• Member Since 20th Jan, 2014
  • offline last seen 2 hours ago


I like things that are spooky or cute. Now with kofi.

That's a lot of bats

Bats! Everyone is a bat! You, me, your waifu, everyone!

Made it this far?

Some people have been asking about it, so I set up a Ko-fi for anyone who wants to get me a coffee. I swear I won't spend it on anything else. It's over here.

Comments ( 1515 )
  • Viewing 1,511 - 1,515 of 1,515

If it gives you any idea what I'm doing, here's a screenshot...


Edit: Look at this one I generated!

--Sweetie Belle

A Lora's actually a special type of AI model that you train, and then it sort of works as an add-on to your existing model. Usually, they'd be used for making it follow a style or generate pictures of a particular character.

Of course, one of the nice things on ipadapter is that it can do similar things without all the training and images a lora needs. With a lora, it could be any character, though, whereas this was pretty much following the images I put in.

One nice thing about ComfyUI is that while it's more difficult then other ai software, it's more flexable. You're basically wiring nodes together to make a workflow, and the benefit of that is that the nodes can be made by anyone. And once the workflow is made, you can keep using it.

One other thing I should really experiment more with is the Krita plugin. There's a plugin for Krita that tells it to connect to ComfyUI, and then within Krita, you can have it do things like generating AI pictures based on a combination of a text prompt and whatever you are drawing at the moment, and you can have it transfer the current ai picture to a layer. All sorts of possibilities there...

--Sweetie Belle

It's impressive it can do it with so few images. Though I guess if you still have the whole comic that's a lot. Maybe I can force the AI to finish the comic for me some day.

And Lora? Do you mean Lilly and/or Iris?

Happy birthday!

They are AI generated. In fact, what I did was fed two pictures from Aetheria Epics to IP adapter, and told it to use them for style and composition, with a minimal text prompt. The image used for the last two was your avatar, of course.

Didn't specifically remember your birthday, I'm afraid, but I was setting up a new workflow in ComfyUI, and adding the new ways to use ipadapter in, and I wanted to see how well it could do that particular art style. They came out well enough I thought you'd enjoy them, though I still want to see if there's any way I can get the actual lines in the pictures to look better.

It's possible to get it to generate non-pony versions too, since the pony 6 XL model also can do anime and furry art...


What I should try some time is to make an Aetheria Epics Lora. It'd take picking out a bunch of pictures and pairing them with descriptions, and probably editing them to remove text and such. I always find myself inevitably feeling like making Loras would be a lot of hassle, though...

--Sweetie Belle

Le gasp! Are these AI generated or did you learn to draw? The last two are very 'me'. Also if you remembered that it was my birthday then I'm very impressed.


  • Viewing 1,511 - 1,515 of 1,515
Login or register to comment

There's more of them!

This page property of Luminary.