[crossfire] Arch repository: layered art files? A bigger picture.

[Mark Wedel]

Kevin R. Bulgrien wrote:

> This spawned an idea that I could actually combine the whole arch set
> into one .xcf file.  Each tile could be reconstructed by turning some
> layers off and on.  If a complete set of graphics was not needed, the
> source image would facilitate re-use of the pre-existing elements of
> graphics needed to render the other views at some time in the future.

  That's an interesting thought.  But from a managability point of view, I'm not 
sure if having 3000 layers would be very useful.

  But things could certainly be done on smaller scale - all the walls of one 
type put in as different layers.

  I'm not sure how well gimp deals with layers of different size when you then 
export the image as a png.  That is to say, if the image itself is 64x64 (say a 
big monster), but the chosen layer is only 32x32, if you save as a png, do you 
get a 32x32 image or 64x64?

  Also, can you copy layers between images?  If so, that does mean to some 
extent that you don't necessarily need to put everything in one one image - you 
just copy the layers as needed.

  In fact, it is probably pretty clear the case that putting everything in one 
image does not make much sense - the overlays for monsters are almost certainly 
different than overlays for walls (monsters would be items, walls would be 
things like cracks (weak walls), or perhaps moss or other effects).

> This concept may not be practical for open development environments,
> but it did seem to imply some side-effect benefits could be realized
> by embracing use of layers.
> 
> 1) Graphical element re-use.  Not a primary rationale for layered
>    files, since copy/paste from the png format is possible, but also
>    not without plausible benefit.
> 
>    Simple "effect overlays" now possible.  Consider Leaf's old idea
>    about adding more images with scorch marks...  Why redo the art
>    from scratch?  Simply make a scorch mark overlay...
> 
>    Similarly, a "weak" version of a wall, for instance, might easily
>    be made with an overlay layer that would turn on/off.
> 
>    Animation sequences would all be in the same file for quick/easy
>    reference where the animation is created by a methodical turning
>    on/off of layers.

  Yep - all that sounds good.

> 
> 2) Size consistency.  When "thinking out loud" on IRC today,  I made a
>    comment about producing more art.  Ryo asked if I meant things like
>    plates, etc.  That, and the wiki page:
> 
>      http://wiki.metalforge.net/doku.php/dev_todo:gfx_needing_fixing
> 
>    gave me an idea.  What if, for example, all food was in one source
>    file, that contained dinnerware?  It could help to keep guide the
>    development of consistent sizes for similar objects.

  I think size consistency is a hard thing to do in crossfire, especially given 
the size of images.  If a character is 32x32 pixels, that means a carrot should 
be somehint like 1x4 pixels for proper scale, but an image that small really 
isn't very useful.

  If I had a choice between images being in proper scale but hard to 
distinguish, or images being of incoherent scale but easy to distinguish, I'd 
take the easy to distinguish any day.

  As an aside, when images were redone at one point, it was found certain images 
on certain backgrounds are hard to see (most weapons are grey, and a fair number 
of dungeon backgrounds are also grey).  There were are least a few people to 
bring up that it was hard to see them, and wanting them more visible.  My take 
was more that maybe grey items on grey background should be hard to see, but 
more that it would be very hard to make every possible item visible on every 
possible background.

  The point here is more that if images are changed so that they are less easy 
to see, I think you may get some number of folks complaining.

> 
> 3) Templates.  For items where perspective or other attributes are
>    important, template layers could assist the artist.  Relevant
>    reference layers could inserted into the source file, not for
>    display when exporting to png, but for future re-work.
> 
>    Template layers might not be saved with each work, but might
>    reside in special template source files.

  Thats reasonable.  Note in the case of base (starting images), a function of 
layers isn't really needed.  An example could be a generic humanoid with the 
different facings - while easier to do in a single file with layers, I'd also 
say that is more a starting point - once someone started making a new monster, 
there probably isn't a reason to keep the template layers in the image they are 
working on - rather they may just modify the different templates (OTOH, it may 
be easier to keep that template and draw over it).

  I certainly think templates would be useful.


> 
> Seeing the banter on the simple saving of layered sources and thought
> about deleting prior works does make one wonder if it is futile to
> think that such an organized plan for producing graphics can
> succeed when people appear/disappear from the project - not a
> criticism as much as an admission that the above ideas might be a bit
> too far on the side of being a utopian idealist.

  That's sort of why I was trying to sort out rules/standard for this. That 
makes it easier for a future person to pick things up if things were done in a 
somewhat standard method.  If things were done more at random, much harder for 
someone to pick up work.

As an aside, if someone actually wants to make graphics, I'd live graphics for 
the following:
1) The different skills (one handed weapons, 2 handed, praying, etc)
2) The different protections/attacktypes (fire, cold, poison, magic, etc)

  With icons, instead of current text in the client, which is too long, a simple 
icon could be used.  These could perhaps also get used in the game somehow, 
perhaps to denote training areas for the skills, or as symbols on the floor.