Add node to read watermarks

[lstein]

Summary

This adds an invocation to read the invisible watermark created by the AddWatermark invocation.

Related Issues / Discussions

QA Instructions

1. Create a simple workflow which pipes an image from AddWatermark to RetrieveWatermark, as per example above
2. [Optional] Change the default watermark text from "InvokeAI" to something of your choice. If you do this, be sure to adjust the watermark string length in RetrieveWatermark appropriately.
3. Run the invocation. The output from RetrieveWatermark should match the AddWatermark string.

Merge Plan

Merge when approved.

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Tests added / updated (if applicable)
• Documentation added / updated (if applicable)

[hipsterusername]

Is the current watermarking mode what we want to commit to long term? Seems like it makes a notable impact on images; although, I’m not sure there’s a better one that’s open source. I know Meta had one, but they put a dumb license on it.

[lstein]

Is the current watermarking mode what we want to commit to long term? Seems like it makes a notable impact on images; although, I’m not sure there’s a better one that’s open source. I know Meta had one, but they put a dumb license on it.

Here's another library which seems to be based on the same algorithms as invisible-watermark: https://github.com/guofei9987/blind_watermark . It lets you embed either a watermark text or an image in the file. I can play around with it and see if the image effects are less noticeable.