How Does Image Size and Pixel Position Work When We Only Have a List of Pixels?

[Rajeev Bagra]

Context

In many introductory image-processing assignments (such as MIT 6.100L Problem Set 5), we are asked to write a function like:

filter(pixels_list, color)

Here, pixels_list is provided as a simple list of RGB pixels:

[(0,0,0), (255,255,255), (38,29,58), ...]

The function applies a color-vision deficiency filter (red, green, blue, or none) by multiplying each pixel with a 3×3 transformation matrix.

However, this raises a natural question.

---

❓ The Question

If pixels_list is just a list of pixels, how does the program know:

1. The size (width × height) of the image?
2. The position (row and column) of each pixel in the image?

---

✅ The Key Idea (Short Answer)

The filter() function does not know the image size or pixel positions — and it does not need to.

This is intentional design.

---

🧠 Why filter() Doesn’t Need Image Size or Position

1️⃣ Filtering is a per-pixel operation

The color filter:

• Works on one pixel at a time
• Does not depend on neighboring pixels
• Does not change pixel order

Mathematically:

new_pixel[i] = transform(old_pixel[i])

So whether a pixel is in the top-left or bottom-right of the image does not matter for color transformation.

---

2️⃣ Pixel position is stored implicitly by list order

When pixel data is originally extracted from an image:

pixels_list = list(img.getdata())
size = img.size   # (width, height)

• Pixels are stored in row-major order
• The order of pixels in the list represents their positions
• Index i always corresponds to the same (x, y) position

Example (3×2 image):

Row 0: P0  P1  P2
Row 1: P3  P4  P5

Stored as:

[P0, P1, P2, P3, P4, P5]

---

3️⃣ filter() preserves this order

Inside filter():

for pixel in pixels_list:
    new_pixels.append(transformed_pixel)

• Pixel order is unchanged
• Index i in → index i out

No rearrangement occurs.

---

4️⃣ Image reconstruction restores geometry

Later, another function (e.g. pix_to_img) rebuilds the image:

img = Image.new(mode, size)
img.putdata(filtered_pixels)

Here:

• size tells Pillow how many pixels per row
• putdata() maps pixels back to their original positions using order

---

🔍 Why This Design Is Intentional

This separation teaches an important computer-science principle:

One function, one responsibility

• filter() → color transformation only
• Image size & geometry → handled elsewhere

This mirrors real-world image-processing pipelines.

---

🧩 Simple Analogy

• pixels_list → a sequence of paint buckets
• filter() → a machine that changes paint color
• Image size → the blueprint used later to paint the wall

The color machine doesn’t need the blueprint.

---

✅ Final One-Line Answer

The filter() function does not determine image size or pixel location because it only transforms pixel colors; pixel positions are preserved implicitly by list order and restored later using the original image dimensions.

---

[Author]

Posts