Hi everyone,
I want to discuss the future of my models, particularly AingDiffusion XL.
I've realized that I've been updating the models too quickly, sometimes without sufficient proof of concept and not in a clear manner. As a result, these updates have produced a range of image quality from excellent to barely passable. My QC team is working hard to ensure the model improves and rejects bad versions. However, I suspect that sometimes they tell me the new model update is better than the last one to make me happy, given the significant amount of money I've spent on training. They don't want me to feel like I've wasted a lot of cash only to scrap a model later.
This situation makes me feel like I'm producing trash at this point. I don't think I can continue working with this kind of workflow.
With this in mind, I've decided to implement a major refresh of AingDiffusion XL. This also means I will be deprecating the old AingDiffusion XL model card/page and considering all versions of AingDiffusion XL except v1.4 as legacy versions. When the refreshed model is finally published, v1.4 will also become a legacy version.
I'm also planning to release the refreshed version of AingDiffusion XL under a new name, since "AingDiffusion" originated from my habit of naming files "Aing" (which means I/My/Mine in Sundanese) when I couldn't think of a name. I feel that the name "AingDiffusion" no longer represents its original purpose. If I can't find a new name, I might call it NeoAingDiffusion or, as a nod to what many Sundanese users have suggested for a sequel name, UrangDiffusion.
Regarding my other models, I stopped the development of SD 1.5-based models a couple of months ago. As for AniEase, since the merge formula is quite "boring," I will conduct further research to improve the model merge. For Pain-G, I must admit that I have been very lazy in developing it because it is costly to train the same dataset on two different models. It will probably become a merge model rather than a trained-merge.
As always, I thank you very much for reading this. I hope you have a great day.
If you have any questions, please leave a comment.
If you want to support me, you can send me some of your Buzz, or you can visit my Ko-fi page.
Best regards,
KayfaHaarukku