The old training engine switcheroo

Ok there's a couple points I want to address now that we are at the last quarter of the year.

The current state of things on the site

I've been asked a couple of times if I'll keep posting models considering there's been changes on the features of the site since people have chosen to leave. I understand perfectly that Civitai is being run as a business and sometimes you need to reduce costs and make improvements on functionality, change things here and there, I get it, I've been on the helm of some projects that were successful thanks to many talented fellas. Ideally I would like to see a more transparent communication from the bigwigs here, it feels that it would definitely help.

I don't exactly know the workings of the resource management but I'll assume that the AI toolkit is less demanding compared to Kohya, and not only that it has stopped receiving support as a reason to make the change of the default engine. From my experience using the toolkit it takes a lot of luck to get results that are on par of Kohya's but it might very well be related to the settings used, we just need to figure them out and finetuning from there. I'll share more of my insights as I find settings that work better, so no need to worry about me keeping it secret.

In my opinion we'll simply have to adapt to the tools that are available to us, we might end up finding that it work better in some cases, who knows.

Also at least from my part as long as I get commissions I'll keep making models, after all, it's pretty much my job at this point, if I consistently achieve my monthly goal on my Ko-fi I can fully focus on training, which take me to my next point.

Terms of commissions change

Thanks to some people and their poor 'haggling' attempts I'll end up with more strict terms on comms I take, such as keeping to a specific season of a show or limiting the number of outfits to be included for characters, I'm rather grateful that I've had no problems with all of the people that have generously commissioned me, you're the ones keeping the lights on here, so there's that, opposed to those demanding multiple models and trying to pay less than the base cost of a single comm, if a character has an extra outfit per episode and every season is more than 50 episodes, don't expect to pay half of the regular, "That's mathematics, son"

I'll also increase my base cost by 5$, to a total of 15$ starting this December, gotta fix some ceiling leaks, I've been told by proper artists (that pick up a pencil and everything) that that they consider that new price still to be low but I need a consistent flow of money, gotta be reasonable with all that supply and demand, not all of us can be suspiciously wealthy furry artists eh.

So in conclusion, yeah we gotta learn to use the AI toolkit, and I like money, gimme.

"End communication."

Update: Oh yeah I also forgot to mention, I don't think I'll ever add a fee of generations on my models, I usually set them to early access to recover a part of what I spent on training, if I get back what I used, great, if I get slightly more, perfect. So no need to worry about that.