I’ve “finished” (in reality I think this is more like the first shot) mapping my jacket, but it’s turned out a bit weird. Here’s the mapping screen. This looks good to me. The map starts at the left of the front zip as you’re wearing it, you can see the cut for the left arm, the back and collar, the cut under the right arm and then back around to meet the front zip from the right:
Here it is playing the built-in “Red-Green XY 2D Sweep” pattern:
The horizontal (red) sweeps look good, but the green vertical sweeps look like they’re being “squashed” in under the underarms. The bright part hits the top pixels of the underarms at the same time it hits the pixels at the collar. I would have expected the sweep to go through the whole (0.0, 0.0) → (1.0, 1.0) “world” and just not be visible in the arm cuts. i.e., the pattern plays out across the whole world and the mapping just dictates the window through which you can see it. Isn’t that the point or do I misunderstand something? Or is there something wrong with my mapping somehow?
It looks like it might be upside down maybe. Try reversing the order of the strips vertically?
1 Like
You mean moving the far left strip to the far right, or setting the bottom left pixel to coordinates [0, 0] rather than [0, 30]?
I wondered about this, but it doesn’t make sense to me that it could be upside down given everything looks the right way up on the mapping screen.
I agree that it looks like the mapping is upside down. I make this mistake often. You have to make sure the wiring order matches the pixel order in the mapper. My recommendation would be to make the top part of your map the bottom edge of your jacket to see if that’s the case.
So it wasn’t “upside down” so much as the rows going the wrong direction—I needed to go up-down-up-down, but it was going down-up-down-up.
I got very confused by the idea that PixelBlaze’s “origin” is nominally at the top left, but the zeroth pixel on my piece is at the bottom left. I had vibe-coded a mapping tool with Claude and went backward and forwards three or four times on where the origin should be for everything to map correctly and the mapping to look the “right way up” in the PixelBlaze interface but actually getting that right caused my problem. It took Claude Opus 4 to fix it 
.
Now my issue is patterns that don’t/won’t work (or are hella slow) on such a big matrix…
