V5 (in testing) - Cinematic and film still datasets, adding more lighting control, trying to improve outdoor ground textures. huge thank you to @Stagnation and @Lovebyte for donating almost all the buzz I needed for v4! V4 - NONE OF THE EXAMPLES USE ANY ADDITIONAL RESOURCES. You may see tags in the prompts that look like they use extra resources, they do not. I just copy pasted some prompts as they were to see what happened, with the exception of removing negative embeddings from the original prompts. Now for the updates; Faces - I feel like this was the biggest improvement with this update. Many prompts (not all) I tend to get different faces from seed to seed. It's not always the case depending on the prompt, but certainly more than I've seen in my previous versions/other pony models overall. I've trained several datasets on nothing but faces, all with at least 'photo of a girl, 1girl' in the captions to try and cover as wide of an area as possible. But I've trained over 1500 faces, every ethnicity I could think of, multiple different ways, and merged in them several times. When using face detailers, sometimes higher strengths get rid of that facial diversity. When possible I recommend using a lower strength (like 20%) to keep unique results. LORA compatibility - Fictional Character and Style loras work very well from my testing (but the model knows many characters on it's own as well. more than I expected honestly.) I've also worked to TRY and improve 'real person' lora compatibility. I've had mixed results so I'm curious to see what you all can do. V3 - I've trained many new batches of my photography, more film, large commercial fashion datasets, and large batches of stock photography into this version. It's capable of some pretty awesome stuff, I'm not happy with where environment textures are and I'm trying to improve that for v4, but portrait wise it's come a long way. All of the main examples above are without Loras, but I have done some testing and it seems to work well with all that I've tried. I haven't messed with too many 'real person' loras though, so let me know how those perform either way. That said, it seems to understand characters very well still without Loras, even some more obscure ones. ALSO, I'm still training more of my personal datasets to merge into this further, so if you like my work please consider donating some buzz to help fund future versions :) (thanks so much for those that have donated so far, it really has helped. I'll be releasing some standalone loras as well themed around different photography elements soon too!) *V2* - This one is special, as I've trained it on some of my personal work as a professional photographer for two decades. Cant wait to see what you all do with it, it definitely has some weaknesses (let me know if you encounter any.) But I'm really happy with a lot of what I've gotten from this version so I decided to upload it. ___________________________________________________ Note - CivitAi uploading doesn't automatically import my metadata, so I have to add everything manually. All of the main examples should have it, if there's any others below you want lmk. This is a pretty massive fine tuned mix, capable of a lot, and trained on a lot of characters. That said, I have no idea what it's fully capable of. I'm pretty new to pony, and most of my images here are prompts I've found on here, and ALL of them are generated without Loras, so I'm excited to see what people can do with all of their extra resources and detailers and whatnot. Most of my testing is without any resources so that I could ensure it was accessible for everyone, but I'm stoked to see what everyone else makes and I REALLY HOPE PEOPLE SHARE THEIR RESULTS. If you like the model and want to help get newer versions out, donating some buzz to help train is greatly appreciated! Hey Fenn, there's a TON of merges out there, what makes yours different? As with my SD1.5 models before, I have only merged original trained datasets. This isn't another 'Deliberate+Dreamshaper' merge. Unlike my SD1.5 models before, I have trained personal Loras and merged them, as well as many other Loras from here. My usual prompt settings: + score_9, score_8_up, score_7_up, photo, realistic, Film still, cinematic, - score_6_up, score_5_up, score_4_up, source_pony, source_anime, source_furry, source_cartoon, worst_quality, low-quality, 3d, cgi,

fennfoto [PONY]

fennfotoPONY_v4.safetensors

V4 - NONE OF THE EXAMPLES USE ANY ADDITIONAL <a target="_blank" rel="noopener noreferrer nofollow" href="http://RESOURCES.You">RESOURCES.You</a> may see tags in the prompts that look like they use extra resources, they do not.I just copy pasted some prompts as they were to see what happened, with the exception of removing negative embeddings from the original prompts. Now for the updates;Faces - I feel like this was the biggest improvement with this update. Many prompts (not all) I tend to get different faces from seed to seed. It's not always the case depending on the prompt, but certainly more than I've seen in my previous versions/other pony models overall. I've trained several datasets on nothing but faces, all with at least 'photo of a girl, 1girl' in the captions to try and cover as wide of an area as possible. But I've trained over 1500 faces, every ethnicity I could think of, multiple different ways, and merged in them several times.When using face detailers, sometimes higher strengths get rid of that facial diversity. When possible I recommend using a lower strength (like 20%) to keep unique results.LORA compatibility - Fictional Character and Style loras work very well from my testing (but the model knows many characters on it's own as well. more than I expected honestly.) I've also worked to TRY and improve 'real person' lora compatibility. I've had mixed results so I'm curious to see what you all can do

Files with SHA256: 5cbacfe393327d1c659f8df16928bf25fe404dd27c4d9d0af375c484ea271af5