"It don't mean a thing .. if it ain't got that sling!"===Flux v2.3This version should be able to generate the use/misuse of slings with a one-shot, with the right seed, or an output that won't require much img2img corrections. It can stack well with other LoRAs such as character based one or anatomy based ones like peen helpers. If your checkpoint doesn't handle male genitals very well to start off with, this isn't going to correct that so you may want to stack a penis based lora if your running into Ken Doll phenomena. It shouldn't require a keyword so long as you use generally descriptive terms such as sling, swing, harness, etc. etc. but the dataset captions do have a commonality term of 'sexsling' in case it helps or a particular checkpoint is giving you trouble.Strength of 1.0 works generally well but dialing it down also helps if you're getting some major body horror with a particular seed #.===Pony v1.0I found that a few base models, mostly Pony, can generate images with sex swings/slings, it is fairly inconsistent and not varied as far as angle and usually just limited to some straps of leather and chains.So I made this one instead of having to battle prompting with the right combo base model checkpoint.It can work on its own generating action shots with people fucking using one, a person just by themselves on it or just a sling on its own. Mostly trained with gay men but threw in some realistic hetero, silhouette and mannequin training images in the mix to help balance it out to get it to be more of a general concept than a type of gay scene scenario. Results were better than what I was expecting, especially for my first generational version of it.You can specify angle with:from abovefrom sidefrom behind (behind the sling facing the top)from belowI didn't go so far as to tag categorization of different types of slings (maybe in a future version), but with the right prompting you'll probably get what you're looking for without too much fuss.Depending on the base model, it is fairly stable with the use of other LoRAs. I threw in some examples in the image gallery. Most of the posted images are how they were generated without img2img retouch or other post image generation editing.Depending how the other LoRA was trained, it'll either work right out of the box with it at 0.6 to 1.0. Some LoRAs just don't play nice with others, so if you find that it's not working after adjusting weighting of either this one or the other LoRA, then you probably are better off doing a txt2img then img2img inpaint workflow.I'd be interested to see some of your variations using it or if you have suggestions on how to make it even more compatible/stable when piling on other LoRAs, then lemme know.So try it out and have a real swingin' slingin' time out there, folks!

Sex Sling - A Slingy Dingy Good Time! (FLUX &  Pony)

Generated from approx 240 HQ image dataset

Files with SHA256: e2c5855e62813230abf490b6b48524ae157b7028d66c99f967b0516aeb4e82e2