Trained on a set of pictures ranging from English countryside to New Zealand Alps, and crumbling architecture reminiscent of Gondor.
Totally new to this, so no idea how this is going to work. It should give (hopefully) decent-ish results at around 15 steps & 1 CFG.
Description
FAQ
Comments (5)
It's been over a day and the safety scan hasn't completed. Odd. Hmm
I probably have no idea what I'm doing, but training that LoRA has been a nightmare so far, the status of it on my training page still shows "submitted" even if it is actually completed, then probably I did something wrong when uploading manually, or there is some behind the scene issue I'm not aware of.
I'm trying to train a SDXL version and it constantly fails, it's all very frustrating.
@favoreddugong That's good, finding out what doesn't work is the fastest way to learn the ins and outs of a new thing. I was moreso commenting on how odd it is that the site hasn't safety scanned it yet, because your outputs show you did a great job on this. Godspeed in your efforts, can't wait to see where you take this Z-Image lora.
@7117 I would like for someone else to test it though because some of the output was less than underwhelming, so the LoRa has definitely its little quirks. Also I have no idea how much of the result is due to the LoRA or the checkpoint, that's why I am trying a SDXL version to see how it goes. But yeah, trials and errors I guess...
Great news, after almost a week and a half the safety scan completed and it was verified. Here's hoping that didn't slow your LoRA's downloads and usage too much! Godspeed