Regards for all your love and support ❤
avoiding negatives can improve result on SDXL
Description
Refined from V1 since i got a lot of responses that they don't like that leosams merge and trained with gpt4v tagged images
FAQ
Comments (29)
could you please make a inpaint model?
v5 was taken down? i was so excited and now i am sad inside
It is there?!
@epinikion Yes, it was CivitAI being CivitAI earlier - page kept reverting to a prior version til everything propagated. All good now, sorry for the alarm.
It is a wise choice not to merge with leosam's model. Leosam's model is one of the worst.
Yeah, it doesn't do faces well, unless you use adetailer to fix stuff.
Finally, somebody else, who actually noticed that as well! I thought I was the only one who had problems getting even half-decent results with it...
@latex_waifu Nope. It definitely creates some horrors. I never use adetailer because it makes generations sooo slow, and I often create loads of images I use for training my MXAI embeddings.
takes forever too given the sampler its recommended to use
Hi epinikion
I would have love to get permission to create a merge including your model and share it as the next iteration of my Mklan, hope you say yes Thanks
Sure go with it. Thanks for getting back to me ❤️
@epinikion thank you
wow i love that you guys are having this comradery. This community is the best. Cool work guys! :D
Hi, does this support inpainting? Could you perhaps make an inpainting version please?
Didn’t test inpainting at all with sdxl, maybe I will have a look into it
@epinikion Understood, seems like inpainting hasnt been well supported with SDXL. If you have the will it would be awesome to see though, inpainting with model such as this would be a true gamechanger for AI generation.
Can you open the generate on the site?
It is open for civitai generation, don‘t know why it is not put into the generator cycle
Hi,
I'm not sure if I've done anything wrong, but the generation speed is really slow? Are some models supposed to be way slower than others? Using epicPhotogasm the speed is super fast, even if I'm having around 70 steps or even more. But this one super slow. I'm using 2M Karras with 25 steps. Also ESRGAN 4x to upscale it. Am I doing something wrong?
Am I supposed to use this as the checkpoint, and SDXL as a refiner?
@Stench Since this is a sdxl model it is used to be slow. Just use this model OR epiCPhotoGasm which is a 1.5 model
@epinikion with nothing else needed? I waited it out with 20 steps and CFG 3 for around 15 minutes, and the result was a basically a corrupted image. Green, blue and screwed up. What am I missing?
@Stench usually with sdxl model you want to use appropriate vae. I'm on mobile now and can't remember the name but google "vae for sdxl" and use that instead of the one you use currently
I check the FinalDestination is same time to process that Juggernaut for example and other models SDXL too.
This garbage model cant draw a penis
Dude, it's probably just not trained on that...there are many other models / loras out there ;)
Bro find the right model for that.
Why would you need to draw it, you probably always have it with you...)))
the peen is dangerous stuff lmao
Details
Files
epicrealismXL_v5Ultimate.safetensors
Mirrors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
REAL_XL_epicrealism_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
epicrealismXL_v5Ultimate.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.