Posts

only one i can think of off the top of my head is this
https://github.com/fpgaminer/joycaption

anyway, a lora of that sort is also something i’m interested in doing.
maybe we could try and work together?

@Cypher
https://civitai.red/models/2658461/my-little-ponymane-six-g5 g5

https://civitai.red/models/2656426/my-little-ponymane-sixequestria-girls, I don’t know who will be interested in this LoRA, but it has the entire G4 six.

@Cypher
ayyyyyy you’re back? awesome!

It helps when you realize that the people who do stuff like are basically wasting their life spewing hatred online and are likely either

A) Miserable people
B) Bots run by miserable people

The best thing to do is just block and move on. There’s a weird, self-harm mindset some people have where they feel they should not block users because it’s somehow ‘rude’, but it’s not rude to block someone who is actively making your life worse. (Note: I have no idea if Civit even has a block user function).

Regarding your LoRAs, I use many of them and I think they’re great! But the most important thing is yourself, and if hosting on Civit is giving you anxiety / stress, your best bet is to find some alternative method of distribution. HuggingFace has terrible searching tools, but you can upload and distribute LoRAs from there, so if you linked them from here on the Character LoRAs section, people who want to find them could find them through the links here on Tantabus.

An extreme alternative is that you could take the route that most online personalities have to adopt as they grow: put a buffer between yourself and the toxicity. Have someone else manage your affairs while you simply create. It might be a close friend, a girlfriend/boyfriend, or a trusted online friend.

Your ‘buffer’ partner

Admittedly, this is far easier to write down the steps for than to find a good partnership to work with, especially in an unpaid context, and you need a partner who is going to be able to ignore, for example, threats from anonymous online people. And of course, all of the insinuations of content that can/shouldn’t/should be made with your LoRAs, which is an unfortunate side effect of AI being… well… AI.

@CappyAdams
The only model is NoobAI. I don’t understand how to train it

@Cypher
ah, I’m so sorry to hear that… I hope this isn’t an issue to ask but are you planning on uploading your loras to google drive and sharing the link here or anythin? I’d love to try them out :>

Heya! I’m curious, are there any uncensored taggers that support neutral language? I need one because I’m working on a pony enhancer for Anima, which is a model that supports both NL and tags

No, I know how to make LoRAs. It’s just that when you get spammed with all kinds of garbage in your DMs and even on your phone for 4 years straight, you burn out. Because of that, I deleted my Civitai account. I wanted to delete this one too, but I changed my mind

@Cypher
If you don’t mind me asking, what kind of hate were you getting? was it “this lora sucks” or “This sucks you should stop training”? just asking because if it’s the first option, I’m more than happy to send you resources for training

After 4 years, I feel like I just wasted my time. I just gave up and deleted my Civitai account because of all the hate I was getting. I won’t name anyone to avoid drama, but I’m just sick and tired of being everyone’s punching bag.

@Equum
Sadly, I do not know of any ways of getting two different character LoRAs to work together other than decreasing both their strengths and hoping for the best or resorting to inpainting.

Thanks for sharing! That seems more approachable than I thought. Do you find you need to do anything special in order to end up using many of your LORAs at the same time without them interacting weirdly?

I think I’m qualified to comment on this subject as I’ve made over 200 MLP character LoRAs for SDXL-based models (mostly Illustrious, though I started with PDXL 6 and have recently shifted to NoobAI).

One thing I have found is that a small, varied and well-tagged dataset is far better than a large one with just passable tagging and not that much variety. I generally use around 30 distinct images of the character (a minimum of around a dozen pics and a maximum of around 50) when training my LoRAs.

After I gather the images for the dataset, I run them all through an autotagger and then touch up its results (adding a tag for the character, removing tags that don’t apply, adding missing ones).

I think I have the capacity to train LoRAs for Flux.2, but I have no clue how to tag images for one (at the moment), so I haven’t done anything of that sort yet.

I’ve been gaining experience with PonyDiffusion and enjoying the easy knowledge it has of MLP characters, but run into some problems with fringe characters it has misconceptions of (characters are the wrong gender, have 90% correct features but random horns or other features, presented in biases acitons or poses, etc). I sometimes find pre-existing LORA’s that help with this, but sometiems the LORA’s themselves cause similar hiccups or interact with existing projects in a negative way.

I’ve recently started dabbling with FLUX2, and while it seems much more useful in creative direction and revision, it really seems to have no concepts of the pony world. Which gets me considering getting into LORA creation. There must be some potential here, seeing how impressive the PonyDiffusion project’s results are for the fandom as a whole and how little the base models seem to know.

I’m under the impression that making a LORA is simply a matter of organizing samples, and relativly trivial to run if you are willing to use a cloud serivce?

I’ve seen some others on here referencing their own LORA creation, and wondering what your experiences have been like. Is this an unrealistic thing to consider doing solo for specific characters, or do any of you find it useful? Is it relatively simple on a base model like SDXL/PXL, and unrealistic on a pony-naive model like Flux?

@Cypher
Well… that’s sucks. I liked these LoRAs… I can’t gen locally so it’s quite a blow for me.

Civit, Channel’s gone. Who cares about the 800 followers, I’m a nobody anyway

@Cypher
On Civit? Or here on Tantabus?

Here on Tantabus, you’ll need to have an administrator do it. On Civitai, you’ll need to go to the bottom of the account settings page (account settings is the gear at the bottom of the top-right dropdown menu)

How do I delete my account?

Chapter 11 added.

@Thoryn
Glad I’m not the only one. My uploads have slowed dramatically.

@Equum Unfortunately, those are limitations of CLIP, which is what a lot of older models use, including PonyV6, Illustrious, and basically almost every model that competently renders ponies. One of the major limitations of CLIP is that it doesn’t really comprehend spatial relationships, which means it really only can do spatial relationships when the specific spatial relationship you’re prompting exists a lot in the training data. It’s not capable of dynamically putting together a new spatial relationship without a LoRA to assist.

Newer models have a much better understanding of spatial relationships (and most of them are also natural-language capable) which makes them much more useful for prompts like what you’re describing. Unfortunately, most of them suck at general pony knowledge; some of them might kinda vaguely get major characters like Twilight or Celestia right, but many of them can’t even comprehend the CMC. They’re also generally heavier models that require better hardware just to get tolerable generation times.

@Equum
No, sadly haven’t had the time to play around with image generation that much the past year because of IRL things.

@Thoryn
Did you end up making progress on this? Settle on any specific workflows?
I’m basically reliving this experience you’re having as I try to get up to speed on all this image generation stuff after stumbling into it in 2026. Lots of the same issues here.