The Filter That Was Doomed: Guidelines for Avoiding 'Low Usage'

I was wondering whether it was possible to compile a list of general "characteristics" that 'Low' usage filters have in common as a guideline for filter authors.

Some of the most important "characteristics" off the top of my head...

[*]Poor Interactive Rendering Performance: One of the prime usage killers in my book. If a filter is too slow for me to be able to tweak it interactively, I'll seldom bother, regardless of the visual quality of the filter. I had to learn this the hard way with 'Synthetic Cubism'. While I love the filter, it's a royal PITA to tweak even on fast machines.

[*]Over-specialization: The filter is so specialized that its potential use is limited from the start. Take my 'Liquorice Snail' as an example: It looks nice but still it has low usage (and that since ages, because it also renders slowly).

[*]Slap-Dash-Authoring: The filter hasn't received any 'love' whatsoever from its author. This is mostly evident in non-existent/trivial descriptions, poor (or less than 9) factory presets and bad control naming.

[*]The Big Yawn Effect: This is similar to the "Slap-Dash-Authoring" in that the filter simply fails to make any impression at all. Filter library users have been so overexposed to large numbers of bland noises, generic patterns and kaleidoscopes, so don't bother to sink another one in the flood. Be creative!

[*]Lifesaver-centric Authoring: The filter looks excellent on the default lifesaver image, but it hasn't been constructed for other images or even tested on them. Big mistake, as the filter will quickly be dumped when this is discovered!

[*]Triviality: The filter does something that is utterly trivial. If that something can be achieved by standard photoshop means in a fraction of the loading-time of FF, why bother with the filter?

[*]Rip-Off: The filter is a clear rip-off of some successful filter. It is likely that users familiar with the filter library will detect such rip-offs instantly and just use the original filter.

[*]Broken Seamlessness: Many filters break when the 'Size, pixels' slider is used. In many cases this indicates bad filter construction and constitutes a major detriment towards earning an Editor's Pick.

[*]Bad Lead Preset: Since the lead preset is the first thing most users see when stumbling upon a filter, it should be as spectacular and representative of the filter as possible. Anything else just sells the filter below its actual value.

Feel free to contribute...

[*]Jack-Of-All-Trades: These are filters that attempt to do everything and consequently fail to do anything really well. My personal advice would be to keep filters focused on specific tasks (naturally there are exceptions to the rule). Keep in mind that filters are usually expected to provide solutions to very specific visual 'problems', whereas a jack-off-all-trades filter best lends itself to experimental purposes.

[*]Undiscoverable: With 4000+ filters on the library, keyword searches become very important. Many filters cannot be reliably found via these searches simply because they sport bad/undescriptive titles, descriptions and insufficient keywords that fail to provide the search engine with something to work with. Undiscoverable filters are a symptom of applied "Slap-Dash-Authoring".

Quote
Mousewrites wrote: (Scab, anyone?)

----Seems like you showed an example of that, and it was for-text. That should maybe almost be a new category really, something that seems to be popular/highly used for commercial brochure/ad/magazine graphic artists. You might title them like "Scab - For Text", or something along those lines, so that people would know exactly how it's going to work best for them, like in what situation. Just an idea, since I swear I recall seeing an example like that.

jffe

Quote
jffe wrote: Seems like you showed an example of that, and it was for-text

Yeah, it's in the comments.

I have several things like that; they look TERRIBLE with the default input, but I have no way of showing what I acutally intend them for without putting an image into the comments.

Of course, that really doesn't help get a HU, becuase people aren't looking in the comments for other example pics. My Midas filter I'm working on is going to be the same way. It doesn't look so good on the lifesaver, (I'm tweaking it so it will, as much as I can) but it does stuff like this:


The only way people are going to know how it really looks is to look in the comments.

I wish that we had an optional 'example pic' we could upload to the filter page, so people could see what the author intended.

Interesting Midas filter, Mousewrites. My suggestion is, in your filter description, mention that you're going to provide better image examples in the filter's Comment Thread. (I think this would get me curious enough to click on the comments link.)



Crapadilla, nice tips.

(I hope this doesn't stifle creativity too much -- as I also have appreciated people's filters even when they break these ideas. That is, even if I won't use it much, I often like the "one-time" examination of the filter in a tutorial/technique sense....)

Quote
jffe wrote: Speaking of CF&M filters, did you ever do an Abstract Growth 4 ? 3 was alright, but I liked 2 the best so far.

Nope never did a 4...Spring of 08 maybe..
I liked 2 better than 3 also..


Back to the thread...

Quote
Carl wrote: so maybe a tip to Checking Library before submitting Smile

Yep I had that in a tut that I wrote a while ago..It was one of the first things
to do before anything...

Bump! This must not be buried, so I'm compiling it into a wiki entry...

Done!

There is now a quick & dirty wiki page for this. Feel free to expand upon it...

Quote
Carl wrote: oh and thanks for not mentioning me by name Wink Big grin

I think I was close to first that violated this one..Broken Seamlessness...
I don't call it poor construction..I call it Creative Construction..

None of the filters I've made have ever been designed for anything else other than some texturing project I was working on, or as a byproduct of one... So technically, I'm not trying to get usage points, just passing on stuff I found useful...

But, what I was wondering is...

Doesn't it matter more that people set their preferences to report usage?

No matter how great a filter is and no matter how many people use it, unless Filter Forge passes that information on, it won't be known.
I'm assuming that is how it works.
I know I've had my various versions set to report usage for a long time, but I don't remember if it was a default setting or if I had to set it... I remember at one point it became "unset" but I reset it.

I bring this up because I know there are a lot of folks that dislike or distrust software "phoning home"... But as far as I know, Filter Forge only reports the filter usage with that setting.
Correct?

"[*]Jack-Of-All-Trades: These are filters that attempt to do everything and consequently fail to do anything really well."

i'll take a small exception to this one because I've made several "multi-filters" which are not designed to be very specific and do try to do "everything". this is on purpose. the role of these filters is to inspire other filters. I throw things together without any care to if a thing looks good or not and I wire the components such that one component may get used in several different series or strings of components. thus, when you hit the randomize control, you are never sure what you're going to get. that's the whole purpose. I've made many filters off of a multi-filter. once I hit the randomize and something interesting shows up, I make a preset and go into the editor. I then hit that preset again and find the string of components that made that effect and then make a new filter out of that series.

saying all that, I almost never publish the jack of all trades filters because most of the randomizing comes up mostly worthless... but every once in a while, I find a gem

It works for me. If you think it got stuck, please switch it to "Always send", restart Filter Forge and switch it back to "Always ask". You can also open c:\Users\your_user_name\AppData\Roaming\Filter Forge 6\Options.xml and check the UsageStatsSendingType parameter is set to 1.