Tiled walls

[RandallDavis]

Ken's idea:

Basic premise is that you can have several size categories for an asset theme depending on the length of the wall to prevent stretching. For unclean divisions it'll find a smaller sized asset if one exists and fit it, or at least the next smallest. The end result being that we're minimizing fidelity loss for long walls and attempting to preserve aesthetics.

|========]|========]|========][++++]

Off the top of my head I'm guessing there would need to be a hierarchy for assets. Wood->wood 2x->wood 4x->wood 8x Stone->stone 2x->stone-4x->wood 8x

Known lengths of the images, and other specific information as some assets might not allow variable size.

[RandallDavis]

My suggestion on implementing this is that we change the way that all global assets are structured so that asset drawing remains agnostic to the type of asset.... feel free to modify / expand on this, but here are my thoughts:

• Asset state objects get a new optional field called 'segmentLength' appended to the end, which is the number of pixels of length it has (this would only be used in createTokenObjectFromAssetBySegment()?). If the field has no value, then it means that this asset can be stretched, but if there is a value, then this asset is exclusively drawn by the specified length.
• Global assets should be made into collections (for example: each floorAsset would become a single-element list of objects instead of just being an object; each wallAsset would become a 2-element list of lists of objects instead of a 2-element list of objects). Each list would have 1 or more assets, with exactly one that has no segmentLength.
• If possible, these asset changes should be made without destroying anyone's existing asset state (see how I did this with https://github.com/RandallDavis/roll20-areaMapper/blob/master/areaMapper.js#L99).
• In createTokenObjectFromAssetBySegment(), loop until the entire segment is drawn: drawing the longest segmentLength asset that can fit in the remaining space and then finishing off with a stretched asset that has no segmentLength.

[RandallDavis]

Discussion point:

It'll probably be pretty daunting to incorporate this into the asset management user interface... what are your thoughts on that? Should it just be that only assets with no segmentLength surface up into the UI (since exactly one is required for every asset list anyway), and any user-created assets are simple cases like we have no with no segmentLength? Should we instead incorporate all of this and let users declare which asset is stretched and have a sub-cycling mechanism for editing assets within an asset list?

[RPGKenL]

I haven't done a deep dive on much of the logic as it being the 4th flew over my head, and Sunday I have appointments I need to keep. From a brief skim as well as the points you've outlined here:

• I agree about the optional field 'segmentLength', for the image length. I'm unclear about the exact implementation but from a high level standpoint, keeping it agnostic is always a virtue to keep (in any software development).
• For global assets, I'm unclear about the exact structure as I'm a bit confused as to why there's an array implementation for what appears to be property access that indexes into another array. I can't clearly comment on this until I deep dive on this.
• I agree, but again I'll need to read a bit more thoroughly.
• Exactly regarding createTokenObjectFromAssetBySegment() from a premise standpoint on using the longest. For the remaining space I'm not sure about stretching the 'unknown' length segment as I feel that a 'best fit' solution feels more sound. More specifically there's a slight tie between 'shrinking' and 'stretching' this 'best fit' segment. I'm imagining a simple absolute difference test between the known lengths of the tiled segments to determine which should fit in the remaining space.

On the discussion point:

I really like how you have a simple select+add graphic for simplicity of operation, but for tiled walls (and perhaps down the road tiled floor assets) there will need to be more input suggestions. Off the top of my head I'd stick in the link a command that prompts the user for dimensions with the default measurements being 0, where 0xN Nx0 are invalid inputs, and 0x0 is free-form scalable.

[RandallDavis]

Everything sounds great so far. :)