NumPy: Brighten without overflow

Adding to uint8 pixels wraps around. Do it the safe way, as np.clip(img.astype(np.int16) + d, 0, 255) would, and count how many pixels hit a limit.

What to build

A program that changes an image's brightness with clipping.

Requirements

  • Add d to every pixel and clip the result to the range 0 to 255.
  • Print the new image in NumPy's integer format, then clipped N, the number of pixels whose raw sum was below 0 or over 255.

Input

Line 1: d. Line 2: h and w. Then h rows of w pixels.

Output

h + 1 lines.

Constraints

−300 ≤ d ≤ 300; 1 ≤ h, w ≤ 20; 0 ≤ pixel ≤ 255.

Example

Input:

50
2 3
0 100 250
200 210 5

Output:

[[ 50 150 255]
 [250 255  55]]
clipped 2

Why it works

Doing the addition in a wider type keeps the true sums, and clipping then maps every out-of-range value to the nearest valid brightness instead of wrapping it around to the other end.

Examples

Example 1 · Input
50
2 3
0 100 250
200 210 5
Expected output
[[ 50 150 255]
 [250 255  55]]
clipped 2
Example 2 · Input
-20
1 3
10 20 30
Expected output
[[ 0  0 10]]
clipped 1

Prompts inside input() are ignored when checking, so input("n: ") is fine.

Hints

Hint 1

Add in ordinary Python integers, which never wrap, then limit the result to 0–255.

Hint 2

A pixel counts as clipped when the raw sum is below 0 or over 255.

Solution

Show one correct solution (try it yourself first)
def fmt2(m):
    w = max(len(str(v)) for row in m for v in row)
    rows = ["[" + " ".join(str(v).rjust(w) for v in row) + "]" for row in m]
    return "[" + "\n ".join(rows) + "]"


d = int(input())
h, w = map(int, input().split())
img = [[int(x) for x in input().split()] for _ in range(h)]
clipped = 0
out = []
for row in img:
    new = []
    for v in row:
        s = v + d
        if s < 0 or s > 255:
            clipped += 1
        new.append(min(255, max(0, s)))
    out.append(new)
print(fmt2(out))
print("clipped", clipped)

Lesson: NumPy in practice: images as arrays →