possible to make non-ovelapping particles available

Or perhaps find a way using a determined group of components already available to be able to have this spacing control and collision detection with bomber and have it as a group component
  Details E-Mail
ThreeDee
Lost in Space

Posts: 1672
Filters: 112
This may be easier to achieve with the new loop component, but there will obviously be a limit to the number of particles.
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Quote
Burt

like the Blocks- seperates and might be able to keep particles from overlapping? Maybe I am reading it wrong but this image shows the Blocks component with 100% contrast without a curve and then with one.



I do not know if this is good or not, but looks interesting what you have done, and mayber is not useful (I really do not know) for this but could be useul for another thing, like it
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
I still wish and think that it would be the best thing and very useful if there could be some "Spacing" control inside the bomber made by FF developers

IN FF 3.0 bomber was greatly updated with the VERY useful of having multiple images input but the problem I find is that they are very randomly positioned when the particles are high, I mean when you rise density.

Or if they do not want to make it or can´t, maybe an alternative group of components that could be used to control the spacing in the bomber result as have already been shown

Quote
This may be easier to achieve with the new loop component, but there will obviously be a limit to the number of particles.


Is this true? Would this be possible in some way with the loops component?

Of course that the would be a limit, but is really not needed a higher number of particles to make something useful
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
My initial thought was exactly that "Oh maybe this loops thing will save our days"

Turns out it's extremely involving. Basically in order for this whole spacing thing to work, the program basically needs to know what the distance between two items is. We can produce something like a gradient that determines the distance between each item. The problem is, doing something like checking for open spaces, checking for minimum distance, or checking to see what size the item needs to be... anything like that etc... the loop becomes extremely sluggish even at low counts, making the whole filter quite unusable (as far as I've played with stuff of this nature.)

So then the question I guess becomes "what about other shapes" or user input? The thing is, it's all the same. We need to know certain variables in order for it to work. So like a set of data is basically what's required for this whole thing to work.

Reminder: FilterForge doesn't do point clouds!

[Spelling edit]
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Ultimately given how FF behaves / works, it's really difficult to chew into the request here. I see it doable in logic. In practice, the whole limitation is basically a brick wall.
  Details E-Mail
ThreeDee
Lost in Space

Posts: 1672
Filters: 112
Right. Loops get really slow really fast trying to do something like this.

My latest attempt some 6 weeks ago was again Bomber-based, but it is too slow to be of any real use.

Non-overlapping Particles - Another.ffxml
  Details E-Mail
ThreeDee
Lost in Space

Posts: 1672
Filters: 112
I had 5 sets of bombers in a row and some fancy gradient look-up for the distance to nearest particle. Already slow at this point. Colors are different Bomber passes.

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Quote
ThreeDee
Loops get really slow really fast trying to do something like this


Anyway thanks for trying it and test that this way is not possible

Quote
ThreeDee
My latest attempt was again Bomber-based, but it is too slow to be of any real use.


Although it may be slow and is true that is not fast making it, although not either too much slow to use it, this is not the real problem, because your attempt is well done, and well thought and surely it works as it can be seen from your filter and makes a very good work isolating the dots and so they are non-ovelapping

BUT the problem is that it seems that the dots is the only thing it can isolate and make non-overlapping perfectly, because if you try to use any other shape or image it will not work as it happens with the dots, there will be a slightly overlapping with other shapes like triangles, squares or stars, although maybe these could be solved in some way, but the worse comes when trying to use external images, even round ones does not work, and also using irregular shapes

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
The above example is done with big shapes, if you reduce the size of these shapes it will be reduced the overlapping and would solve in some way the problem, but reducing the size also will add more empty space between the elements and add more distance between them

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Here is the modified filter that I have used to make the examples above and are stored as presets inside the filter

I have included in the about tab the following text

SELECT SOURCE

1 - Dots
2 - Triangle
3 - Triangle Gradient
4 - Star
5 - Star Gradient
6 - External Image 1
7 - External Image 2


I have used two external images, because I wanted to put in the presets one circular shape (FF Lifesaver) and one irregular shape (FF Kingfisher)

The presets are also ordered as the source number with and without rotation chaos

The filter as the original above is for FF 4.0

Non-overlapping Particles - Another - Images.ffxml
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
ALSO is important that all the above examples are done all with exactly same size, and maybe it will be even more difficult with variable sizes

So I have made an update to the filter to include also the Variable Size inside the bomber with Size Chaos

And have also added the corresponding presets with the Size Chaos too

Now instead of the checkbox to add rotation chaos you have the following

Rotation/Size chaos slider in se ttings

1 - Normal
2 - With rotation chaos
3 - With rotation chaos and size chaos

Non-overlapping Particles - Another - Images_2.ffxml
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Here is an example using the rotation and size chaos from the uploaded in the post above here

As can be seen with the stars the overlapping is very low, but not with the images, that is much higher the overlapping and does not work in the same way as with the stars

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Although the ThreeDee atempt is good and well thought, as I have tested with the 3 post showing the available examples I have put above, is not good for images, only can be good for shapes when reducing the source size a little, and so after testing this I agree totally with Skybase and what he have written above

Quote
Skybase

Turns out it's extremely involving. Basically in order for this whole spacing thing to work, the program basically needs to know what the distance between two items


I think that this may seem something not very difficult, but I think that is much more complex and difficult to make than it seems as you have to know not only the distance between the two items, you must also analyse and know the whole and full shape contour and know all the space it fills so no other shape, fills the same space, and making circles none overlapping is much more easier, than making non-overlapping other shapes, or even more complex using irregular shapes.

ONLY REAL POSSIBLE WAY WITH FF IS WITH A LUA SCRIPT?

I do not want that this is considered as something I am requesting for others to do, is like a challenge, and of course that I will never ask and request that you make it, is just optional to anyone that would like and know to do it, but is not needed that you spend the time making this and do not bother otherwise


I think that the only way to make a working non-overlapping filter with FF is to make it with some LUA script that have one of the already done special algorithm that have been done by others experts and available on internet. Although I really do not know if there could be some possible way to convert an algorithm from the original source language to lua.

Sorry that I do not know nothing about Lua script, and do not know how different is this from the Processing.org language, because I know that there are already

Compilation of Random Space Filling Tiling articles for FF Filters

For examples of algorithms already done, examples are shown on this other thread

Other threads related to this one

ThreeDee made this other thread - Pattern Filling

How would you make this variable dots pattern in FF?

Make this in FF? - Digital Circlism - Artwork with hundreds of circles
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Thanks very much ThreeDee for making and sharing this filter, as finally I like how it is done smile:)

After some more testing I have seen that although it does not work on the non-overlapping way perfectly, is not at all a bad idea, and it works well in some way for having a way to control the distribution of the bomber particles and try to avoid as much as possible the overlapping, and get a good random but ordered and organized in some way distribution

I have made a new filter based on this one to use only images instead of shapes and have added some controls

I like the results even if they are not non-overlapping, but the overlapping is only slighty in some parts, and surely better than the default bomber result.

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Here is the filter I have done based on ThreeDee source and that it uses images instead of dots or shapes

Thanks to ThreeDee

Non-overlapping Particles - Another - Real Images.ffxml
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Here is an example with a real photo of a flower using the filter uploaded here above to show how the distribution of the filter works.

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
I have thought that as there is still some overlapping, it would be good idea perhaps to add some drop shadows and add some depth and separate the particles in layers, but when adding the drop shadows I got a bad thing, the result takes nearly 4 TIMES MORE to render than without shadow and is only using one blur component

I have used the most simple drop shadow as shown by crapadilla in his wiki page



If Threedee was thinking that his filter was slow, wait until you add drop shadows this way and multiply nearly by four the time smile:D smile:cry:

  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Renewed stuff from a couple years ago.

  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Really quick self note about the method above: Impractical but works with imagery.

SR, ThreeDee's method relies on gradients. Each iteration needs a much more distinguished boundary to feed into chance. Otherwise it's not doing justice.
  Details E-Mail
voldemort
voldemort
Posts: 872
Filters: 649
Could my filter framed tile snippet be used for your spacing by using an alpha instead and individual sliders for the scale options?
lets all whine for a wine port
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Quote
voldermort

Could my filter framed tile snippet be used for your spacing by using an alpha instead and individual sliders for the scale options?


Thanks or the suggestion and answering, but I have seen your filter, but I am sorry that I do not understand what you mean by using an alpha instead (instead of what?), and even if I could add individual sliders, this is not using the bomber, so all the images are always straight, and there is no random rotation or random resize, and how would you multiply the images so they they are not always in a square pack fixed?

And in your filter you can only use one image alone, so you would have to multiply the components for each image you want to add, and then how would you mix them? The only way I know to mix multiple images is the bomber, because the multiblend would not work.

So I am sorry that I do not know how your filter could be used for the spacing, ir you could explain it more, would.be good, Thanks
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Quote
Skybase
Renewed stuff from a couple years ago.


Is possible to know HOW you have done this or maybe you want to keep it secret and I understand it and will not ask for it

It looks very good and nice with different shapes, random sizes, and random rotation, different colors, and more important is that all of them are mostly NON-Overlapping, although there is much space between them, but anyway it is good


Quote
Skybase

Really quick self note about the method above: Impractical but works with imagery.


Why is impractical? Is slower and worse than ThreeDee option?
And if this can work with images would be good.

Quote
SR, ThreeDee's method relies on gradients. Each iteration needs a much more distinguished boundary to feed into chance. Otherwise it's not doing justice.


OH Yes! You are right that each iteration needs good and clean distinguished boundary, and adding the shadows before the bomber as I have done is wrong, so now I have added it after the bomber to each one of the results and works better and faster. Thanks
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Part of the reason why I don't upload stuff on public space these days isn't about trying to keep secrecy, it's more that I have difficulty uploading source material that have certain commercial value, provided some of the works were used commercially as well. However, it is more so that I simply hope to encourage people to work out their own solutions and methods for their filters rather than stuff becoming a "copy-and-paste and modify" snippet thing. As much as I encourage openness, it should also be noted that I sometimes feel people rather just find their own ways with their work. While certain things like equations, algorithms, and base recipes remain fundamental, I strongly feel certain specifics should remain rather an inspiration for new ideas.

With it being impractical, the issue is more so the amount of control you have with the whole thing. It's kinda mediocre in terms of effort.

Lastly, it's not really wrong, more like it doesn't spare the effect of them looking like "separate entities", rather ended up looking like a result from 1 bomber with the density set at 2.
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
I agree with you Skybase, and you are totally free to do what you want and of course that I understand and agree with any reason you may have for not uploading, and you do not have to explain why you do not want to do it, nobody in the forum is force to share something that they do not want, someone could ask for it, and you have the right to say no, and nothing happens by saying no, you have your own reasons not to do it. I respect and do not feel bad for this.

I agree with you also on the "copy-and-paste and modify" snippet thing as I am one of them, but doing it this way, I try to understand and learn WHY you have done it this way, and how it is working and not just copy and paste and not care why I am doing it, unless is something complex that I do not understand.

The problem is that you could encourage people to find their own solutions and do it their own way IF they know and are FF experts with high skills and lots of experience like you and many others in the forum, but for me is still difficult to do it even if I try to do it because I really CAN´T know to do something that I do not know how to do it

You could have some control if instead of sharing in the forum you could do it privately as you have said you are already doing with others in the forum
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
Thanks to Skybase that have made me go the right way, when I have seen that it was much better to put the shadows AFTER the bomber and not before I have made a new version that works faster and gives better shadows than before and I have added also the color customization, so you can have only one source but can have it with up to 5 colors

And as you can see in the screenshot below the render time is now more than half less than before and this is considering that now there are 5 individual shadows groups and before was only one but before bomber (it may be only one although maybe is really like 5, as it had links to each bomber)

Note to me: It must be considered and think carefully WHERE you put the components and how they work together combined and not put it in anywhere inside the filter, even if you get a good result, because there can be others better and more efficient ways to do it.

  Details E-Mail
SpaceRay
SpaceRay

Posts: 12299
Filters: 35
And until now I have used only one source for all the 5 bombers of this filter, BUT what will happen if you use 5 different sources instead?

Will it also work in the same way?

The answer is YES!! it also works well smile:)

The render time for pnly 1 source or 5 is the same, but when using the shadows is not the same and the time is higher, but have not used the same settings, so it really can´t be compared.

  Details E-Mail

Messages 46 - 73 of 73
First | Prev. | 1 2 | Next | Last 

Join Our Community!

Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!

33,712 Registered Users
+19 new in 30 days!

153,534 Posts
+27 new in 30 days!

15,348 Topics
+72 new in year!

Create an Account

Online Users Last minute:

28 unregistered users.