Long story short. Before joining this site, I have been a user at Civitai which I have been enjoying, and still do (even with problems). Months ago, I found interest on this site as the content that I do generate does cater to this community, and also that there are less restrictions. However, there are some issues. One, Civitai does require premium currency in order to generate NSFW contnet, and two, the uploading guidelines here makes it almost to nearly guarantee that inpainting is required.

Fortunately, I am someone with a MacBook that is able to run AI image generators locally, I have used Draw Things every once in a while. I know the basics of using it, but there are still plenty of things that I still need to know to really know what I'm doing. There isn't a lot of tutorials out there on the internet that really teaches you how to use Draw Things, let alone one for furry content. I have used inpainting a few times before, but only know how to use them for eyes as I do have a LoRA for that. Other specific parts of the face, especially teeth have been a pain to deal with. Does anyone here have more experience using Draw Things? If so, I'm curious about your workflow when using it.

With proprietary image editing models having more capabilities with each iteration, I have been using them as part of my workflow. My first post that I uploaded from Civitai (now my profile pic here) did use gpt-image-2 to do a small edit for an existing image that I have generated. That post has been approved (and that made me very happy). I was then looking through my Civitai profile to see what other images can I upload (that were also edited or using a character as a reference for consistency with proprietary models). Unfortunately for me, the next two posts (both used Nano Banana Pro) were taken down, either by not meeting quality standard or a small nitpick (that should easily be fixed). I'm not exactly mad by it, but it kind of left me frustrated, and also made me wonder if using proprietary image editing models are even worth it when it comes to using them to edit images or making comics and posting them here.

I'm aware there is a Discord server for potentially some of the stuff that I'm trying to figure out (probably will join there later), but I do want to hear some perspectives here before I do so. I know there are a growing amount of users here that do like what I post, so I do want to improve in generating images so that others can enjoy them without having to worry that they could be taken down, especially when it comes to NSFW content, and to prevent my upload limit from being reduced (this part worries me the most as funny as it sounds).

Thanks for anyone reading and responding to this.

Hello and welcome :)

I am surprise that I never heard of "Draw Things" (probably because does not have too many Mac user?).
So, I can't help you much for that, but, I know that a good amount of people in the Discord (myself included) using the image editor Krita + plugin AI Krita Diffusion.
Which is also supported for mac.

There is a way bigger chance having people that can help with AI Krita Diffusion then "Draw Things", but, of course, no need to change if you don't want, a lot of people use different tools.

As for using proprietary image editing models, it is up to you.
If it help you, good. If you feel like it is not good enough to fix stuff, fine as well, use a different model and/or tools until the problems are fixed.
The only thing that is evaluated is the result.

You say you use a LoRA to fix eyes when inpainting, which good if works, but in most case, you probably don't need any LoRA at all to fix eyes, nose, teeth, paws (...).
You often just need the same model you used to gen your raw image.

Also, the more LoRA you use, the more chaotic the result might be.
As someone said in the Discord (forgot who!), each LoRA is like another cook in the kitchen, too many and no one can agree on how to do a specific meal.
Best would be no LoRA at all, but, of course, LoRA exist for a reason, to make a model understand a character/concept/etc that it does not know.

If you feel like it, I would recommend you to join the Discord, will be easier to help you.
We have a channel named "gen-help" for doing just that, helping people :P

You can also DM me here or in Discord (same username) if you prefer (but, that said, I am most likely going to sleep soon, so won't have a quick answer from me most likely).
Or wait for more answer here, but, there isn't a tons of people checking the forum.

oarebt said:
Hello and welcome :)

I am surprise that I never heard of "Draw Things" (probably because does not have too many Mac user?).
So, I can't help you much for that, but, I know that a good amount of people in the Discord (myself included) using the image editor Krita + plugin AI Krita Diffusion.
Which is also supported for mac.

There is a way bigger chance having people that can help with AI Krita Diffusion then "Draw Things", but, of course, no need to change if you don't want, a lot of people use different tools.

As for using proprietary image editing models, it is up to you.
If it help you, good. If you feel like it is not good enough to fix stuff, fine as well, use a different model and/or tools until the problems are fixed.
The only thing that is evaluated is the result.

You say you use a LoRA to fix eyes when inpainting, which good if works, but in most case, you probably don't need any LoRA at all to fix eyes, nose, teeth, paws (...).
You often just need the same model you used to gen your raw image.

Also, the more LoRA you use, the more chaotic the result might be.
As someone said in the Discord (forgot who!), each LoRA is like another cook in the kitchen, too many and no one can agree on how to do a specific meal.
Best would be no LoRA at all, but, of course, LoRA exist for a reason, to make a model understand a character/concept/etc that it does not know.

If you feel like it, I would recommend you to join the Discord, will be easier to help you.
We have a channel named "gen-help" for doing just that, helping people :P

You can also DM me here or in Discord (same username) if you prefer (but, that said, I am most likely going to sleep soon, so won't have a quick answer from me most likely).
Or wait for more answer here, but, there isn't a tons of people checking the forum.

Okay, good to know. Probably going to stick with Draw Things for the time being despite now knowing that it is not widely used (as I'm more familiar with it). I'll use proprietary image editing models in a case by case basis when and if needed. Will keep these tips in mind going forward. Thanks for the information.