Number of Controls

Personally I would prefer to have the chance to post filters with more controls.

I make an example, just to be clear.
Today I was going to post a filter.

The structure was composed basically of 6 polygons that, blended together in a bomber, gave some nice effect over an image.
The filter was not too complex as structure and not too difficult to understand.

As you may imagine, when you got a free polygon, you add a control for the color, another for the number of sides, then...the bevel, the radius, etc etc.
After that, you can give to the user some option ( invert, noise, or others.... ).
Clearly all these controls are " x 6 " because you have 6 polygons.
You can't apply the same controls to all polygons, otherwise the result would be very boring ( all polygons would get the same shape, color, .... ).
What did I get in that filter? 150 controls.

FF library only allows filters with less than 100 controls.
....
Deleting 50 controls would make that filter useless.

Somebody would say "That means that the filter was a bad idea".
A bad idea is a filter that doesn't look good, no matter it has 1 or 50000 controls.
When a filter looks good but the library doesn't let you upload it, I can't exactly say that it is a "bad idea".

FF can surely put a limit to the controls number but..... higher than now.
Considering that the most simple effects are already done by others, doing completely new things is becoming very difficult.
If we limit the number of controls and the size of the filter too much, we may end doing filters that only repeat what other people did before ( the only big difference will be given by the new versions of FF, adding more precision and quality in the components ).

Oh sweet a great tutorial idea. If I have some time I'll probably cover it.

Oh wait I'm gonna throw in a debate about this.

I argued on the stance with "Monster Filters" with SpaceRay before on several of his gigantic filters. Not that "Monster filters" are wrong, it's just that FilterForge's current UI doesn't really accommodate it well. Hence in many cases, it leads to rather frustrating filters. FilterForge is also not "real-time", hence feedback for each slider comes at a cost of time. This really depends on the complexity of the filter, but in many cases, FilterForge's super awesome image processing capabilities do come at a cost of time.

When it comes to the UI FF's little control tab wasn't made to fit many things in the first place. On most monitors when you got 20... or 30 controls of any kind, some of them will go off view. So you have to scroll to get to some of the parameters leaving you no option to view it all in 1 unified view. The UI controllers are also a factor in adding to the confusion. For example, XY coords should best be controlled with an XY pad, not sliders. Certain nodes with "options" should provide menus instead of controls through int-sliders. The UI itself should disable portions when certain chains are not used. etc etc... many of which have been discussed in the past.

Since 2.0 there have been many many changes to FilterForge making it quite the ultimate tool, but several factors like the UI hasn't been exactly up-to-date with it. And we have to come by to understand that there are certain principles at work here. One of which is that FilterForge is supposed to be part of a process, not intended to be the "art generator" or "the thing that makes images look like other things". It's intended to work with Photoshop in a quick, intuitive way to deliver you textures and effects rapidly. However, since the updates, it's opened up a huge parade of possibilities leading up to this discussion.

Unfortunately, I'm a bit busy so I'll have to cut this train of thought here... and return later in a video encompassing the points here. I think it's worthy of talk. If not, I'll just write in more specifics. We'll see!

About the UI :

I agree with Skybase.
It needs an update.

This period I'm working a lot with the XY settings.
I perfectly understand what you mean.
As you said, using sliders to control them is not the best. a pad would be much more intuitive.

An idea could be :

- Creating a filter, collecting the commands in areas ( someway like we already do for the Groups ).
Example : we make a filter that generates a sphere and then adds some effect to that same sphere. We could split the filter in a first area to create the sphere and other areas for every effect.

- These areas could appear in the Settings menu as folders ( same as we already do for the filter selection ). Each folder would have a name and it would be easy to understand what that particular folder and the commands inside are used for.

- If the filter maker doesn't create any area, the filter Settings would appear with all controls viewable in a single scrolling row, like now ( this way there would be no problem with old filters, with very small filters, or with who simply doesn't want to use the areas )

Yeah, I get where you're stemming from haha. Sorry we're kinda going off topic but it's one of those topics that do get brought up every now and then. I just think it's important because this discussion DOES relate to many many genres of computer graphics and to me, it's one of those topics that rarely get discussed in depth because people think it's intuitive.

I just come from the sort of CGI / VFX background doing work like this all the time. I get tools that are hundreds and hundreds of parameters deep and gets very technical quickly. So we have macro controls for those. A set of 10 simple sliders and knobs that just work itself out. This alleviates our task of digging out that 1 control, instead lets us just focus on what we need to focus on: design.

Gotta think about end users too!

And please forgive me for extending the discussion.