John6666

/

iniverse-mix-xl-sfwnsfw-guofen-v15-sdxl

like 0

This model is a .safetensors that was placed in Civitai and converted to Huggingface Diffusers’ format. I have no idea where it was first published, but if you keep an eye on the following you won't miss it.
iNiverse Mix XL(SFW & NSFW)
https://civitai.com/models/226533/iniverse-mix-xlsfw-and-nsfw?modelVersionId=608842

thank you !

can you upload fully real XL 10 ?

https://civitai.com/models/227059/fullyrealxl

this model has been removed but if you find it elsewhere can you upload it ?

Sorry, I don't have one.😅

hey @John6666 can you upload cookie-run-character-style?

https://civitai.com/models/16068/cookie-run-character-style

i like this model by the way :p

I did it, just put the base model repo name in README.md since it's LoRA.
https://huggingface.co/John6666/cookie-run-character-style-v1-sd15-lora
Any LoRA that is fine with weights=1.0 should work this way. (as long as there is a working base model on the huggingface)

alright thanks you

It looks like the HF people are working hard on debugging, so I'm just free while I wait.
If the HF people were throwing away the debugging, even I would run away.😇

By the way, I don't know what other people's demands are, so it's actually easier if I get a moderate number of requests.
If there are too many requests, it would be difficult.
Maybe this is the same for HF people and developers in general.

thank you for uploading models John !

When you'll have time could you upload the V2.0 of the same model : https://civitai.com/models/578500?modelVersionId=667338

Comments says its the best version of this model

I've heard that sometimes a well-done old version is created that the author can't get past.
Anyway, it is very easy to copy and paste more than 90% of the README.md.
https://huggingface.co/John6666/yaprm-yet-another-pony-realistic-merge-v20-sdxl

i go always for last version and then i can miss good old version. It has already happened to me with juggernauth model. Thank you !

hello john,

Do you know when AI architecture will offer 2048x2048 native pictures ? 1024x1024 was a huge shift from 512x512 but 2048x2048 will be so great !

No, I have no clue about the roadmap.
That said, I'm scared of how many times more VRAM will be consumed by unet alone if it's twice as long and twice as wide.
The unet (transformer) file for Flux is also roughly 4 times the size of the SDXL...

Personally, I think the text encoder needs to be improved first. That is especially the starting point for image generation AI development in non-English speaking countries.

So let's cheat with up-sampling for now.

Hello John,

is there anyway to upload lora Flux model from civitai and use it on inference api page like this author flux model on hf : https://huggingface.co/punzel ? he managed to put lora model and we can compute on api inference but i dont know how long this lora will be up.

there is a lot of author in civitai specialized on Flux lora who do amazing work we can check there https://civitai.com/tag/celebrity?sort=Newest or