Woman with perfect and large breasts with a great cleavage. Easier with the lora trained for this purpose. The images have been tested with various semirealistic and realistic checkpoints. Trained on Civitai with 180 images (v3 with 350 images and work with nsfw style). Work better alone or with max 2-3 loras. Read info.ILLUSTRIOUS version ----&gt; <a rel="ugc" href="https://tensorhub.art/models/998648422791655363/Massive-cleavage-ILL-v1">here</a>It is easier to increase the dimensions of the breast by adding "big breasts" or "huge breasts" and you can also put "sagging breasts" in the prompt. The Lora has been trained with "covered nipples" with not NSFW images.SD v1 - Weight : 0,1-0,4 with a lora character and 0,5-0,8 for a style.Prompt example:Masterpiece, 4k, detailed, realistic, a beautiful girl, wearing bikini, in a pool, massive_cleavage, sagging breasts, long hair, outdoor, covered nipples, ...SD v2 / 2.1 - better cleavage and form of breasts. Weight: 0.6-0.8. Tested with my RealPhoto model or other realistic model (not modded), (not perfect compatible with Boobalicious and Boobs Planet model, make breasts too big). Use for SFW images.SDXL v1 - Weight: 0.4 to 0.7. Trained and tested with my realphotoXL model. Remember to use trigger words. See prompt. Clip skip 1 or 2 (trained with 1 but i use 2).SD v3 - This new version for SD is able not only to improve the breast, but also to enlarge it and also improve the general details of the image. I recommend using it with maximum weight of 0.8, but you can reach up to 1.0. At most weight we will have better details and bigger breasts. The words "Breasts" or "Big Breasts" or "Gigantic Breasts" should not be written in the prompt because the Lora makes everything necessary. The most important thing is to insert the trigger words "massive cleavage, sagging breasts". This version work with NSFW style (put in positive prompt "nipples, large areolas, heavy engorged breasts" for nsfw and "nipples, nsfw" in negative prompt for SFW images).SDXL v2 - Is not a slider, but you can have a better breast. Trigger words are facoltative, but you can use this "heavy engorged breasts, massive cleavage, sagging breasts". Weight 0.4 to 0.9. Tested with my RealPhoto XL model.Use this negative prompt : DeepNegative_xl_v1, unaestheticXL_Jug6, bad pose, deformed, 3D rendering, fat body, watermarkIMPORTANT, PLEASE READ<span style="background-color:rgb(26, 27, 30);color:rgb(250, 82, 82);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">: If you get different results, the prompt was probably not followed and an unsuitable checkpoint was used. <span style="background-color:rgb(26, 27, 30);color:rgb(253, 126, 20);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">If it doesn't work well for you, the first thing you should do is try the Lora by itself, without embeddings or other Loras<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">. All my LoRAs have been trained on realistic models for A1111. The videos I post represent what you can achieve; even though the video is animated, it reflects the quality and what the LoRA can do. The first frame of the video is the image I created using the LoRA and then animated, using the parameters I recommended (see other note in this descriprion). To achieve a realistic result, a <a target="_blank" rel="ugc" href="https://tensorhub.art/models/869285461682562510/Illustrious-SDXL-Realistic-v1">realistic</a> ILLustrious non-merged checkpoint should be used (or <a target="_blank" rel="ugc" href="https://tensorhub.art/models/968604476942323137/Real-Illustrious-SDXL-Final-version">Semirealistic</a> based on realistic checkpoint) <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">. If you use merged models with anime models, the result may be very different from mine. Do not use corrective loras<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> in addition to the main lora because they often only ruin the result and try lower weight first. If possible, do not exceed 150 tokens in prompts<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">. Too many words in prompts do not improve the result. On ComfiUi you could have different results.<h3 id="please-donate-buzz-and-vote.-to-make-these-loras-civitai-asks-for-over-1500-buzz-!-if-you-donate-buzz-i-can-continue-to-train-new-loras.-thank-you.">Please donate Buzz and vote. To make these Loras, Civitai asks for over 1500 buzz ! If you donate Buzz I can continue to train new loras. Thank you.</h3><hr />I hope you like it, I will see from your votes and pictures. If the result is positive (as I hope), I will continue to share other goodies.<hr />Recommended for a better result: ADETAILER mediapipe face mesh with impaint 0.55Setting : Sampling Euler-a or DPM Karras, HiresFix on, R-ESRGAN 4x+ anime, Denoising 0.4, Upscale 2x, Sampling steps 35+, Resolution 512x768, Upscale 4x Ultra Sharp next passage.<hr />It is strictly prohibited to distribute outside civitai this model or hosting services in third-party image generation services, or sell my images on other site. You can use this model for personal use on Civitai or A1111 on your pc. Not for sell.If you like the character, always leave a vote and an image, and if you give some buzz you will allow me to train other LORA or MODEL.

Massive_cleavage SD1.5 - SDXL

massive_cleavage.safetensors

Massive cleavage

SD v1.0

Files with SHA256: 5ba732430b84b256364ce6686be3aaeb2d58ce06d80463c0001edf74d9f49119