Going Forward


Hey, been a while, sorry 'bout that. Been busy with life and such, and Twine updated, so I've been having trouble adjusting since they completely reworked their UI, still not quite used to it, since I've been busy, but I'm making progress to being able to properly use it on my laptops tiny screen. I put emphasis on that because Twine made it so that when you open a segment of the game to edit or create it, it makes a small side window. This side window shrinks when you open another one, which makes it difficult for me to navigate the pages.


That aside, I think moving forward I will be using AI art to detail what is happening. Not only will it be cheaper, but if I can get a decent mix of AIs it looks like I'll be able to make some really great art. The main issue is that it seems there is a large learning curve to using AI art, in particular with merging different AIs for making artwork, which is a tactic commonly employed for the really good pieces of AI art. I'm still not quite sure about how it works, but I will continue to try, and hopefully, I can use it in the near future to get back to work!


Any advice on how AI artworks work is much appreciated, hell, if someone would be willing to walk me through merging different AI checkpoints, I would be over the moon. But, well, I don't know anybody with expertise in this area, and all the online tutorials are awful, so I'll be going in blind. Additionally, any recommendations on mixes of AI checkpoints would be greatly appreciated, as I only know a few, and while they look pretty good, they have flaws I'm not happy with and that I don't see in some AI art, so I will be on the lookout for the best possible merges.


Again, sorry for the slow progress, but I am very much so based on what I am taught, and having to adjust systems so suddenly and greatly has slowed my working speed, especially since I got busy just as it changed so drastically. Thank you for reading my rambling, and have a wonderful day!

Comments

Log in with itch.io to leave a comment.

Deleted 274 days ago

Sorry for the delay. I sent a request, thank you!

(+1)

No idea how far you've got looking into this, but here's my thoughts.

Berry's Mix is a very good merge focusing on anime, you can find instructions on how to merge here. It includes using the leaked NovelAI model, which you may find you'll have excellent results with anyway, without merging. The other model you might do well with is Anything V3.0. These are both anime/cg style.

Character art may be the biggest issue. As far as I'm aware, to use AI to create character art and pose and stuff, you need to train it on your character, which you'd need to look into and have a system capable of doing so, as I have no experience on that.

And finally, as stated above, your system has to be able to run the software. I run it on a 1070, but I think it can run on on a 1660 too. It can be run on Google Colab though, so if you don't have a great system that's a decent plan.

I see, thank you for the assistance! I've got a fairly decent machine, using a 2070 and ryzen 7. I've gotten a base model working, but I'll be sure to check out how to merge as it sounds far more effective. I've noticed my biggest limiter at this moment in time is RAM, as I use 16gb, which is enough to use, but only nets me so much at once and I'd rather have a larger pool of images to experiment with.


Thank you for the advice, I'll be sure to look into AI training!