(0.6 - 0.8)
its nice on its own, but use INPAINT for CA$H MONEY result.
Keywords:
Cherub + (baby, angel, floating, flying, multiple wings, metal details)
Servitor + (skull, floating, flying, metal details)
Cherub and servitor (both will come out)
Equipment: (TYPES OF EQUIPMENT TO CHOOSE)
(holding) +
scroll
wire
trumpet - (usually tough for inpaint, cause u gotta be able to draw trumpet or sillhoute of a trumpet, i mean how many people can actually draw trumpet, like best i can draw is prolly flute but thats not even that good, fuck i suck at drawing.)
Please let me know if you're enjoying the model's performance so far. Additionally, if you encounter any amusing, failed, or successful prompts during your prompting journey, I would be delighted if you could share them @AI.failed on Instagram. This is a fantastic opportunity to contribute to the AI community's meme collection. Your support and engagement are greatly appreciated!
Description
FAQ
Comments (6)
There seems to be smth wrong with this LORA. When I click on it in the panel in a1111 UI, it ads "<lora:Angron MK1 by CARAXES:1>" to the prompt, which can't be right. The resulting image are all messed up. Changing the LORA text manually to "<lora:Cherub & Servitor MK1 by CARAXES:1>" (took it from the example pics) doesn't help either - the images still look messed up. Am I just dumb and doing something wrong?
hey man, it could be your a1111. maybe try to reload or restart it. else check for update? im no expert at this, i myself is still learning. Also, this lora been posted for quite sometime. and if there's a problem, some one would've notified me. if there's a file mixed up it would've been improbable, since Angron and cherub files are separated by 1 week of different uploads. i dont mean to be defensive, but this seem to be a quite isolated case.
@caraxes Thanks I'll check again if it's something on my end and report my findings. Thanks btw for all the 40k stuff you're putting on this site. I'm more of a Midjourney guy and your LORAs are the only reason I've installed and started learning Stable Diffusion 😀Your LORAs allow making characters in SD that are either outright impossible or excruciatingly difficult with Midjourney!
@caraxes It seems the problem has been a mix of both: one hand, I've been trying to use the LORA incorrectly, and on the other, the metadata indeed has an error - it outputs the wrong name into the prompt box. But I was able to make it work once I substituted the string manually with the correct one <lora:Cherub & Servitor MK1 by CARAXES:1>. What I've been doing wrong was trying to use the lora without inpainting - using it in the overall prompt to add the servo skulls/cherubs together with the rest of the composition. With a high weight, the LORA was trying to make everything in the image look like a cherub/skull resulting in the picture looking messed up 😅. Works just fine with inpainting, though 👍👍Funnily enough, if used in a prompt with a low weight (I messed with 0.25-0.35) it can act as a "wh40k filter" for any base model, making the image look a bit more grimdark 😅
@_Calgar_ yeah just leave it at inpaint. dont try to mix let say guilliman lora and mix it with servitor in hope to produce picture of 'guilliman accompanied with servitor' , i have done it and it will not work. INPAINT is a must to produce such result. then again im still bad at inpaint myself and its kind of willy nilly result.
same thing with TITANS, i wanted to pose it along with CUSTODES but it just dont work in INPAINT. maybe GPU also plays a factor in the output. since the bigger the canvas the more shit you can plonk in to the drawing. and since my gpu is measly mini2060super it can only do so much.
@caraxes Yeah I haven't mastered in-painting either. For me it's easier edit generated images in Photoshop than feed them back to SD with img2img and let the ai smooth it out
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.