Open mr-september opened 2 weeks ago
As you can see - for about a year this project has been abandoned by the author and from updates only acceptance of Pull requests. It remains to wait until someone decides to create the same project, but actual or wait for someone kind who will send Pull requests
Although it can still be used in a limited way and most of the features still work on sd1-SDXL.
I remember in the old Readme, Noise Inversion was part of the recommended settings for upscaling.
It is no longer there, and not mentioned in the wiki guide and nobody discussed the parameters in https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111/issues/3. The string 'noise inversion' only appeared once, off-handedly.
Can the authors or any experienced users please clarify if Noise Inversion is still a generally recommended setting?
And relatedly, is the workflow in wiki guide still considered the "general" SOTA?