Thanks folks!

@Fersus - Yes, I wondered about a hexagonal grid. However, over the past year, I've been involved in discussions about the use of movement grids in RPGs more generally elsewhere, and the consensus has been almost exclusively for square grids (or occasionally none at all). They're such a feature of most published dungeon maps, it's hard to break away from that, I suspect. Plus because hexes don't scale equally in all dimensions, a square grid works better for estimating distances for the GM by-eye.

Given the whole point about participating in the Atlas, and contests like this too, is to try different things, at least from time to time, my intention was to try to make each of the ten Palace maps different. So for the second one, I picked the Old School Blue style from CA12 (also known as the Create Your Own Style pack). Blue, ice, seemed apt!

For continuity, I've decided to stick with the Mason Serif Bold font that comes with CC3+, however.

Along the way, I let myself get a bit distracted, and decided a CA3 portrait of the Winter Queen might be an interesting addition to the map set, with a hint of the SS2 Bitmap A Snow Frozen Lake fill texture for the border design, which thanks to the standard CA3 bevel effect, looks rather like fine marble now:

She's intended to be between 8 and 9 feet tall in (imaginary) reality, 2.4-2.7 metres.

So having chosen my ten snowflake templates, it was time to test things out using CC3+. Summer Palace 1 was drawn using DD3, with the snow fill from SS2 Bitmap A as the backdrop:

The Palace size doesn't encourage adding too much interior detail, as it starts to look cluttered very quickly, so I settled for showing simply the major, fixed features instead. There'll be text notes and a PDF to accompany all the maps in the final Atlas version, as normal. Each Palace version is intended to have the same seven main areas (Hall of Mirrors is one of the seven, incidentally), though not always in the same order - it is meant to be the same place, just with a different appearance each day, after all.

I was a little concerned about the narrowest passages, for all I'd checked with the grid on as I was drawing it - the narrowest are between four and five feet wide in the connecting passages. I decided early on that it wasn't going to be necessary to ensure access to every nook and cranny that didn't require direct access. It is meant to be a magically-grown, natural creation, and when real snowflakes can be so complex, that seemed an ideal recommendation to follow!

There was a degree of experimentation in this, as I wasn't sure how best to work the complex outlines at first, as you may find when you have access to the FCW file later - such as several overlapping floor segments in places. As I was just using the normal straight polygon drawing tool though, not fractal, I probably shouldn't have been too concerned. That's easy to say after the event though ?

When I started thinking about this, because all my previous Community Atlas mapping has involved a degree of random design, I started looking at random snowflake creation systems online. This is one design I made from the Misha Studios site run by Misha Heesakkers for instance:

However, this generates only an SVG file if you're using Chrome or Firefox, which is not ideal.

Another interesting site, Snowflake Generator by Fabian Kober creates fractal, fully adjustable, PNG download files, such as this:

While these are fascinatingly wonderful - and the fractal versions can be incredibly intricate - they're also perfectly symmetrical, which wasn't really what I wanted, so I started searching for images of real snowflakes online. There are a lot of these! However, when I struck upon the many nicely contrasty black and white photos by Wilson Bentley (1865-1931; Wikipedia link), I decided to make my selections for this project chiefly from those. Wikimedia Commons has a lot of options, for instance.

I thought it might be useful to give the two image generators here though, as they may be useful for those wanting to create symmetrical snowflake design mazes or labyrinths, for example.

For the random element in the maps, I opted to stick with that being primarily in where the palace can appear, and in what form.

The basic premise for this set of maps is that of a wandering sub-surface palace, which is occupied by/part of a powerful humanoid creature, possibly a deity (nobody's very sure). The palace magically shifts elsewhere every day, and changes its form at the same time. It is composed entirely of ice, and has the shape of a gigantic snowflake. It can only appear in an iceberg, in pack ice, an ice cap, or a glacier, or in a huge snow-cloud high in the air. There is always only a single entrance on the surface, surrounded by delicately intricate ice-sculptures of summer flowers and foliage (yes, even in the cloud); the rest of the palace is buried and completely hidden at all times (again, yes, even in the cloud!). The palace fits within an area at most roughly 600 feet by 500 feet, though its size changes from day to day.

My idea is to prepare a location map showing ten potential places the palace may appear, thus giving a simple random option of where it may be using a 1D10 roll. Another 1D10 roll will give the form and size the palace has that day, so I'm working towards having ten different versions of the palace map. The following map is an early version of the locations map, to give an idea of what I'm meaning:

For this, I've basically reworked Shessar's beautiful world map into a very simplified version using one of Monsen's recent blog posts about representing subtly inset features on CC3+ maps. The numbered snowflake symbols (Wingdings font letter "T"s) indicate the potential palace locations. These are not going to be defined more precisely than this, so if the random roll comes up twice the same for two separate days, this just means the palace has moved, but not beyond the general area indicated - so "6" again means it's still somewhere in southern Ezrute, for example. It'll be up to the GM to decide exactly where in the general areas the palace actually is. This is not the final version, which will be a bit larger, and have more information; I've adjusted a couple of the markers since this version was prepared as well.

The 1 & 9 markers are for icebergs/pack ice somewhere on the great northern and southern oceans; the 2 & 10 ones are for the aerial snow-cloud versions, which will be found exclusively somewhere high above those same two oceans.

The palace and its occupier are intended to be things that might be useful to find for very esoteric information, but which are nearly impossible to guarantee locating at all quickly, even by those who know of their existence. I thought this might be an interesting addition for Nibirum more generally.