This is definately better than generating hundreds of images hoping it would spell out what someone wants... not that i've ever done that... multiple times... for hours :D
This is great. I've been adding text fairly successfully using an external image with the text loaded into controlnet, but this is much more convenient and better overall.
Thought we wouldn't see decent text for open source for a month. Will wait till it comes out for comfyui hope it will be able to Japanese as well. Catching up quickly to Dalle-3, going to be a great year for open source.