Most appropriate component
Image (p5.Image, image, etc.)
Describe the bug
When calling copy() or blend() on a p5.Image instance that has pixelDensity > 1 (e.g., Retina / high-DPI displays), the destination coordinates (dx, dy, dw, dh) are not scaled by the destination image's pixel density.
In p5.Image.prototype._copyHelper (lines 920-960), the source coordinates are scaled by s = srcImage.canvas.width / srcImage.width, but the destination parameters are passed directly as raw logical values to dstImage.drawingContext.drawImage().
Because dstImage.drawingContext operates directly on the physical canvas of dstImage without an automated DPI transformation matrix, drawImage() draws into only [dx, dy, dw, dh] physical canvas pixels instead of [d*dx, d*dy, d*dw, d*dh].
For pixelDensity(2), this results in the copied image covering only 1/4th of the intended area in the top-left corner, leaving the remaining 75% of the intended target area completely untouched (transparent / blank). Because p5.Image.prototype.blend() delegates to copy(), it suffers from the identical issue.
Steps to reproduce
let src = createImage(50, 50);
src.loadPixels();
for (let i = 0; i < src.pixels.length; i += 4) {
src.pixels[i] = 255; // Red
src.pixels[i + 3] = 255; // Alpha
}
src.updatePixels();
let dst = createImage(100, 100);
dst.pixelDensity(2); // canvas backing store is 200x200
dst.copy(src, 0, 0, 50, 50, 0, 0, 50, 50);
// Sample a point in the bottom-right quadrant of the logical image
let col = dst.get(35, 35);
console.log(col); // [0, 0, 0, 0] instead of [255, 0, 0, 255]
Expected behavior
The destination coordinates should be scaled by d = dstImage.canvas.width / dstImage.width so that copy() and blend() properly fill the intended logical region on high-DPI images.
Most appropriate component
Image (p5.Image, image, etc.)
Describe the bug
When calling
copy()orblend()on ap5.Imageinstance that haspixelDensity > 1(e.g., Retina / high-DPI displays), the destination coordinates (dx, dy, dw, dh) are not scaled by the destination image's pixel density.In
p5.Image.prototype._copyHelper(lines 920-960), the source coordinates are scaled bys = srcImage.canvas.width / srcImage.width, but the destination parameters are passed directly as raw logical values todstImage.drawingContext.drawImage().Because
dstImage.drawingContextoperates directly on the physical canvas ofdstImagewithout an automated DPI transformation matrix,drawImage()draws into only[dx, dy, dw, dh]physical canvas pixels instead of[d*dx, d*dy, d*dw, d*dh].For
pixelDensity(2), this results in the copied image covering only 1/4th of the intended area in the top-left corner, leaving the remaining 75% of the intended target area completely untouched (transparent / blank). Becausep5.Image.prototype.blend()delegates tocopy(), it suffers from the identical issue.Steps to reproduce
Expected behavior
The destination coordinates should be scaled by
d = dstImage.canvas.width / dstImage.widthso thatcopy()andblend()properly fill the intended logical region on high-DPI images.