Fix: 'ImageDraw' object has no attribute 'textsize' (Pillow 10+)

Quick fix

Replace w, h = draw.textsize(text, font=font) with:

left, top, right, bottom = draw.textbbox((0, 0), text, font=font)
w, h = right - left, bottom - top

For font.getsize(text), use font.getbbox(text) the same way, or font.getlength(text) if you only need the width.

What the error looks like

label.py

from PIL import Image, ImageDraw, ImageFont

img = Image.new("RGB", (400, 120), "white")
draw = ImageDraw.Draw(img)
font = ImageFont.truetype("DejaVuSans.ttf", 40)

w, h = draw.textsize("Hello, Pillow", font=font)

Tested on ✓ Linux ✓ Windows ✓ macOS

Traceback (Pillow 12.3.0)

AttributeError: 'ImageDraw' object has no attribute 'textsize'

The font method fails the same way:

label.py

from PIL import ImageFont

font = ImageFont.truetype("DejaVuSans.ttf", 40)
w, h = font.getsize("Hello, Pillow")

Tested on ✓ Linux ✓ Windows ✓ macOS

Traceback (Pillow 12.3.0)

AttributeError: 'FreeTypeFont' object has no attribute 'getsize'

Why it was removed

textsize() returned one width and one height, which hid two things: where the ink actually starts (text rarely starts exactly at the point you draw it), and how wide the text is for layout purposes (its advance, which includes side bearings and can differ from the ink width). The newer methods return each of these explicitly.

Version	Change
8.0.0 (2020)	`textbbox()`, `multiline_textbbox()`, `textlength()`, `FreeTypeFont.getbbox()` and `getlength()` added.
9.2.0 (2022)	`textsize()`, `multiline_textsize()`, `getsize()`, `getoffset()` and `getsize_multiline()` deprecated.
10.0.0 (1 July 2023)	All of them removed.

Replacement for every removed method

Removed	Use instead
`draw.textsize(text, font)`	`draw.textbbox((0, 0), text, font=font)` for the ink box, or `draw.textlength(text, font=font)` for the width
`draw.multiline_textsize(text, font)`	`draw.multiline_textbbox((0, 0), text, font=font)`
`font.getsize(text)`	`font.getbbox(text)` or `font.getlength(text)`
`font.getoffset(text)`	the first two values of `font.getbbox(text)`
`font.getsize_multiline(text)`	`draw.multiline_textbbox((0, 0), text, font=font)`

textbbox vs textlength: what the numbers mean

measure.py

from PIL import Image, ImageDraw, ImageFont

img = Image.new("RGB", (400, 120), "white")
draw = ImageDraw.Draw(img)
font = ImageFont.truetype("DejaVuSans.ttf", 40)
text = "Hello, Pillow"

left, top, right, bottom = draw.textbbox((0, 0), text, font=font)
print("textbbox:  ", (left, top, right, bottom))
print("ink size:  ", right - left, "x", bottom - top)
print("textlength:", draw.textlength(text, font=font))

Tested on ✓ Linux ✓ Windows ✓ macOS

Output

textbbox:   (0, 8, 241, 43)
ink size:   241 x 35
textlength: 240.578125

Read the output like this:

textbbox is the rectangle the ink covers, relative to the point you passed. The non-zero top means the tallest glyph starts below that point, because the default anchor is the top of the ascender line, not the top of the letters.
textlength is the advance width: how far the pen moves. Use it to place the next word or to wrap lines. It is a float with 1/64 px precision.

Diagram of the word Typography with its textbbox in red, the advance width in blue and the drawing point in green — Real output of the script below: the red box is `textbbox`, the blue line is `textlength`, the green dot is the `xy` you pass to `draw.text()`.

bbox_diagram.py

from PIL import Image, ImageDraw, ImageFont

font = ImageFont.truetype("DejaVuSans.ttf", 64)
label = ImageFont.truetype("DejaVuSans.ttf", 18)
text = "Typography"
img = Image.new("RGB", (620, 190), "white")
draw = ImageDraw.Draw(img)

x, y = 48, 60
draw.text((x, y), text, font=font, fill="#0f172a")
box = draw.textbbox((x, y), text, font=font)
draw.rectangle(box, outline="#dc2626", width=2)
adv = draw.textlength(text, font=font)
draw.line((x, y - 18, x + adv, y - 18), fill="#2563eb", width=3)
draw.ellipse((x - 5, y - 5, x + 5, y + 5), fill="#059669")

draw.text((x, 18), f"textlength = {adv:.1f}", font=label, fill="#2563eb")
draw.text((x, box[3] + 14), f"textbbox = {box}", font=label, fill="#dc2626")
draw.text((x - 10, y), "xy", font=label, fill="#059669", anchor="rm")
img.save("bbox.png")

Tested on ✓ Linux ✓ Windows ✓ macOS

A drop-in replacement for textsize()

If you are upgrading a large codebase, add one helper and replace calls mechanically:

compat.py

from PIL import Image, ImageDraw, ImageFont


def text_size(draw, text, font, **kwargs):
    """(width, height) of text, as the removed draw.textsize() returned."""
    left, top, right, bottom = draw.textbbox((0, 0), text, font=font, **kwargs)
    return right - left, bottom - top


draw = ImageDraw.Draw(Image.new("RGB", (1, 1)))
font = ImageFont.truetype("DejaVuSans.ttf", 40)
print(text_size(draw, "Hello, Pillow", font))
print(text_size(draw, "Two\nlines", font))   # multiline works too

Tested on ✓ Linux ✓ Windows ✓ macOS

Output

(241, 35)
(94, 71)

Sizes can be a pixel or two different from what textsize() returned on Pillow 9, because the bounding box measures the actual ink. That is usually an improvement, but check pixel-exact layouts after upgrading.

If you only have a font object (no image), use the font methods directly:

font_metrics.py

from PIL import ImageFont

font = ImageFont.truetype("DejaVuSans.ttf", 40)

# No Image or Draw object needed:
print(font.getbbox("Hello, Pillow"))     # replaces font.getsize() / font.getoffset()
print(font.getlength("Hello, Pillow"))   # advance width, for layout and wrapping

Tested on ✓ Linux ✓ Windows ✓ macOS

Output

(0, 8, 241, 43)
240.578125

The centring bug after switching

The most common regression: code that centred text using the old size now draws it slightly too low. The bounding box has a top offset, and using only bottom - top as the height ignores it. Here are three approaches side by side, rendered by the script below:

Three panels: red text below centre, blue text centred by anchor mm, green text with its ink exactly centred — Real output. 1: centred by size only (too low). 2: `anchor="mm"`. 3: bbox offset subtracted, ink exactly centred.

centre_text.py

from PIL import Image, ImageDraw, ImageFont

font = ImageFont.truetype("DejaVuSans.ttf", 56)
small = ImageFont.truetype("DejaVuSans.ttf", 16)
text = "Pillow 10"
W, H = 360, 200


def panel(title):
    img = Image.new("RGB", (W, H), "#f1f5f9")
    d = ImageDraw.Draw(img)
    d.line((W / 2, 0, W / 2, H), fill="#94a3b8")
    d.line((0, H / 2, W, H / 2), fill="#94a3b8")
    d.text((10, 8), title, font=small, fill="#475467")
    return img, d


# 1. Wrong: uses the bbox size but ignores its top offset, so the text sits low.
a, d = panel("1. size only: too low")
l, t, r, b = d.textbbox((0, 0), text, font=font)
d.text(((W - (r - l)) / 2, (H - (b - t)) / 2), text, font=font, fill="#dc2626")

# 2. anchor="mm": centres the line box (ascender to descender).
m, d = panel('2. anchor="mm": line box centred')
d.text((W / 2, H / 2), text, font=font, fill="#2563eb", anchor="mm")

# 3. Subtract the bbox offset: centres the visible ink exactly.
k, d = panel("3. bbox offset: ink centred")
d.text(((W - (r - l)) / 2 - l, (H - (b - t)) / 2 - t), text, font=font, fill="#059669")

out = Image.new("RGB", (W * 3 + 40, H), "white")
for i, p in enumerate((a, m, k)):
    out.paste(p, (i * (W + 20), 0))
out.save("centre.png")
print("top offset of the bbox:", t, "px")

from PIL import Image, ImageDraw, ImageFont

font = ImageFont.truetype("DejaVuSans.ttf", 56)
small = ImageFont.truetype("DejaVuSans.ttf", 16)
text = "Pillow 10"
W, H = 360, 200

def panel(title):
    img = Image.new("RGB", (W, H), "#f1f5f9")
    d = ImageDraw.Draw(img)
    d.line((W / 2, 0, W / 2, H), fill="#94a3b8")
    d.line((0, H / 2, W, H / 2), fill="#94a3b8")
    d.text((10, 8), title, font=small, fill="#475467")
    return img, d

# 1. Wrong: uses the bbox size but ignores its top offset, so the text sits low.
a, d = panel("1. size only: too low")
l, t, r, b = d.textbbox((0, 0), text, font=font)
d.text(((W - (r - l)) / 2, (H - (b - t)) / 2), text, font=font, fill="#dc2626")

# 2. anchor="mm": centres the line box (ascender to descender).
m, d = panel('2. anchor="mm": line box centred')
d.text((W / 2, H / 2), text, font=font, fill="#2563eb", anchor="mm")

# 3. Subtract the bbox offset: centres the visible ink exactly.
k, d = panel("3. bbox offset: ink centred")
d.text(((W - (r - l)) / 2 - l, (H - (b - t)) / 2 - t), text, font=font, fill="#059669")

out = Image.new("RGB", (W * 3 + 40, H), "white")
for i, p in enumerate((a, m, k)):
    out.paste(p, (i * (W + 20), 0))
out.save("centre.png")
print("top offset of the bbox:", t, "px")

Tested on ✓ Linux ✓ Windows ✓ macOS

Output

top offset of the bbox: 9 px

Panels 2 and 3 are both correct; they answer different questions:

anchor="mm" centres the font's line box, from ascender to descender. Text without descenders (no g, j, p, q, y) sits a little above the middle, but every string sits on the same baseline. That consistency is what you want for buttons, labels and table cells.
Subtracting the bbox offset centres the ink of this particular string. That looks best for a single word or number on its own, such as a badge or a thumbnail title.

centre_fix.py

from PIL import Image, ImageDraw, ImageFont

img = Image.new("RGB", (600, 300), "#0f172a")
draw = ImageDraw.Draw(img)
font = ImageFont.truetype("DejaVuSans.ttf", 56)

# Option 1: anchor="mm" puts the middle of the line box at the point you give.
draw.text((img.width / 2, img.height / 2), "Centred", font=font, fill="white", anchor="mm")

# Option 2: centre the ink, subtracting the bbox offset.
left, top, right, bottom = draw.textbbox((0, 0), "Centred", font=font)
x = (img.width - (right - left)) / 2 - left
y = (img.height - (bottom - top)) / 2 - top
print(round(x, 1), round(y, 1))

Tested on ✓ Linux ✓ Windows ✓ macOS

Output

188.5 119.0

Anchors also cover other placements, for example "rs" (right edge, on the baseline) for numbers in a column, or "ms" (horizontal middle, on the baseline) for captions. See How to add text to an image for anchors, wrapping and outlines.

Font file not found? These examples load DejaVuSans.ttf, which is installed on most Linux systems. On Windows try "arial.ttf"; on macOS use a full path such as "/System/Library/Fonts/Supplemental/Arial.ttf". If you get OSError: cannot open resource, see that fix.

The outputs and images on this page are the real results of running every example with Pillow 12.3.0 on Python 3.13 (Linux) on 2026-10-15. Every example is also re-run automatically on macOS (Darwin 25.6.0, Pillow 12.3.0), Linux (Linux 6.17.0-1022-azure, Pillow 12.3.0), Windows (Windows 2025Server, Pillow 12.3.0); the line under each example shows the result per operating system. The DejaVu Sans font files sit next to each script so font examples behave the same everywhere. Sample photos: NASA/Sam Lott, public domain. Spotted something wrong? Tell us.

What the error looks like

Why it was removed

Replacement for every removed method

textbbox vs textlength: what the numbers mean

A drop-in replacement for textsize()

The centring bug after switching

Related