Version 2 of Skatter is basically a complete rewrite from scratch. Part of this process includes questioning the current UX/workflow.
There are some modifications pretty obvious, but we stumbled upon a dilemma that we can’t resolve. We need your opinion:
Should we merge the Grid and Uniform distribution types?
They are basically the same, except that in Grid the Jitter parameter is not exposed in the UI and is always 0, and in Uniform it defaults to 100%.
Arguments for merging:
It’s just simpler, less confusing
Less redundancy, it takes less space in the UI
Arguments for keeping them separate:
The Jitter at 0% or 100% produces a completely different distribution, this might not be obvious to new users. When the Jitter is at 100%, it doesn’t look like a grid at all, but is closer to the Random distribution.
I use uniform much more often than random, i mean i like better that with uniform i can give it the amount of randomness i want, this i like better than the random that has no value to reduce the randomness, so uniform is my distribution of choice. i hope i explained myself this time
I use both so I’d prefer to keep them both. Sometimes you need an even distribution and other times more random. Without knowing how the clumping feature will work it’s a bit hard to know which way is the best.
Just to be clear, all the options you already have will stay. You will still be able to do exactly what you can do in v1.
If we merge Grid and Uniform, you will still be able to get the results of both Grid and Uniform. And Random is not going anywhere.
If we merge the two, it is basically like removing the current Grid type, and renaming the current Uniform type as “Grid” and setting the Jitter to 0% by default. Or to see it the other way, we are removing Uniform and adding Jitter to Grid. One way or the other, all the options will stay.
The question is directed toward the feature discovery for new users. Is it useful to keep the two separate if we can merge them while retaining the same amount of control?
How about merging them and adding a tooltip to the word “type” in the Distribution tab? It should contain some sort of presets for begginers.
Simple UI is better. But it’s very helpful to have some kind of info about the result you will achive by changing some parameters to a specific value, especialy if you are a new user.
Yes we plan on having tooltips everywhere. Also tooltips will probably have two states: a simple line of text displayed when you hover an element for a short time (1 second or so). And a second state with more text and possibly pictures and gifs, displayed when you continue to hover for a longer period of time.
I think I’m with Ronen here, GRID means something very different from uniform & random - even uniform is to some degree misleading to the layman as it can produce a random (seemingly) distribution. Underlying though, I prefer the control that uniform offers. I think GRID should stand alone.
Ronen is on the right track. It’s primarily an issue of naming. Random has to be called random, no question there.
The other category should be called Uniform, as uniformity defines a grid, but a grid does not define uniformity. It should be apparent to me users that a grid could be achieved with a uniform distribution. I would suggest 0% jitter as default for the uniform distribution setting, as that is similar to the opposite of random distribution.
Frankly I’ve always hated the term jitter, but maybe that’s because I don’t completely understand what it is mathematically. I know it’s regularly used in CG applications, but I feel like that term makes this issue somewhat confusing for new users, as it is an uncommon usage of the term outside of CG. Can’t you just call it ‘offset’ or ‘randomness’ or ‘scatter’ or ‘scatteritude’ (I know this is not a winning battle…)
As an experienced user it makes sense to merge uniform and grid, but that probably would have thrown me as a new user, so I would suggest keeping the redundancy.
What I’d actually like (I suppose this is a feature request) is 3 kinds of random:
uniform (grid with jitter)
blue noise (points that are tightly-packed, but no closer to each other than a specified minimum distance generated with e.g. a Poisson-disc sampling algorithm)
Blue noise (#3) is particularly good for crowds and swarms. I currently have to do this myself with my own clunky code. I’d much rather do it within Skatter.
I’d also like to be able define the distribution by choosing a specific number of points for the host rather than choosing either a grid dimension or a density. That is, I want to select a host and say: “put 2,467 objects on here at random” [or whatever number].
I agree with Ronen. The naming of RANDOM and UNIFORM is a bit confusing. I think Grid can stand on its own and have it’s specific parameters. RANDOM can have more random and less random (i.e. UNIFORM) unless I’m miss defining “UNIFORM.”
On another note, is there a rough timetable on the release date for version 2? No pressure I’m just excited to see what if can do!