Is there a way to save a coalesced gif?

[lucasvickers]

Hi All,

I'm working on a project that requires fully coalesced gifs. Based on this gist, I'm able to confirm that I'm saving a non-coalesced gif.

Is there a current way to achieve this? If not, any ideas on what would need modification to allow this?

My code:

  from PIL import Image as PilImage

  first = PilImage.open(image_files[0])
  sequence = []

  for i in range(1, len(image_files)):
    sequence.append(PilImage.open(image_files[i]))

  first.save('pillow-out.gif',
             save_all=True,
             append_images=sequence,
             optimize=False,
             compress_level=0)

And the currently resulting gif:

[radarhere]

The gist you linked to is 11 years old. We've significantly improved our reading of GIFs since then, so if you want to extract all of the frames from a GIF, you should no longer have to be concerned about pasting and palettes. You should just be able to

from PIL import Image, ImageSequence
with Image.open("pillow-out.gif") as im:
  for i, frame in enumerate(ImageSequence.Iterator(im)):
    frame.save(str(i)+".png")

But your actual question is about saving. Sorry, I don't know exactly what you mean when you say 'coalesced'. Could you describe that in more detail for me? At the moment, I see 25 frames of the number 72 shrinking and then expanding again. What are you hoping to see?

Ideally, if you could attach your input images (image_files), that would also be helpful.

[lucasvickers]

When I say coalesced, I mean that every frame is an unoptimized complete set of pixels. This is different from frames that only contain the differential pixels. They describe it in imagemagick docs, seems like it's not a universal term :) In the gist I referenced it's referred to as full vs additive, I want full.

So given these source images: test.zip, I want a gif where each frame is a full 600x600 pixel image. And you are right, it's just a number pulsing against a gradient.

I'm only using the highlighted section of the gist (L28-L51) that checks if the gif is full ( what I'm calling coalesced) or additive.

ImageMagick is able to do this using -coalesce but it's overall gif handling is poor and it renders some terrible artifacts:

[radarhere]

Ah, ok. It's not possible to do this using Pillow's public API. However, you can workaround that by adding in the following code when generating your image.

from PIL import GifImagePlugin
GifImagePlugin._getbbox = lambda a, b: (0, 0) + a.size

[lucasvickers]

Awesome, thank you!

[Yay295]

I think you're talking about the disposal method used. You can set this when saving. I think you probably want method 0.

https://pillow.readthedocs.io/en/stable/handbook/image-file-formats.html#gif-saving

[lucasvickers]

Thanks for the idea! I tried 0-3 but all of them have differential rendering (i.e. update_region_dimensions != im.size)