The size of png images will increase after being merged into pdf

[yang-521]

I browsed through the official documentation. No matter how a png image is inserted, it will cause the pdf file to become larger, but a jpg file will not. The size of the pdf file obtained in jpg format is approximately the same as the original image size. Both formats use the same compression parameters when saving files

[JorjMcKie]

Provide reproducing data please. Which options did you use on save?
JPG and PNG use utterly different compression algorithms. So much depends on the image itself whether PNG of JPG yield a smaller compressed size.

[yang-521]

doc = fitz.open() for jpg in jpg_list: img = fitz.open(jpg) rect = img[0].rect pdfbytes = img.convert_to_pdf() img.close() imgPDF = fitz.open("pdf", pdfbytes) width, height = rect.width, rect.height page = doc.new_page(width=width, height=height) page.show_pdf_page(rect, imgPDF, 0) doc.save('outfile.pdf', deflate=True, garbage=3)

[yang-521]

This is my code. When replying, I don't know why there was a problem with the line break

[JorjMcKie]

Use this option and choose "code block":

/

I am not so much interested in your code but in your image files examples

[yang-521]

[yang-521]

[yang-521]

The jpg files are basically the same, but the png files are all about 50% higher

[yang-521]

In fact, this png was originally larger. I processed it with PIL. Could it be that the image was restored due to re-decompression during the pymupdf merge process

[JorjMcKie]

Image insertion effectively happens under the control of the base library. Currently, we cannot influence the compression parameters used here.
So it may be that some PNGs indeed increase under the insertion's re-compression.
This is especially true when PNGs contain an alpha channel (transparency), which cannot happen with JPEGs.

[yang-521]

图像插入实际上是在基础库的控制下进行的。目前，我们无法影响此处使用的压缩参数。 因此，某些 PNG 图像在插入的重新压缩下确实可能会增大。 当 PNG 图像包含 Alpha 通道（透明度）时尤其如此，而 JPEG 图像则不会出现这种情况。

It might indeed be for this reason. I tested merging the png images directly converted from get_pixmap into PDFS, and the sizes were basically the same. However, the processed png will be restored by pymupdf