Can the ModelTiepointTag help you determine alignment?
✅ Yes — if you read it correctly.
The ModelTiepointTag defines a pixel-to-world coordinate mapping:

text
Copier
Modifier
[ i, j, k, X, Y, Z ]
i, j, k: pixel coordinates in the raster (usually 0,0,0)

X, Y, Z: corresponding world coordinates (usually in map projection units, like UTM)

From this and the ModelPixelScaleTag, you can derive the affine transform:

js
Copier
Modifier
originX = X
originY = Y
pixelSizeX = scale[0] // from ModelPixelScaleTag
pixelSizeY = -scale[1] // usually negative (north up)
So if both bands:

Have the same tiepoint (i.e. same upper-left world coords), and

Have the same pixel scale (resolution),

Then yes, they should align exactly in map space — even if their pixel dimensions differ.

✅ This means that cropping a central window (defined in map space) will give you aligned imagery, even if the original arrays are different shapes.

🔍 How to check alignment using tiepoints (pseudo-JS):
js
Copier
Modifier
const tie1 = image1.getTiePoints()[0];
const tie2 = image2.getTiePoints()[0];

const scale1 = image1.getFileDirectory().ModelPixelScale;
const scale2 = image2.getFileDirectory().ModelPixelScale;

const sameOrigin = tie1.x === tie2.x && tie1.y === tie2.y;
const sameScale = scale1[0] === scale2[0] && scale1[1] === scale2[1];
If both sameOrigin and sameScale are true, then you can safely align them using geotransform math.