Yes, the problem is when the curves fan out larger than the tile size or squish to unreasonably small sizes. You cannot simply rely on a UV divisioning, which is the quick and dirty. When spans become too large, you have to start adding in new courses. 
The method I was sketching earlier looked AWFUL. You couldn’t tell that it was supposed to be a radial pattern at all. I won’t even share a screenshot 
But this is something more like the inspiration image, and keeps all the tiles in nice neat rows which will ensure the mason’s sanity. I’d leave it up to them to divide each row into tiles. They’ll ignore your drawing if you try and tell them anyway 
Segment and tile width are adjustable
NOTE: To make sure the segments and offsets go in the right direction, each source surface must have the same edge order. Easiest way to ensure this is to copy the first one and edit the CPs to make it fit. Otherwise you have to do a bunch of filtering and sorting and flipping and transposing which is just a load of tiresome overhead (that I do not fancy implementing right now!)
mosaic rows.gh (45.1 KB)
Nice!
I think this could maybe be optimized by looking at which side you offset from produces less cutoffs Having two cut courses coming together would look horrendous.
nice
Hello
not sure it is is useful but as giving this thread I made a little tool that Pack (not far for nesting) shapes using the direction given by a Field.
For the field I use you curves but as there are some places without curves I also use the bounding curves.
Here some results
Calculation was not finished, it tooks 8 min ! With 1 000 000 (one millions) of test.
Field calculation is surely a bit slow with the curves.
Here the GH that shows the field …, if interested I can publish more rapidly an update of Nautilus.
I internalise result so no need of plugin.
test mosaic.gh (773.2 KB)
Yes field calculation is not enough fast. With other type of field, it takes less than 1 min for the same calculations.
1min42 s for this one ( 31 000 bricks and ~4 M pixels)
That’s very nice. I love the way the density ebbs and flows. An absolute nightmare to set out in the real world mind you! I think you’d have to print out 1:1 templates, or invent a mosaic robot …
I agree in some domain lets artist or artisan do what they are good at. Like in Asterix and Obelix “Mosaic takes Time” !!
I love what you have done here @laurent_delrieu , and I agree @Tom_Newsom that this would be a huge task to undertake in the real world. But as a designer, I am very keen to see how closely our work in GH can be translated into a real-world application for mosaics. That said, the final result will depend on the tile artisan’s finesse, and their effort depends on me being able to translate the idea in my head into a very effective drawing or sketch.
So far, a combination of all the work on this forum has helped me get this result and has pushed me forward, helping me make better design decisions. I really appreciate all the help! I will continue looking into how we want to execute this design and its overall feasibility.
Thank you so much, everyone!
Looks great Laurent. I am thinking if I were to make this myself, I would map the ‘tile’ segments to a unit length, so the entire work could be done with a minimum various size pieces. Similar to a script you published a few years ago a ‘paint by number’s’, except this would be ‘tile by lengths’. Great work