# Convert PDF to Images with Python: PyMuPDF, pdf2image and Pillow
Extracting PDF pages as PNG or JPG images is a common task: document previews, diagram extraction, OCR processing, or visual archiving.
## Tool Comparison
| Tool | Speed | External Dependencies | Custom DPI |
|---|---|---|---|
| **PyMuPDF (fitz)** | ⚡ Very fast | `pip install pymupdf` only | ✅ |
| **pdf2image** | Good | Poppler (external binary) | ✅ |
| **Wand (ImageMagick)** | Medium | ImageMagick (binary) | ✅ |
## PyMuPDF (`fitz`) — The Fastest Option
```bash
pip install pymupdf
```
```python
import fitz
import os
def pdf_to_images(pdf_path, output_dir="pages", dpi=150, fmt="png"):
os.makedirs(output_dir, exist_ok=True)
doc = fitz.open(pdf_path)
zoom = dpi / 72
matrix = fitz.Matrix(zoom, zoom)
paths = []
for page_num in range(len(doc)):
pixmap = doc[page_num].get_pixmap(matrix=matrix, alpha=False)
path = f"{output_dir}/page_{page_num+1:03d}.{fmt}"
pixmap.save(path)
paths.append(path)
print(f" {path} ({pixmap.width}x{pixmap.height}px)")
doc.close()
return paths
images = pdf_to_images("document.pdf", dpi=150)
```
### Specific Pages Only
```python
import fitz
doc = fitz.open("report.pdf")
matrix = fitz.Matrix(200/72, 200/72)
for idx in [0, 2, 4]: # pages 1, 3, 5
if idx < len(doc):
pixmap = doc[idx].get_pixmap(matrix=matrix)
pixmap.save(f"page_{idx+1}.png")
doc.close()
```
### Transparent Background (Alpha Channel)
```python
import fitz
doc = fitz.open("logo_in_pdf.pdf")
pixmap = doc[0].get_pixmap(matrix=fitz.Matrix(2, 2), alpha=True)
pixmap.save("with_transparency.png")
doc.close()
```
### Save as JPG
```python
import fitz
doc = fitz.open("catalog.pdf")
for i, page in enumerate(doc):
pix = page.get_pixmap(matrix=fitz.Matrix(150/72, 150/72), alpha=False)
pix.save(f"page_{i+1:03d}.jpg", jpg_quality=85)
doc.close()
```
## pdf2image — Poppler for Maximum Fidelity
```bash
pip install pdf2image
# Ubuntu: sudo apt-get install poppler-utils
# macOS: brew install poppler
```
```python
from pdf2image import convert_from_path, convert_from_bytes
import os
pages = convert_from_path("document.pdf", dpi=150)
os.makedirs("output", exist_ok=True)
for i, page in enumerate(pages):
page.save(f"output/page_{i+1:03d}.png", "PNG")
# From bytes (web APIs)
with open("document.pdf", "rb") as f:
pages = convert_from_bytes(f.read(), dpi=120)
pages[0].save("first_page.jpg", "JPEG", quality=90)
```
### Advanced Options
```python
pages = convert_from_path(
"report.pdf",
dpi=200,
output_folder="./output", # write to disk without loading into RAM
fmt="jpeg",
jpegopt={"quality": 85, "progressive": True},
first_page=2,
last_page=5,
grayscale=False,
size=(1200, None), # fixed width, proportional height
thread_count=4,
)
```
### Large PDFs Without Exhausting RAM
```python
pages = convert_from_path(
"large_book.pdf",
dpi=150,
output_folder="large_pages",
paths_only=True, # returns paths, not PIL objects
)
print(f"Generated {len(pages)} images")
```
## Post-Processing with Pillow
```python
from pdf2image import convert_from_path
from PIL import ImageEnhance
import os
pages = convert_from_path("scan.pdf", dpi=200)
os.makedirs("processed", exist_ok=True)
for i, img in enumerate(pages):
gray = img.convert('L')
enhanced = ImageEnhance.Contrast(gray).enhance(1.5)
binary = enhanced.point(lambda x: 255 if x > 128 else 0, '1')
binary.save(f"processed/page_{i+1:03d}.png")
```
## Extract Embedded Images from PDF
```python
import fitz
doc = fitz.open("catalog.pdf")
for page_num, page in enumerate(doc):
for img_idx, img_ref in enumerate(page.get_images(full=True)):
xref = img_ref[0]
image = doc.extract_image(xref)
name = f"p{page_num+1}_img{img_idx+1}.{image['ext']}"
with open(name, "wb") as f:
f.write(image["image"])
print(f"Extracted: {name}")
doc.close()
```
## OCR with Tesseract
```python
import fitz, pytesseract, io
from PIL import Image
doc = fitz.open("scanned_text.pdf")
all_text = []
matrix = fitz.Matrix(300/72, 300/72) # 300 DPI for OCR
for num, page in enumerate(doc):
pix = page.get_pixmap(matrix=matrix, alpha=False)
img = Image.open(io.BytesIO(pix.tobytes("png")))
text = pytesseract.image_to_string(img, lang='eng')
all_text.append(f"--- Page {num+1} ---\n{text}")
print(f"Page {num+1}: {len(text)} characters extracted")
doc.close()
with open("ocr_output.txt", "w", encoding="utf-8") as f:
f.write("\n\n".join(all_text))
```
## DPI by Use Case
| Use case | Recommended DPI | Approx. A4 size |
|---|---|---|
| Web preview | 72–96 | ~400–600 KB |
| On-screen reading | 120–150 | ~1–2 MB |
| Standard printing | 200–300 | ~3–8 MB |
| OCR / archiving | 300 | ~5–10 MB |
| Professional offset | 300–600 | 10–50 MB |
## Convert PDF to Images Online
KaijuConverter allows direct PDF to PNG or JPG conversion in the browser, page by page or in bulk.
Guide