"Jingle Butts" Edition*Previous Thread: https://archived.moe/y/thread/3066855*Request Thread: >>>/y/airt>To make a new thread, wait for page 8 at bump (and/or image) limit in catalog.>ONLINE GENERATORShttps://tensor.arthttps://civitai.com>METADATA & TAGS EXTRACTION/LISThttps://www.metadata2go.comhttps://rentry.org/airt-rs3https://rentry.org/povlist>/airt/'s RESOURCEShttps://rentry.org/airt-rs1 (FAQ, etc.)https://rentry.org/airt-rs2 (Setup)>QDIFFUSION (LOW-END PC)https://github.com/arenasys/qDiffusion>LOCAL FORGEUI *NEW*https://github.com/lllyasviel/stable-diffusion-webui-forge>LOCAL WEBUI SETUP *OLD*https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki>LOCAL COMFYUI *OBSOLETE*https://github.com/comfyanonymous/ComfyUIhttps://comfyanonymous.github.io/ComfyUI_examples/https://openart.ai/workflows/templates>LOCAL MODELShttps://civitai.com/models/184813/animeconfettitunexlhttps://rentry.org/5exa3 (hll+fluff) *OLD*https://gitgud.io/gayshit/makesomefuckingporn (use wayback machine) *OLD*>RESOURCESFeatures: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/FeaturesTraining: https://rentry.org/59xed3 | https://github.com/derrian-distro/LoRA_Easy_Training_ScriptsTags: https://danbooru.donmai.us/wiki_pages/tag_groups | https://danbooru.donmai.us/related_tagControlNet: https://rentry.org/dummycontrolnetIOPaint (LamaCleaner): https://huggingface.co/spaces/Sanster/iopaint-lama | https://www.iopaint.com/installAnimation: https://rentry.org/AnimAnonWildcards: https://rentry.org/NAIwildcards4chanX Catbox/NAI prompt userscript: https://rentry.org/hdgcbNAI API/prompting in auto1111: https://github.com/Metachs/sdwebui-nai-apiPony-related: https://rentry.org/ponyxl_loras_n_stuff | https://rentry.org/ponyxl_lora_previews | https://lite.framacalc.org/4ttgzvd0rx-a6jfAI Boorus: https://aibooru.online/ | https://e6ai.net/>RELATEDFurry (Male) AI art: >>>/trash/msdg
>>3072373**THREADLY NOTE**Need to find a character's template/LORA? https://civitai.com/ | Built-in tags/artist styles (Pony): https://docs.google.com/spreadsheets/d/1m2W-pZEvHuEpfHcNHrxCSr-Aw1mgtUUYho6sz9LChEA/Need inspirations and/or couldn't find the right words to describe what you want? Try this: https://danbooru.donmai.us/wiki_pages/tag_groups>>>/y/airt is also good for all sorts of questions/advices regarding AI (and requests).*Highlights from previous thread (1/5)*
*Highlights from previous thread (2/5)*
*Highlights from previous thread (3/5)*
*Highlights from previous thread (4/5)*
*Highlights from previous thread (5/5)*
Prompt: https://civitai.com/images/45450343My first LORA, gotta say, I much prefer Illustrious over Pony nowaday, it's so much more consistent.
Prompt: https://civitai.com/images/45449252Last batch
>>3072380>>3072382LORA: https://civitai.com/models/1035650Ofc i forgot the one thing i was supposed to post...
Merry Christmas!
I'm so damn excited for NAI v4's smut potential once the full version releases.tl;dr for anyone who didn't read the announcement post:- Separate prompt boxes for each character, up to 6- Rough positioning map for said characters within the image- New action tag syntax to specify who is doing what to who- Better understanding of natural language and positional phrasing- Bunch of other stuff but who caresAlso character recognition is WAY stronger. None of the prompts in picrel (except the first Boyd prompt, Kensei, and Vincent) provided any description of the character beyond name and series, but it was able to pretty accurately recreate characters with very low tagging counts, the lowest I saw being 22 (the numbers under their names are their tag counts on danbooru). Most of those were first try as well.It's definitely not perfect (couldn't do Dragon Knight from Dota: Dragon's Blood to save its life even though he has 74 posts for example), but it's pretty damn good considering no LoRAs.
Large group shots with the new separate character prompts feature can unsurprisingly be a bit uncanny/janky (like occasionally dropping or duplicating a character, or one of them having a weird expression), but the fact that it can even somewhat consistently do group shots with 5–6 distinct characters, even obscure ones, with only minimal prompting (I was mostly just doing names and hair colors), is pretty crazy. (Excuse the women, I wanted to see if it could do gender mixes and New Class VII.)
Couples/trios it handles very well though.
I'm really hoping the full version is as good at using these new tools to override bias as the SFW preview is, because if so, it's going to be huge for people who like atypical sex compositions (AKA anything that's not a single young light-skinned character getting penetrated by faceless offscreen men).Especially people who enjoy monster bottoms, older bottoms, fat bottoms, dark bottoms, anyone who the AI always wants to put on top when paired with dissimilar characters.And also people who like the tops to be in the picture, since the AI always wants to shove them out of the frame.Don't mind the middle image, I was testing how consistently it handles characters with opposite designs. (Also I got it to do a penis with inpainting. Nice.)
>>3072447Forgot to mention it earlier but all those were same prompt, same settings, same seed, just different model.Anyway here's some actual porn to make up for the sperging
>>3072448That's great. The Kiseki male characters are pretty accurate.
>>3072447NAI as in Noob-AI or Novel AI?I'm currently catching up (Illustrious), i've just finished with Pony and already feeling a bit behind...
>>3072459Sadly most of my favorite Kiseki males aren't recognized...I'm probably going to upload as many high quality references of my obscure favorites as I can to danbooru now that I know NAI can recognize them with so few images. No idea how long before that'll be reflected in the model's training though.>>3072497Novel AI, forgot there're two NAIs now sorry
>>3072517Can relate, the one I find the hottest is Nial and has like 3 entries and doesn't even appear directly on any of them D:
>>3072517May i ask if NAI still require payment to use? (Or are there "other" ways around?)
>>3072531>nial>larryKino>>3072538I think you're stuck paying for Novel AI unfortunately anon. To my knowledge v3 didn't leak and v4 probably won't either. There's that old leak (v2?) but I tried it and it seemed pretty bad compared to other models...
anyone have a list of characters, artist, or stuff you'd like for a lora for pony on tensor. I am going to start training new ones soon going to do them on tensor
>>3072497Illustrious is also compatible with Pony LORAs btw
>>3072619deep sea king (one punch man) for illustrious please
>>3072619Beowolf from Skullgirls.
>>3072724yes but they're noticeably diminished
>>3072733Not the person you're replying to, but I've done one recently for NoobAI on Civitai; I can do a Pony version as well if there's demand but Noob is a much better modelhttps://civitai.com/models/1081091/beowulf-skullgirls-noobai-xl
>>3072743Fucking fantastic!
>>3072725I will be making an improved versionhttps://tensor.art/models/814017901976864955
>>3072733mines more cartoony but its on tensor and I will be retraining to improvehttps://tensor.art/models/814014268434464079/Illustrious-Bara-Beowulf
SDXL
NTR Mix Illustrious 4.0
Stable Diffusion
ReForge
SD
>>3072805
style lora for noob (epsilon) if anyone wants: https://litter.catbox.moe/veizia.safetensorsyou probably have to manually put it into the prompt since people in the past said my loras aren't detected automatically
