Beach, Orchard, Mud metatile alignments, and reducing symbol size for allotments

[polarbearing]

Summary:

Providing a fix for #299 - this changes the symbols for beach and orchard,
as well as mud.png exactly the same style, that would clear all 10px cases in #937
allotments.png just reduced in size without visual change.

Details:

For Beach, Orchard, Mud:
To maintain pixel-crispiness while bending to the Power of Two the density of the pebbles
has been reduced from 4+1 pixels to 7+1 pixels.

This follows the opinion in fixing the allotments in #770, where the wider style was accepted as
adding more clarity to the map.

New tiles:

edit: Adding 3+1 style for direct comparison, from discussion below:

Examples of current metatile artefacts:
http://www.openstreetmap.org/#map=18/52.37324/12.87641

Resolves #299.

[pnorman]

#770 was 6px to 8px, this is 5px to 8px. Could you throw together a 4px based one?

[polarbearing]

@pnorman - 4px won't hold the 7+1 pattern style (7px yellow beach, 1px grey pebble).
Allotments are currently 16x16, but could be cut to 8x8.

I have added the mud.png exactly the same style, that would clear all 10px cases in #937

[polarbearing]

Actually allotments came from 30x6 to 16x16, and is now cut to 8x8 as well. Checked in my tilemill, works fine.

[pnorman]

@pnorman - 4px won't hold the 7+1 pattern style (7px yellow beach, 1px grey pebble).
Allotments are currently 16x16, but could be cut to 8x8.

I was after a look at 4px repeating instead of 8px repeating.

[polarbearing]

This is an example of beach in 4x4 px, i.e. 3 px yellow and 1 px pebble. Hurts my eyes.

I have added the example in the first post for direct comparison -- you see that the higher percentage of grey pebbles gives the beach a darker, dirtier look.

[imagico]

In case you want to try for the beaches - here is a pixel aligned irregular pattern with the same point density as a 5x5 pixel grid pattern:

[polarbearing]

How does the pattern symbol size influence the performance? Should we keep them as small as possible, or maybe vice versa?

For the random effect to be maintained over a certain area you need a minimum size of the seed tile.
I have tiled up @imagico's example 8x8 for a 1028x1028 image, and looking closely you still recognise the pattern repetition. (view in 100% size, git scales to column width)

[matkoniecz]

I am not fan of irregular patterns - but especially for orchards it makes no sense. This regular dot style is fitting, because typically trees are planted in regular patterns.

[polarbearing]

I agree here with @mkoniecz, though in organic production you also have the idea of meadow orchards with scattered trees:
http://en.wikipedia.org/wiki/Orchard#Meadow_orchard_.28Streuobstwiese.29

Anyway, we have the discussion about randomising in #938.

I'd propose to fix the regular pattern first so we can close the issues in #937. Once this PR gets merged i'd volunteer for squaring the other tiles.

[imagico]

As i wrote i meant the irregular pattern for beaches, not for orchards. Repetition is not an issue, it is easily possible to produce such pattern in larger - the tool i linked in #938 can be used for that.

But i understand regular grid patterns have a long tradition in the standard OSM style and introducing something different would be a big step. On the other hand beaches would be a nice test case to try this out and collect some opinions from people on the general concept.

[matthijsmelissen]

You're also changing mud and alotments. Is that intentional? If so, could you change the title of the pull request?

[polarbearing]

@math1985
yes, beach/orchard/mud is changed with slight visual impact, as in the first post,
fixing #299 and clearing all 10px cases in #937
allotments.png just reduced in size from 16x16 to 8x8 without visual change, to have these pattern tiles consistent.
Title updated, added summary to first post.

[matthijsmelissen]

I have tested this PR, and I think it can be merged.

[matthijsmelissen]

@polarbearing Small suggestion, if you include 'resolves #299' / 'closes #299' / 'fixes #299' in your PR description, Github will automatically close the issue once the PR is merged. I changed your PR description to add it now.

[matkoniecz]

@math1985 According to https://help.github.com/articles/closing-issues-via-commit-messages it is necessary to modify commit message, not PR description.

[matthijsmelissen]

PR message works as well nowadays: https://github.com/blog/1506-closing-issues-via-pull-requests