If I understand correctly, this is not just for passing an image object directly to IPython, like in #1090 (comment)

from PIL import Image
im = Image.new("RGB", (512, 512), "blue")
im

and not just for IPython.display.display, but specifically for IPython.display.display_jpeg, yes?

If I understand correctly, this is not just for passing an image object directly to IPython, like in #1090 (comment)

from PIL import Image
im = Image.new("RGB", (512, 512), "blue")
im

and not just for IPython.display.display, but specifically for IPython.display.display_jpeg, yes?

Yes this is display_jpeg.

Release notes have now been added - https://pillow--7135.org.readthedocs.build/en/7135/releasenotes/10.0.0.html#support-display-jpeg-in-ipython

I'm wondering what the purpose of this change is. I'm interested to know if it's mainly to support calling display_jpeg on an image?

If so, this seems to be having unintended consequences as I'm now getting both a PNG and JPEG encoded image saved in each cell. For example, a cell I just created includes:

     "data": {
      "image/jpeg": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsI>
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAC7kAAAu5CAAAAABw4BI+AAEAAElEQVR4AQBSgK1/AdfANnmuHKdSoc6W0wL98OSB+b>
      "text/plain": [
       "<PIL.Image.Image image mode=L size=3001x3001>"
      ]
     },

which might well double resulting .ipynb file sizes, depending on how either encoder deals with the image.

I'm interested to know if it's mainly to support calling display_jpeg on an image?

Yes. That is what the original poster was after, and that is what we have announced.

Discussion of double images can continue in #7259