Feat: Does it support adjusting the quality of the generated image?

[a523]

Feature description and goals

We have a requirement to generate preview images for mobile use. By default , the user accesses a low quality compressed image, until he click the ’raw image‘ button.
So, I want to use the project not only to generate preview images but also to generate the compressed images, If the raw file is a image.
This way, I don't have to look for additional compressed libraries, because I think they should have almost the same dependencies, and I want to put them in the one project of mine.
That mean is I don't want to change the width and height of the image, but I want to change the quality (file size) of it

Prerequisites

[raphj]

Hello @a523, thank you for your feature request. Seems like it would be a useful thing to have.

@inkhey it seems like we could add a quality parameter to PreviewManager.get_jpeg_preview, what do you think?

You'd still need to pass the correct width and height though.

[inkhey]

it's a good idea, but depending on the builder behind the parameter may have different meaning.

According to the current pillow builder, there is already few parameters that make sense for jpeg:

• compression level aka quality
• resampling algorithm

and maybe:

• optimize
• progressive
• background color ?

preview-generator/preview_generator/preview/builder/image__pillow.py

Lines 14 to 28 in ee5b0ee

	# INFO - G.M - 2019-06-25 - some configuration for jpeg saved,
	# see https://pillow.readthedocs.io/en/latest/handbook/image-file-formats.html?highlight=JPEG#jpeg
	# optimize -> extra pass to select optimal encoded
	# quality -> compression level from 1 to 100
	# progressive -> permit to see a blured version of file instead of partial image when image
	# is not fully downloaded, nice for web usage.
	DEFAULT_JPEG_OPTIMIZE = True
	DEFAULT_JPEG_QUALITY = 75
	DEFAULT_JPEG_PROGRESSIVE = True
	# Pillow algorithm use for resampling, bilinear is not the fastest but is fast
	# enough, this permit good enough image (using faster algorithm return much more
	# aliasing. see https://pillow.readthedocs.io/en/latest/handbook/concepts.html?highlight=Bilinear#filters
	# for algorithm list.
	DEFAULT_JPEG_RESAMPLE_ALGORITHM = PIL.Image.BILINEAR