It would be nice to have these basic things explained since they all share the same page, wouldn't it?

They're version numbers. Higher is better. A means alpha. B means beta. Use 3b, it's the latest and best.

And it is explained. Says it uses the updated lazymix base. So you go the lazymix model page and find out what's changed.

@nonamebrand Version numbers say little sometimes because Higher being better is not always the case. Often models will be better for an older version for X type of gens and higher version for Y, etc...many model pages on this site testify to that sort of thing, so it's not unreasonable for me to ask for a better description of model differences since you have ZERO currently mentioned vs. sending me to another page. Thanks anyway.

@clevnumb I'm not the creator of the model and even I could figure it out lol

@nonamebrand Good for you. I prefer it done how most models on this site are done is all.

I am not getting realistic pics. do you have any suggestions?

@TulisP "deformed, anime" ONLY in the negative. do not add "low quality, worst quality, funky arms, mutated pussy" shit.

try "dark lighting" or "studio lighting" in the prompt. again, no "masterpiece, ultra HD 8K, highly detailed skin" garbage.

Kaylazy did not disclose the recipe, When I used this great model, I could see that it is close to Reliberate - v2.0, but more spicy. In any case, the confirmed ones are Dreamlike Photoreal 2.0 and SubReddit V3 models

@Diioo That is something I found here on civit as a model string on an image post by Kaylazy: 0.8(0.75(0.67(0.5(LazyBase_v1-2 + 0.95(subredditV6 - sd-v1-4)) + 0.5(urpmv13)) + 0.33(basedModel_v10)) + 0.25(epicrealism_pureEvolutionV3)) + 0.2(aZovyaPhotoreal_v2) + 0.15(N_sxd_10Pruned - Models_v1-5-pruned-emaonly) I just have no clue what basedModel and N_sxd_10 are. I know the other models but not those 2 well and LazyBase

https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/How-to-make-your-own-Inpainting-model

It's fairly straightforward to make your own! The lazymix inpainter I made following these instructions works wonderfully.

@j8888498159 could you maybe share yours?

save the seed and use same variation seed. then use same prompt.

@KDOG Y como puedo hacer esto, podrian ayudarme por favor

It can. its just a horny model so you might have to be firm with the horny bat if you get what I mean.
Like using negative prompts like nude, naked, nsfw and describing the clothes the subject is wearing to deter it from drawing the subject naked

@firedelta5 I see, thx

@FriskTemmie You could download Nag_'s Un-horny LoRa and turn it all the way the max.
At least that worked for me.

Link: https://civitai.com/models/101567/concept-un-horny

Use some of the NSFW Loras, they do a pretty good job

you can make your own: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/How-to-make-your-own-Inpainting-model

i created one for you: https://mega.nz/file/WIEFRSyK#6umi_a0YL50BzzGA1Z9xrNmkGSg2Nfw1teo4VHIJyyQ

@pkmngotrnr Serious question. Why do you need inpainting? I find using auto1111 it works with any of them. I know it is a noob question but if you have time let me know.

@KINGLIFER can you maybe tell me? i have no clue and research in this topci seems like overwhelmingly lots of conntect that is outdated....any keywords or specific tutorials to look out for?

@pkmngotrnr hey man thanks for the checkpoint merge. I tried to do my own, but I effed up at some point I guess and was getting weird images, so I just downloaded yours and it works for inpainting

@KINGLIFER you don't, unless you're using it for inpainting specifically. Checkpoints do not work for inpainting unless stated otherwise (some have an inpainting version as optional download). Same with LORA's, some work, some don't. But if a lora isn't working for inpaint, nothing will happen. If the checkpoint is not meant for inpainting, you'll get completely screwed up images.

Seed #'s. Play around with them. Keep he first few digits of a seed you like (say the first 4) and randomize all the ones there after, repeat.

Repeat again.

i would suggest using reactor. download/save the face model you want and then use reactor so it will faceswap on the last step of generation

@Legend_of_Chavo_G thanks!

@blackfirelouis thanks!

trigger words are for loras, this is a checkpoint. so like the image base that then gets influenced by loras, if you select it as a checkpoint in lets say a1111 it'll be used in every single txt2image or img2img prompt you provide, no need for trigger words.

Is a post-processing tool, find it in your webUI settings

Mix - two or more checkpoints mixed together. Vae - a decoder/encoder for latent image. It's not really important, the creator just mentioned that if you want to recreate his pictures you'd need to use a different vae that's not included in this checkpoint.

It's made with 512*512 in mind. Generate at lower res and use an upscaler.

For inpainting you usually need a model that's specified it's for inpainting. I recommend EpicPhotoGasm, it has a version that's good for inpainting.

I agree. I usually go to EpicPhotoGasm for good and accurate hands, and this one for everything else when generating pictures of girls