Join my new Discord here! <a target="_blank" rel="ugc" href="https://discord.gg/cnVvnkgtPb">https://discord.gg/cnVvnkgtPb</a>Belle Delphine (Face &amp; Body LoRA) - the infamous e-girl. This LoRA includes Belle Delphine's face AND body. Use the trigger words to get the desired result. Infinite Belle pics await you.NSFW Previews: <a target="_blank" rel="ugc" href="https://jpg.pet/aihotgirls">https://jpg.pet/aihotgirls</a> <h2 id="recommended-settings">Read Me!</h2>TWO VERSIONS:<ul><li>Belle Delphine RHM v1: Trained on <a target="_blank" rel="ugc" href="https://civarchive.com/models/63482/real-hot-mix">Real Hot Mix</a>, my own mix. I use this LoRA for extremely high quality photorealistic results<ul><li>For txt2img &amp; img2img: LoRA weight between 0.9-1 for best results</li></ul></li><li>Belle Delphine SD1.5 v1: Trained on SD 1.5 base model, compatible with all models<ul><li>For txt2img &amp; img2img: LoRA weight between 0.5-1 depending on checkpoint model for best results (lower for anime/non-photorealistic models)</li></ul></li></ul><h2 id="recommended-settings">Recommended Settings</h2><ul><li>Hires fix: 0.2-0.3 for face, 0.2-0.4 for full body</li><li>Face is best at 1:1 resolution (make sure inpainting is 1:1 resolution)<ul><li>Other resolutions work as well (see Training Info: resolution sizing for training dataset aspect ratio info)</li></ul></li><li>'makeup' is a trigger word. Put it in negative and use 'brown hair' to get Belle with no makeup</li><li>This LoRA is NSFW skewed. Use 'clothes' in positive and '(nude:1.5), (breasts:1.5)' in negative if you want SFW pics</li><li>SOMETIMES:<ul><li>If generating on not 1:1 aspect, her face may become slightly elongated. Maybe 1 in 5 images</li><li>'pink hair' occasionally makes humongous hair. Not sure why this happens. I use 'huge hair, long hair' in negative if that seems to be happening with my prompt</li></ul></li></ul><h2 id="recommended-settings">Training Info</h2><ul><li>5 epochs, 25 repeats, 36 images<ul><li>21 brown hair, 10 pink hair, 5 blonde hair pics (if unspecified, brown hair will usually default)</li><li>22 face, 14 full body pics</li></ul></li><li>For resolution sizing, enable buckets = true (25 = 1 image)<ul><li>bucket 0: resolution (384, 640), count: 25bucket 1: resolution (384, 704), count: 75bucket 2: resolution (448, 576), count: 100bucket 3: resolution (448, 640), count: 25bucket 4: resolution (448, 704), count: 100bucket 5: resolution (512, 576), count: 25bucket 6: resolution (512, 640), count: 75bucket 7: resolution (576, 576), count: 450bucket 8: resolution (640, 448), count: 25</li></ul></li></ul><h2 id="background">Background</h2>I've been experimenting with LoRA training and seem to have found a sweet spot for epic photoreal accuracy for people. I didn't realize there were already so many Belle LoRAs while making this, but wanted to share anyway. I believe this one is the BEST one on Civit.

Belle Delphine (Face & Body LoRA)

5 epochs, 25 repeats, 36 images<ul><li>21 brown hair, 10 pink hair, 5 blonde hair pics (if unspecified, brown hair will usually default)</li><li>22 face, 14 full body pics</li></ul>

Beldel LORA RHM 5E 25 576 Cosine.safetensors

Belle Delphine RHM v1

Belle Delphine SD1.5 v1

Join my new Discord here! <a target="_blank" rel="ugc" href="https://discord.gg/cnVvnkgtPb">https://discord.gg/cnVvnkgtPb</a>Belle Delphine (Face &amp; Body LoRA) - the infamous e-girl. This LoRA includes Belle Delphine's face AND body. Use the trigger words to get the desired result. Infinite Belle pics await you.NSFW Previews: <a target="_blank" rel="ugc" href="https://jpg.pet/aihotgirls">https://jpg.pet/aihotgirls</a> <h2 id="recommended-settings">Read Me!</h2>TWO VERSIONS:<ul><li>Belle Delphine RHM v1: Trained on <a target="_blank" rel="ugc" href="https://civitai.com/models/63482/real-hot-mix">Real Hot Mix</a>, my own mix. I use this LoRA for extremely high quality photorealistic results<ul><li>For txt2img &amp; img2img: LoRA weight between 0.9-1 for best results</li></ul></li><li>Belle Delphine SD1.5 v1: Trained on SD 1.5 base model, compatible with all models<ul><li>For txt2img &amp; img2img: LoRA weight between 0.5-1 depending on checkpoint model for best results (lower for anime/non-photorealistic models)</li></ul></li></ul><h2 id="recommended-settings">Recommended Settings</h2><ul><li>Hires fix: 0.2-0.3 for face, 0.2-0.4 for full body</li><li>Face is best at 1:1 resolution (make sure inpainting is 1:1 resolution)<ul><li>Other resolutions work as well (see Training Info: resolution sizing for training dataset aspect ratio info)</li></ul></li><li>'makeup' is a trigger word. Put it in negative and use 'brown hair' to get Belle with no makeup</li><li>This LoRA is NSFW skewed. Use 'clothes' in positive and '(nude:1.5), (breasts:1.5)' in negative if you want SFW pics</li><li>SOMETIMES:<ul><li>If generating on not 1:1 aspect, her face may become slightly elongated. Maybe 1 in 5 images</li><li>'pink hair' occasionally makes humongous hair. Not sure why this happens. I use 'huge hair, long hair' in negative if that seems to be happening with my prompt</li></ul></li></ul><h2 id="recommended-settings">Training Info</h2><ul><li>5 epochs, 25 repeats, 36 images<ul><li>21 brown hair, 10 pink hair, 5 blonde hair pics (if unspecified, brown hair will usually default)</li><li>22 face, 14 full body pics</li></ul></li><li>For resolution sizing, enable buckets = true (25 = 1 image)<ul><li>bucket 0: resolution (384, 640), count: 25bucket 1: resolution (384, 704), count: 75bucket 2: resolution (448, 576), count: 100bucket 3: resolution (448, 640), count: 25bucket 4: resolution (448, 704), count: 100bucket 5: resolution (512, 576), count: 25bucket 6: resolution (512, 640), count: 75bucket 7: resolution (576, 576), count: 450bucket 8: resolution (640, 448), count: 25</li></ul></li></ul><h2 id="background">Background</h2>I've been experimenting with LoRA training and seem to have found a sweet spot for epic photoreal accuracy for people. I didn't realize there were already so many Belle LoRAs while making this, but wanted to share anyway. I believe this one is the BEST one on Civit.

Files with SHA256: 71c7d762df5fcce65d5cef412a814b0efd236ed62bb44df7c6d77914583b98f7