Get Big Love Eryn1, Ultra5, Photo6, Zuna1 or Zia1 for $5.50 and Gwen1 or Hyper1 for $10 on Fanvue. Click 'Follow', buy a post, check your Fanvue DMs! You can get Klein2 for $10 on Tensor Art. A subscription includes the latest release and new ones within 30 days.
To create up to 85 free images per day in highest quality you can use Big Love here. For discussing photorealistic image generation join me on Discord.
Big Love Gwen is a finetune of Qwen Edit 2511 and can do NSFW image editing. It produces more realistic and sharp images with better skin texture & genitals than other Qwen checkpoints/loras. Txt2img is less reliable, but generates nice results too.
Big Love Gwen1 allows commercial usage without additional fees (if no auto-generation involved) unlike Big Love Klein. With its better anatomy it is easier to get good editing results. As fast at 6 steps as Klein at 4 steps.
The fp8 mixed version is for 16+ GB VRAM (highest quality), the nf4 version is for 12 and 16 GB VRAM (25% faster), the GGUF Q5_K_M version is for 16 GB VRAM (softest) and the nvfp4 version requires 24+ GB VRAM (on RTX 5090 2-3x faster, otherwise 2x slower, lower quality than fp8 mixed, less saturation, textured).
It is recommended to use ComfyUI. The Gwen1 workflows offer a skin enhance feature that is not available in Forge Neo. Move the Gwen1 model file into the diffusion_model sub folder of the models folder of ComfyUI, qwen_2.5_vl_7b_fp8_scaled.safetensors into the text_encoder sub folder and qwen_image_vae.safetensors into the vae sub folder.
Settings are 6 steps (3 steps for upscaling), cfg1, Euler sampler, Beta scheduler, 832x1216 or 1024x1536 pixels.
Big Love Eryn is a fine-tune of Ernie-Image(-Turbo). You can run it locally with ComfyUI and Forge Neo. The fp8 version works with 8 & 12 GB VRAM. Use the (pruned) bf16 version if you have 16 GB VRAM or more. The "Full" versions are only meant for training a lora (coming soon!). Put the downloaded model into the diffusion models sub folder of ComfyUI or the StableDiffusion sub folder of Forge Neo. You additionally need to place ministral-3-3b.safetensors in the text_encoder sub folder and flux2-vae.safetensors in the VAE sub folder.
Settings are 4-6 steps (3-4 steps for upscaling), cfg1, Euler sampler, Beta scheduler, 832x1216 or 1024x1536 pixels.
Big Love Hyper supports generating at image sizes of 1664x2232 (4 megapixel) and higher with SDXL. Can be easily upscaled to 9 megapixel unlike the normal low-res images. Hyper currently has not fully stable anatomy, so needs more runs. Some prompts work better than others. It produces highly detailed and life-like images close to real photos. It has a special quality that cannot be achieved by just upscaling. The Hyper versions of Big Love are pay-only and special license conditions apply (see below).
Settings are DMD2 lora with strength 1.0, LCM Exponential, 8 steps, cfg 1, clip skip 2, 1664x2432 or 1536x2240. Upscale with img2img with DMD2 lora with strength 1.0, same prompt, LCM Exponential, 1.25-1.5x, 4-8 steps, cfg 1, 0.3-0.5 denoise, clip skip 2.
Big Love Klein is a fine-tune of Flux 2 klein. It can do txt2img but also edit images with prompts. Txt2img is currently experimental and requires more prompt & seed hunting. Image editing works stable. It only needs 4 steps, so is quite fast (5 sec on 5090, 15 sec on 5070). You can run it locally with ComfyUI and online on Tensor Art. Use the GGUF Version with 8 GB VRAM, the fp8 version with 12-16 GB VRAM and the (pruned) bf16 version with 24+ GB VRAM. The nvfp4 version (full nf4) renders 2-4x faster on RTx 5xxx GPUs with the latest CUDA drivers. The base version (full fp32 or full bf16) is meant for training a lora only. Put the downloaded model into the diffusion models sub folder of ComfyUI. You also need qwen_3_8b_fp8mixed.safetensors in the text_encoders sub folder (Qwen3-8B-Q5_K_M.gguf wth 8 GB VRAM) and flux2-vae.safetensors in the vae sub folder. Forge Neo is not recommended, because it currently produces too low image quality with Klein, which will hopefully change soon.
Settings are 4 steps (2 steps for upscaling), cfg1, Euler sampler, Beta scheduler, 832x1216, 1024x1536, 1280x1920 or 1536x2240 pixels
Big Love ZT, Zia or Zuna are fine-tunes of Z-Image(-Turbo) on the Big Love dataset. You can run it locally with ComfyUI and Forge Neo. The fp8 version works faster with 8 GB VRAM but produces more noise so may sometimes look more realistic, but details are not as good. Use the (pruned) bf16 version if you have 12 GB VRAM or more. The "Full" version is only meant for training a lora. Put the downloaded model into the diffusion models sub folder of ComfyUI or the StableDiffusion sub folder of Forge Neo. To make Big Love ZT/Zia/Zuna work you additionally need to place qwen3_4b.safetensors in the text_encoder sub folder and ae.saftendors in the VAE sub folder.
Settings are 8 steps, cfg1, Euler or DPM++ 2s a RF sampler, Normal or Simple or Beta scheduler, 832x1216, 1024x1536 or 1280x1920 pixels. For upscaling 8 steps, cfg1, Euler sampler, Normal or Simple or Beta scheduler, 1.5x upscale.
Big Love Ultra supports generating at image sizes of 1280x1870, 1360x1984 and higher. Can be easily upscaled to 6 megapixel unlike the normal low-res images. More photorealistic and detailed. A special quality that cannot be achieved by just upscaling. Ultra produces more details & sharpness when upscaling and detailing low-res images. The Ultra versions of Big Love are pay-only and special license conditions apply. See below.
Settings are DMD2 lora with strength 1.0, LCM Exponential, 8 steps, cfg 1, clip skip 2, 1280x1864 or 1360x1984. Upscale with img2img with DMD2 lora with strength 1.0, same prompt, LCM Exponential, 1.25-1.5x, 4-8 steps, cfg 1, lora strength 1.0, 0.3-0.5 denoise, clip skip 2.
Big Love Photo and Insta1 are finetuned versions of XL. Lust1 is a finetuned version of Lustify merged with Photo. They were trained on thousands of images and dozens of new concepts. Why they are the most realistic and versatile. For more details and prompting Photo1 read this article.
Settings are DMD2 lora with strength 1.0, LCM Exponential, 8 steps, cfg 1, clip skip 2, 832x1216 or 1024x1496. Upscale with img2img with DMD2 lora, same prompt, LCM Exponential, 1.25-1.5, 4-8 steps, cfg 1, lora strength 1.0, 0.3-0.5 denoise, clip skip 2.
Big Love XL is a combination of 5 different training branches of SDXL: original SDXL, bigASP, NatVis, Anteros and Pony. With a bit of Pony in it, some Pony loras (poses, characters) work too. It outputs a more photorealistic & creative look than the Pony versions and can also do paintings & cartoon.
Settings are DMD2 lora with strength 1.0, LCM Exponential, 8 steps, cfg 1-1.5, clip skip 2, 832x1216 or 1024x1496. Upscale with img2img with DMD2 lora, same prompt, LCM Exponential, 1.5-2x, 8-12 steps (XL1/XL2), 4-8 steps (XL2.5-XL4), cfg 1, lora strength 1.0, 0.3-0.5 denoise, clip skip 2.
Big Love Pony produces different images than both Pony & SDXL-based models. Pony tags are understood, but interpreted a bit differently. Some images lean more in the one or other direction depending on the prompt. It works with Pony loras as well as normal SDXL ones. It can turn anime/cartoon into photorealistic images in img2img. Pony3 includes a bit of Illustrious.
Settings are DMD2 lora with strength 1.0, LCM Exponential, 8 steps, cfg 1-1.5, clip skip 2, 832x1216 or 1024x1496. Upscale with img2img with DMD2 lora, same prompt, LCM Exponential, 1.5-2x, 8-12 steps, cfg 1, lora strength 1.0, 0.3-0.5 denoise, clip skip 2.
ComfyUI Workflow:
Download one of the following images and drag'n'drop it onto Comfy:
Klein, Eryn and Gwen workflows are embedded in the showcase images.
Z-Turbo/Z-Base txt2img & Upscale Workflow (recommended)
SDLX DMD2 txt2img & Upscale Workflow (recommended)
SDXL DMD2 txt2ing & Upscale for Big Love Ultra
SDXL DMD2 img2img Workflow
SDXL Lightning Workflow
Trained Concepts
Newer versions of Big Love include the concepts of older versions, e.g. Photo4 can do everything that Photo1 to Photo3 can do and more. The only exception are the Insta concepts, but they have been partially added to never versions too.
Photo6/Ultra5: ai-style, bodyscape, butt plug, candid amateur, cumshot, illustrious-style, outdoor sex, partner fingering, photoart, innie pussy, ray-style, rebellious, reflection, sensual, sexy ass, sweetheart, wedgie
Photo5/Ultra4/Hyper1: alluring, anal sex, artistic photo, artistic pinup, charming, cum portrait, cute asian, cute pinup, cute portrait, double oral, dreamy-style, face pov, fantasy pinup, fantasy-style, kai-style, lowkey, mj-style, outdoor flashing, photo art, pole dance, public indecency, public nudity, rock climbing, scify-style, sword pose, vaginal sex
Photo4/Photo4.5/Ultra2/Ultra3: 2girls kissing, acrobatic sex, average face, beautiful face, amateur pinup, anal fingering, artistic portrait, cuddling, cum swapping, doggystyle pov, even skin color, fashion photo, female masturbation, female pov, finger licking, food porn, footjob, french kiss, full nelson, funny movie, funny tongue out, insta cute, legs pov, low contrast, mirror portrait, nude pose, oral pov, orgasm face, outdoor nude, penis licking, pinup photo, pro photo, pro portrait, pussy licking, sex pov, sexy portrait, vaginal fingering, waterfall
Photo3/Lust1/Ultra1: 360 degree photo, 3d selfie, 69 position, adorable, alt beauty, anal gape, anal sex, average face, beach life, beautiful, beautiful face, candidness, cum, cute, deepthroat, dildo insertion, double exposure, enthusiastic, extreme pose, faces, fashion photo, female pov, fisting, follow me pov, girl next door, gorgeous, handjob, heroin chic, huge feet, huge hands, huge nipples, light rays, light streaks, lighting, low/mid/high contrast, lowkey/midkey/highkey, low/mid/high saturation, middle finger, motion blur, object insertion, detailed eyes, teeth, pro photo, pussy gape, real life, rooftop, round ass, sad, sexy pose, skin texture, skin tone, spread pussy, squirting, street portrait, testicle licking, testicle sucking, vaginal sex, voluminous hair, water splash
Insta1: amateur photo, fashion photo, insta selfie, pinup photo, pro photo, real life, big ass, bimbo, cute, luxurious, adorable, enthusiastic, average face, beautiful face, detailed eyes, sexy legs.
Photo2: 3d selfie, 69 position, adorable, alt beauty , amateur photo, ball licking, ball sucking, beach life, bubble butt, low/mid/high contrast, dildo insertion, double exposure, enthusiastic, extreme pose, female pov, follow me pov, forced perspective, girl next door, gorgeous, handjob, heroin chic, huge feet, huge hands, huge nipples, light rays, light streaks, lighting, middle finger, motion blur, no tan lines, object insertion, pussy gape, rooftop, sad, low/mid/high saturation, sexy pose, squirting, street portrait, water splash.
Photo1: amateur photo, pro photo, lighting, lowkey, midkey, highkey, beautiful, cute, candid, real life, average face, remarkable face, voluminous hair, skin tone, skin texture, 360 degree photo, spread pussy, deepthroat, fisting, anal gape, anal sex, vaginal sex.
Training a (character) lora
My advice is this: Take 10-30 high quality images (at least 50-100 for styles or poses), download OneTrainer, click the SDXL lora preset, choose Big Love as the base checkpoint, add your images as a concept, add tags on the Tools tab, and start training. If you train on Big Love Ultra change the training resolution to 1536 and provide 1248x1832 images.
You will only get the full Big Love quality by training on real photos or extremely photorealistic images, which is usually not the case with a fictional character. It is all about image quality and not choosing boring images. Don't add posing images to a character lora, because Big Love provides the posing later for it. Focus mainly on good portraits with a few upper and full body images. Rather fewer higher quality images than a lot of average images. Read here for traing a lora on Big Love ZT2, ZT3, Zia1, Zuna1 or Eryn1.
Fixing Anatomy
If you want to keep the seed and repair anatomical problems in an image, activate the Extra checkbox in A1111/Forge and set Variation Strength to 0.01 to 0.1. Then generate until anatomy is fine. In Comfy there are sampler nodes that support variation seed (e.g. Inspire Pack), which do the same thing. Also possible to change steps or add commas to the prompt, but I do not recommend it. Extra/Variation Seed is more powerful and convenient as it gives you endless variations with just one click. There is always a variation which looks similar or even better with great anatomy.
You don't need negative prompts for this. They are rather ineffective. You can also change the positive prompt to suppress things. Anatomy problems are also created by the prompt if you prompt contradicting poses that the model cannot combine. Better to fix the prompt first then.
Reducing Saturation, Warm Colors and Sharpness
These SDXL loras reduces the saturation and change the color temperature without reducing image quality. The lora strength defines the effect. Negative values increase saturation or add more warmth. To make an image softer use a lower strength for the DMD2 lora, e.g. 0.95, 0.9, 0.85.
Big Love Hyper/Ultra/Photo Sizes
Big Love Hyper supports these image sizes:
1664x2432 (recommend)
1536x2240
Big Love Ultra supports these image sizes:
1360x1984
1280x1872 (recommend)
Big Love Photo supports these image sizes:
1024x1496 (recommend)
832x1216
You can also use the 16:9 or 2:3 versions of these image sizes as well as landscape orientation, but the above should be more reliable.
The higher the image size, the lower the probability of good results. If you get too many problems with a higher image size, switch to the lower one. Some prompts do higher sizes fine, some do not. Do not use a lower image size than the recommended one, otherwise image quality will be lower. With higher image sizes anatomy problems occur too often.
Remove fill words and unusual tags in the prompt. They can cause a quality reduction with Big Love Ultra/Hyper as they were not trained. Shorter prompts should work better as it is less likely that bad or untrained tags are in it.
SFW Output - No Tits Please
"Is there a trick so that Big Love doesn't always generate raised shirts or visible breasts?" I get this asked so often, while it is so obvious. Big Love knows what men want so it gives it to them. Negatives don't work that well for it. So you got to speak to it in the positive prompt like you were taking to a nun. "Breasts? How dare you speak such an obscene word to me." Don't mention breasts or any of these sexy parts. Instead describe clothes a lot. Don't write down your impure thoughts. God forbid! Yeah, I know, so hard. Describe her body shape, which also implies certain tits underneath the clothes.
XL Prompting Advice
SDXL prompts & prompts from various SDXL checkpoints work with it. Pony prompts too, but they create a more photorealistic look. Natural language prompts tend to create artistic glamour images with less skin detail while normal tags do more photo-style images. Big Love XL gives individual words more attention, so you don't need to rework prompts like Big Love Pony requires sometimes.
XL1/XL2 has a tendancy towards cuteness and bright skin. If you get too much of it, remove words like cute, sweet, pale/fair skin & flash from the prompt. It also creates realistical analog photos with reduced image quality. Sometimes better to remove terms like analog or grain to increase quality. Decrease cfg to get a more natural look.
XL2.5 can sometimes get too sharp. Reduce cfg or use less steps in img2img to make images softer.
Pony Prompting Advice
Pony prompts & prompts from other checkpoints work with it. Score tags and fill words like masterpiece or perfect skin have as good as no effect usually. A negative prompt is often unnecessary unless you want to make something vanish or avoid underage, but it can also change the style of the image.
Pure booru tags and some other samplers may produce Pony style. Photographic terms or a simple prompts ensure a photo look. Sometimes it refuses to produce medium shots, portraits or close-ups unless you weigh the term heavily or remove feet, shoes, high heels, boots etc. from the prompt. wavy hair sometimes does not look so great.
If it refuses to do porn, try simplifying the prompt. If you want to do a portrait, don't describe surroundings too much. If you want to do porn, describe the action mainly and don't focus on the people or surroundings in the prompt. The largest part of the prompt should be about the main subject of the image.
If a prompt does not produce the intended image, you often only need to modify it a bit to make it work perfectly.
Using Character Loras
Big Love Pony supports Pony characters, but they look like real humans. There were some complaints about them not working, but the problem was not the model itself so far. So here are some tips:
* Try Pony as well as SDXL character loras
* Use a lora weight up to 1.5 if it does not reduce quality
* If it is a well known character, add the name in the prompt and also give it a higher weight if necessary.
* Check your negative prompt for problems. Delete it if necessary and rewrite it from scratch.
* Render a lot of images to be able to pick those with the best resemblance.
More SDXL Photorealism:
You can highly improve the images generated with this model by using img2img, upscaling, detailing etc. The SDXL Lightning lora enhances the look and constrast despite only 8 steps. Alternatively, the SDXL DMD2 lora produces a sharp, but more natural look with less contrast. Also check out the Subtle Style loras, which nicely enhance images.
Big Love License
Big Love SDXL versions are licensed under CreativeML Open RAIL++M (dated July 26, 2023), the Big Love Z-Image and Ernie versions are licensed under Apache 2.0, and the Big Love Klein versions are licensed under the FLUX Non-Commercial License (Black Forest Labs Inc.), with the following additional terms:
For clarity, the following version tiers are defined:
Restricted versions: Ultra, Hyper, Klein, Zia, Zuna, Eryn, Gwen
Open versions: Pony, XL, Photo, Insta, Lust, ZT
1. You may use Big Love without crediting the creator.
2. You may sell the images that it generates.
3. You may only run the official Big Love versions of the SubtleShader account on Civitai and Tensor Art. Running it on other public or shared servers commercially (including other Civitai and Tensor Art accounts) requires a separate license or permission from the creator.
4. You may not sell this model or merges of this model.
5. You may not merge, share merges of, or share LoRA extractions incorporating Restricted versions. You may merge and share merges of Open versions.
6. When sharing the allowed merges of Open versions, as well as finetunes of them, you must apply these same additional license terms.
7. Redistribution of Restricted versions is prohibited. They may only be obtained from official sources designated by the creator.
8. All use-based restrictions of the CreativeML Open RAIL++M license apply to SDXL versions and their derivatives. All use-based restrictions of the FLUX Non-Commercial License apply to Klein versions and their derivatives.
9. Any restrictions in these additional terms may be waived only with explicit written permission from the creator.
TLDR: This is basically the same license as before (just stated more clearly), so nothing changes for the Pony, XL, Photo, Insta, Lust and ZT versions of Big Love. However, restrictions were added for Ultra, Hyper, Klein, Zia, Zuna, Eryn and Gwen versions (no merging, no redistribution). The "Share merges" condition below the license on the right hand side was only deactivated because of the restricted versions. Merging is still allowed for the open versions.
Many thanks to the creators of SDXL, Z, Flux klein, Pony, bigAsp and Lustify for making their wonderful models available. Big thanks to RaymondLuxuryYacht for allowing me to train on his fabulous images.
Description
Based on SDXL. Crazier & more spontanous images, cuter faces, more dynamic poses, better prompt following. Still a bit of Pony in it, so supports Pony loras. Prints text nicely. Less strong texture by default, but you can increase it by adding texture, skin pores to the prompt and upscaling with up to 12 Lightning/DMD2 steps.
FAQ
Comments (201)
Hey man!
Xl version is better with Lightning or without it?
Whats the best method to prompt complex hardcore? Tags or nat language?
Lightning is always better. Tags is easier. Nat language works too.
@SubtleShader What is the difference between XL and Pony version?
@akersamir228 The Pony version works better with Pony-like prompts & is closer to the original Pony model. The XL version is based on SDXL models and contains only 12.5% Pony. The image style is more creative & photographic. Prompting is also different.
Allright, the XL version is officially better, than Lustify. Why? Because, it has total celeb likeness.
I will update.
Update 1: I was too happy early, my first prompt was Selena Gomez:) with perfect result. Less famous ones has no likeness, who works on general all arounders.
Ok, did not try celebs. Good to know. But got a perfectly looking Daenerys when prompting "Dragon rider, Game of Thrones". 😁
@SubtleShader It seems to work in both world, Xl and Pony, but its picky, something works, some not. But very good. It would be Lustify beater with constant good celeb and character likeness.
i have no words for this model, incredible detail great job! I do anime but sometimes i play around with realistic models to see which one is good amongst others, this one is the better trust me, with the right prompts
does pony tags work on the sdxl hybrid model
XL1 does not use a Pony clip & prompting is a bit different. It can do Pony prompts, but you have to see yourself if good enough. But even Pony2 is different to other Pony models.
Since my connection is sucks, i can't really download one by one to test it. So which version is recommended, XL1 or Pony 2. Which is best for using with pony prompting and loras?
You can run all 3 Big Love models on TensorArt with 60 free daily images. See the link in the red text at the top of the page.
XL1 has a more creative photo style while Pony2 looks more classical & supports Pony loras better as well as Pony prompting.
@SubtleShader Thank you, i will try them on tensorart.
Thank you for the SDXL love, we are feeling left out. Excited to try some Pony loras with this one
Feeling left out? 🤔
@SubtleShader Its understandable, because there were new good releases barely, but now its the best time for Xl users. They cant complain, with Lustify and Biglove they can feel, what Pony users feel for a long time.
@SubtleShader everything is either flux or pony or one of its derivatives now
@TomatoAlien Flux is not ready for prime time. Still needs to mature. And Pony still does not produce enough photorealism yet. My Pony2 model does enough realism, but at the expense of prompting. Still, the new XL1 version is more photorealistic. So, SDXL is currently the best there is.
@SubtleShader I agree, pony always was too unrealistic for me. Which is why I appreciate this one.
Thanks for XL1. Had a question on the "Pony prompting" vs "SDXL Prompting". In Pony, we generally use those score tags along with 1girl,1boy etc. How does SDXL prompting differ? I've not used SDXL much.
Big Love Pony2 does not need score tags either. But you can use 1girl, 1boy with Big Love XL1 too. There is no standard SDXL prompting. There are dozens of ways & each SDXL model interprets prompts differently.
@SubtleShader So what would be your recommended prompting method for XL1?
@Melty1989 Natural language prompts tend to do artsy glamour images with XL1. Tags produce more normal photo-like images. Honestly, all kind of prompting styles work. XL1 can even reproduce Juggernaut-like images. Just look at the prompts of images that you llke here & try to imitate it.
@SubtleShader Alright thanks. So, one final question - sometimes with natural language prompting with 2 characters, the clothes I specify on either char tend to get interchanged or both use the same clothes. Is that expected?
@Melty1989 You could try putting the word BREAK between the descriptions of both characters. Should help.
@SubtleShader Thanks, also I noticed just adding "Big Breasted" to the prompt results in the breasts being shown outright. Any way to prevent this?
Edit : Looks like describing it as "A woman with medium breasts", tends to show breasts outright less, but occasionally does trigger it
I am not getting any success with several different Pony Loras, even with strength 1.5, maybe I am prompting wrong?
Likely it is the model. Try Big Love Pony2 with them. Or let me know which ones you tried, so I can test myself.
@SubtleShader I was trying the XL model on InvokeAI. The description says that some Pony loras should work, I am not expecting them to be as good as they would be with a Pony model but they seem to have no affect at all. I am using the XL model at 20 steps DPM++ 2M SDE. Here are some of is the loras I tried: https://civitai.com/models/242629/humantaurs // https://civitai.com/models/43965/photo-comparison-phone-view-beta-v2-xl
@TomatoAlien With XL, that wont work. Thats pure Pony degeneration:) There are full blood realistic pony models(it means no XL in it), try on them.
@damarienezel296 The Humantaur lora starts to work more or less with a lora weight of 3-4. A bit better with Big Love Pony2. But did not get any usable image. Probably needs more renders.
The Photo Comparison lora does not work at all. Too abstract I guess.
You can try to find similar lora that were trained on SDXL.
@SubtleShader thank you for trying. I cannot get pure Pony models to work on my setup for some reason. It's fine, there are alternatives.
@TomatoAlien And how is that possible they dont work? Pony is based on SDXL.
@damarienezel296 Yes, but Pony is a completely new trained model that overwrote the old SDXL base data. Only bit by bit parts of other SDXL models were added in more photorealistic Pony based models. But depends if a lora was trained on Pony V6 or a newer model.
@SubtleShader "I cannot get pure Pony models to work on my setup". What setup should be that?
@damarienezel296 @TomatoAlien Usually clip skip 2 is not set if a Pony model does not work. If you give me more details about the problem, I might be able to be more precise.
@SubtleShader Clip1 and 2 has difference only on 1.5. I never experienced any difference on XL and Pony. So it should be on automatic2. I tried on Cyber (full blood), no difference at all.
@damarienezel296 InvokeAI
@SubtleShader InvokeAI does not have an option for clip skip
@TomatoAlien Then it should work flawlessly, Clip doenst have impact in Xl and Pony.
@suveermarkon910 then the problem isn't with clip skip, it's probably just with Invoke and how it works. Or is user error, maybe both. Either way it's fine.
@TomatoAlien Allright, we all tried to help you, and now, ok its fine:)) I will check Invoke:))
@TomatoAlien Doesnt load? Give error? Grey output?
@suveermarkon910 @SubtleShader really sorry if I came across as angry or sarcastic, I didn't mean to at all. I appreciate all the advice and of course the model.
When I use Pony models in Invoke it gives lots of artifacts and deformed images, like SD 1.5 at 5 steps or something.
When I said Fine I meant it literally, no hard feelings.
@TomatoAlien What sampler you use? What checkpoint did you try?
@TomatoAlien No problem. Never got that impression.
Im happy, its a success. I wrote to you, when i think V1 pony dropped, you should make it XL. They are hungry.
we are haha
Awesome checkpoint.
With lightning lora and 10 to 12 steps and the cfg 1.5-2 (in 2.5 it starts to break) the outputs are pretty much state of the art for SDXL. Need to test further for complete evaluation of the checkpoint.
Hats off.
You can use DMD2, Hyper or Lightning. I still prefer Lightning, because it produces the highest quality fine details and tends towards glamour.
@SubtleShader i've been testin with lightning for hours... And you're right. 12 steps and cfg 2 render superior output. I'll edit the message. Thanks.
I run it on 8 steps and 5 or 6 CFG and it still looks good
@TomatoAlien 9-12 steps is only for img2img upscaling if you want to add more texture or grain. For txt2img you should always use 8 steps.
@SubtleShader cool that's what I'm doing
Yes, it's better than Lustify Endgame. This is now the best checkpoint for SDXL... so far.
The anatomy is a lot stronger and coherent than lustify. I guess it's part of the pony's influence although BigAsp2 might help too in that regard. Mind you! It's far from perfect, this is not flux after all, but its rather outstanding and state of the art for a (not pony) sdxl checkpoint.
Image quality is surprisingly good too.
I think BigLoveXL would be a good base model for future loras.
SDXL is getting giant leaps of quality these last months. Keep this thing rollin'. This train never stops!
Great job all in all. Lookin' forward to BigLove2.
"It's far from perfect, this is not flux after all"
Im late to party. Since when can Flux make hardcore?
@ronaldovihaan383 i was addressing "anatomy-wise". You have to agree that anatomy in general is better on Flux. And no, Flux can't make hardcore.
Thanks. I'm surprised myself how versatile LX1 is. It is basically Lustify Endgame with a lot of other things added on top. It combines 4 different training branches of SDXL. Why it can do many different styles.
Maybe I'll do a Flux version of Big Love, but depends on the effort & if I'm able to improve on existing models.
@ElectricDreams If we want judge Flux anatomy, first we must see in action, right?
Im not talking about hands.
@SubtleShader NOO! But if you can make it to do a proper handjob, Flux users will cum, and you will be king.
another thing I noticed lately is that this model seems not to force subjects to look straight at the viewer, something that is incredibly difficult to avoid on some checkpoints
XL1 might be the best NSFW model around as far as I can tell, at the very least the best XL one (but for me it's more overall)
it's the best XL model period
Extremely good BJs with this prompt:
(your subject woman),
((Woman kneeling between your legs in the cramped airplane bathroom, her lips eagerly wrapped around your shaft, the hum of the plane engines drowning out her soft moans))), ((((thick ropes of cum shooting onto her flushed face, some landing on her tongue as she sticks it out, her fingers gripping your thighs tightly)))),
\nFace focus, her cheeks red with heat, saliva dripping down her chin, her hair slightly tousled from the confined space, the dim fluorescent light accentuating the glistening cum on her lips and skin, ((soft vibration of the airplane adding a subtle thrill, sterile white surroundings creating contrast, slight turbulence making her movements even more intense))),
\nCinematic POV, close-up of her glowing skin and messy, eager expression, the combination of cramped intimacy and the risk of being caught heightening the experience, electric and unforgettable, raw passion confined in a tiny space.
<lora:sdxl_lightning_8step_lora:1.0>
Feel free to post such images. Without adding a name 😁.
My images are all completely smudged.
Everything looks great and then in the last step, the image gets destroyed. I don't understand that. I only have this problem with SDXL. I use Forge, DPM++ SDE, SGM Uniform. I have already tried 3 different VAE. Nothing wants to work. :/ Could this be related to Clip Skip? I no longer have the option to set it in Forge. Please help me :D
there's no clipskip using the xl UI in forge, change to sd UI, or use "all" to use clipskip in sdxl, I've had a similar problem and what helped injitially was upping the resolution, you can try an upscaler with a low denoising value. I also changed to Euler a Automatic which fixed most of the problems with the body, and adetailer for the face, I also have SDXL style selector add-on and have "ENHANCE" selected. you should also try different cfg scale values, go town to 2.5 and all the way to 7.5 and see which works best while giving suitable contrast. Your sampling steps should also be high
Forge sets clip skip automatically, so always right. Must be some other setting that is wrong. No idea at the moment. Maybe reset Forge to defaults
No need for plus VAE, its build in. Try other sampler scheduler combo, and see, if its still shit.
First of all, many thanks to all the commentators. I've found out what the problem is. It is the HiRes fix. The image looks fine without it. As soon as I use HiRes Fix, the image gets destroyed in the last step.
I tried different upscale models, same problem everywhere. I don't know how to fix it, but at least I can generate images without HiRes.
@Halodri Better use img2img for upscaling. Better quality & more versatile.
@SubtleShader I will try it, thank you!
@Halodri Your hires settings has some problem, it shouldnt happen this.
Figured it out! I was using Uniform instead of "SGM Uniform" for schedule type.
@ermbaby Quite a difference with Lightning. Glad you found the problem.
aaaaaaaaaaaaaaaaawesome...
BigLove XL is the best checkpoint ever as of Dec. 17, 2024. It's the best for realism and NSFW among XL checkpoints. It's very good result responsible to the prompts. I am using BigLove XL more than Flux now.
BIGLOVEXL tm is the new Flux!
@ElectricDreams Are there even a single decent flux for adult content that doesn't look terrible?
@tootedandburped119 no.
@ElectricDreams Yeah, it's been pretty disappointing. At this rate, looks like we will be using SDXL forever. At least we have @nutbutter and @coyotte
Impressive! It's amazing that SDXL can produce images this realistic in spite of its terrible VAE.
Can't be that terrible 😆
@SubtleShader Haha, maybe "terrible" was too harsh. I do think it's showing its age compared to the newer 16-channel VAEs, but your model practically closes the gap!
Great model! Many natural woman faces, real people like!
This is an amazing model that has given me incredible results by just simply following the guide that the creator typed up, and I am an amateur!!!
Some really nice images you did there! Unfortunatelly a lot of people don't listen to my recommendations & just use other settings of their own with less good results. But easy to just drag & drop one of my images into the prompt box of A1111/Forge and clicking the ↙️ button to apply the settings. Or drag & dropping it into Comfy.
@SubtleShader I can follow instructions and I do! That is one thing I can do correctly hahaha! I followed your directions to the letter!
@jayt6676 Your newer images are even better!
@SubtleShader Thank you!
I would be coriuous, what would happen, if you merge with an all-arounder, only for celeb, lora compatibility.
With all-rounder, I guess, you mean a popular SFW model? Big Love XL1 is more of an all-rounder than most models with its diverse merge base. But if I would merge with a sfw model, it would not be as easy to do NSFW images. You would have to use stronger prompting and there is a limit on how strong you can prompt.
A bigger problem with these SFW models is that they are not as realistic as models based on bigASP or NatVis. So it would make the merge less photorealistic & push it more towards semi-realism.
Tell me which model is doing celebs better than Big Love & I will have a look. I also don't see a lora compatibility problem apart from loras that were trained on non-photorealistic models.
@SubtleShader I dont think, they are semireal. All-arounders are the top popular realistic models, which are compatible with everything, but uncapable for hardcore. Thatswhy i asked, what would happen. Realism wouldnt hurt, but hardcore capability maybe.
@SubtleShader Sorry, but every all-arounder has better compability, thatswhy i asked.
@sukhmanpatton839 They are definitelly less realistic if they are not based on bigASP or natVis. Big Love can generate everything they can plus more. Lora compatibility is a lower priority for me than NSFW & photorealism.
You can always use another model for txt2img and then use Big Love for upscaling in img2img.
best all rounder with best lora compabillity has been always the cyberrealistic series. It has nsfw capabilities too (at least is a lot better than Juggernaut, Realvis and such...). If i was interested in merging checkpoints that would be my first candidate. Skin quality is pretty good too, that is one of the -few- weak points of BigLove. The OP should try that.
@ElectricDreams Do you mean CyberRealistic Pony or XL? I will improve skin texture in XL2. Tried to optimize XL1 more for pretty & cute faces.
@ElectricDreams I would, but with simple merger, i achive nothing good. I use Forge and supermerger doesnt even work, and i dont even understand, how those weights work.
@SubtleShader If you want serious textures, i recommend this model. It has so sharp texture, you must use the softest sampler, scheduler combo (Euler, SGM uniform) with very low cfg.
https://civitai.com/models/1022696/pony-babel?modelVersionId=1146790
@SubtleShader I was talking about Cyberrealistic XL. I haven't tried the Cyberrealistic ponys... so i can't talk about it. The cyberrealistic series for SD 1.5 are unmatched to this day in image quality and lora fidelity (especially the classic version). On the other hand, The SDXL one is not as good as the 1.5 checkpoints but is what i consider the best SDXL allrounder (i've always found XL iq a little lacking...). EpicRealism is good at details too but it totally lacks NSFW capability... unlike Cyberrealistic XL. But try it for yourself and see if it suits your needs. The strong feats are lora compatibility, image quality and better lightning.
Yeah Biglove is lacking in skin detail (like a lot of checkpoints and that includes Lustify) but its an inherent "sdXL" thing. It sorta comes with the territory I guess. It's like the characteristic "white/bright lightning" of SDXL which distracts me every time.
In any case, going for a better skin detail is a really good call for BigLove2. I wish you the best of luck.
@sukhmanpatton839 examples of the checkpoint seems a bit "overdone" (oversharpening?) and grainy for my taste, but that's on me.
@sukhmanpatton839 Not sure if that is the type of texture that I want. But thanks. Will try more of Tame first.
@ElectricDreams@ElectricDreams Thanks. I was a fan of the SD1.5 version of CyberRealistic but not of the XL version. I had a look at the images & it is not what I'm going after. Lora compatibility is no priority. Big Love does not really need many loras.
@SubtleShader yeah the Xl version is sort of meh compared with the legendary 1.5 checkpoints. I don't know. The problem is a lot of creators are now dedicated to flux and the consolidated "alrounders" for sdxl haven't been updated in a long time (Realvis, Juggernaut). Very few people is giving love to SDXL these days. It doesn't help the creator of Lustify is not interested in making nsfw checkpoints anymore either.
Anyway, i'm sure you'll figure it out.
@SubtleShader Lora and celeb compability must be the first priority for every hardcore makers.
We have good hardcore checkpoints on XL, they are almost perfect(you, Lustify). What is lacking only compability with celebs and loras.
I dont know, what realism you want more, its perfect.
Why do you think, there are gazillion loras? People want unique girls.
@maliqmassimiliano722 yeah it's pretty important. But let @SubtleShader cook.
In any case, i think BigLove SDXL improves significantly with the pony scores
score_9, score_8_up, score_7_up,
and the negatives
score_6, score_5, score_4, worst quality, low quality,
Reminder that BigAsp2 was partially trained with those tags too.
Ok, i uploaded some fun celeb pics, maybe will be deleted. But only the most famous ones work.
What resolution is everyone using with the 8 step lightening lora + upscale? Everything seems pixelated. I'm using the "Highest Quality" workflow that was posted in the main post.
I'm using 832x1216 or 1024x1536. Then upscale 1.5x or sometimes 2x in img2img. Sometimes I don't upscale 1024x1536. After upscaling I manually detail face, breasts and/or pussy by another 2x or more seldomly use ADetailer.
Do you mean the Comfy workflow? Try to reduce steps in the second sampler. Comfy produces too much grain with Lightning. Why I mostly use Forge.
@SubtleShader hi, can I please ask how you manually detail the face, breasts and pussy by another 2x exactly? I don't know how to do that
@idevst Select the image section in Photoshop, copy & paste into Forge img2img, render with 2x upscale till the result looks good, copy & paste back into Photoshop, scale down by 50%, paste back into the selection & edit the layer mask with a brush.
This shit is next pony :o
Small tip: Adding texture, skin pores to the prompt improves images with Big Love XL1. Makes them less soft. I will use it with each new prompt from now on. Also has an effect if only added in img2img when upscaling. Using 10 or 12 steps with Lightning in img2img additionally improved texture & sharpness for soft images.
why am i getting so many artifacts and grain while using dpm ++sde and lightning LORA? Everyone else's images are so sharp
Only use 8 steps in txt2img. Make sure you use the SGM Uniform scheduler. If you have ConfyUi, don't use more than 3 steps in img2img. In A1111/Forge you can do up to 12 steps for XL1 in img2img.
Recommended TensorArt settings
To get great results with Big Love with a free TensorArt account use these settings: https://civitai.com/posts/10998584
As Lightning does not work with a free account (no SGM Uniform scheduler), the idea is to use DMD2 instead. Image quality is quite good. Lets you render 62 free images per day. With 4 steps & lower quality even 124 images.
I updated the render setting in the descriptions above for DMD2. I will likely use DMD2 more than Lightning in future. Reasons: Looks more natural, more coherent bodies, better skin texture, nicer details, renders faster.
DMD2 was designed for cfg 1, which means that negative don't work. Higher cfg is possible by reducing the DMD2 lora weight, which makes negatives usable again, improves backgrounds and increases contrast. Tonemapping and cfg rescaling works too, but looks less natural.
Just tried it and it works well.
So is DMD supposed to be CFG 1 or upto 1.5? Your instructions in the summary say it can be set from 1 to 1.5
@Melty1989 I just updated the instructions to: cfg 1-2 (with lora weight 1-0.75). If you want to use a higher cfg for more contrast & better background, reduce the DMD2 lora weight down to 0.75.
@SubtleShader Thanks, will try it. Also, I'm not certain if you use FaceDetailer in ComfyUI - but I use that and I was wondering if you had an opinion on the Sampler,cfg and steps for that? I noticed using Euler in FaceDetailer really changes the face quite a bit (adding a lot of skin pores for instance).
@Melty1989 I mostly use Forge. I use the same img2img setting for detailing faces. I prefer not to change faces too much.
@SubtleShader Got it, maybe i'll use it for SD strictly. I prefer ComfyUI since it basically can do any model including Hunyuan and LTX-Video.
@Melty1989 I use Comfy too for things not supported by Forge. Merging models is also more convenient with Comfy.
i'm your biggest fan Subtleshader!!!
Wow! Great tip changing the DMD lora strength. Looks amazing now!
I noticed some strange behavior when using a character LoRA and ADetailer in Forge. The faces turn out completely weird and distorted. I can only fix this by adding the LoRA and its keyword to the prompt specifically for ADetailer. This issue doesn't seem to occur with other checkpoints.
Which lora?
@SubtleShader There were several of them, but unfortunately, I don’t remember exactly which ones. It was a couple of days ago—probably related to my Forge installation, some LoRA models, ADetailer cache, or something similar. I’ll try to provide more details if I encounter the issue again.
Thank you for creating such a versatile model, by the way! It’s one of my favorites so far!
Check out the Cuteness lora. Works nicely with Big Love: https://civitai.com/models/1115264/cuteness
Great, even fabulous model indeed. Love many of the basic prompting outputs, however prompt understanding is limited. Any chance of building it on top of Illustrous which (out of SDXL based models) shines in that aspect?
Thanks for the idea. Just merging Illustrious would reduce photorrealism. Would certainly require some training to compensate for that. Certainly not easy to get right. I'm sure, some photorealistic versions of Illustrious will emerge over time.
@ElectricDreams Thanks! Looks good!
Original Illustrious checkpoint only has 33k downloads since September. And I don't see any xxx images done with it. So I don't think it is worth the trouble. I'd rather spend the time on doing XL2 and Flux versions of Big Love.
Hi there, your checkpioint "Big Love Pony2" is amazing!!
But when I’m creating images through the workflow, I tried using the "4x_NMKD-Superscale-SP_178000_G" model for upscaling as you mentioned, but the results turned out really bad, even though I followed the parameters from your tutorial. Could you provide an example of the workflow? The workflow in your tutorial only uses Double Sampling and doesn’t include the Upscale use mode node.
I don't use any upscaler. Just plain img2img produces better quality in most cases. I just mentioned the upscaler because some people prefer to render & upscale in one step with A1111/Forge. The double sampling is just fine. That is plain img2img.
@SubtleShader I’m kind of new to this kind of stuff so apologize for the maybe obvious question: what are you doing in img2img that’s mimicking upscaling?
@Bigreds There is no mimicking. The image is conventionally upscaled (Lanczos) & then rendered again.
@SubtleShader I think I see where I was confused. I'm generating on Tensor.art but I think you're talking in terms of local generation and as good as tensor can be it's not exactly the same options as when generating locally on something like Forge.
@Bigreds Yes, more options locally, which translate into higher quality if you know what you are doing.
@SubtleShader Thank you for you responed pal,I got it! now I'm using DMD2 and everything solved!!
great one again thanks dude
incredible, this model is really great, I have very few anatomical errors compared to other models. The poses are aesthetic and it works wonderfully with a lot of Lora. Thank you thank you! A V2 ? :)
Thanks. Working on a XL2 version, but really hard to improve apart from adding more texture & contrast for certain prompts. Like all bigASP/SDXL based models porn prompting could be better compared to Pony, but not that easy to train. Adding more Pony improves that but also reduces photorealism a lot.
@SubtleShader have you tried with illustrious? it's better with photorrealism (like the cyberrealistic one). IMO It's still "not photorealistic" but the lightning is a lot a better than pony.
@ElectricDreams if I would want to make XL1 less photorealistic, I would add more Pony or Illustrious. The only way to improve it Is to either use an updated version of its merged models (not available yet), add another very photorealistic model (none that I know) or do some training. Maybe will add some of my loras.
@SubtleShader yeah don't be in a rush and take your time. I totally understand it's gonna take -at least- a few months to improve the current checkpoint. Take as much time as you need.
First off, this model has been a game changer. Absolute banger! I have a specific question: has anyone had any luck getting pegging images to work? Was never really my thing, but I’ve had some people asking for them lately and I’m kinda lost.
Just search for pegging above. A lot of loras will show up. Try the ones for SDXL. Pony ones should work too. Even if the example images look anime, Big Love might still produce photorealism with the lora.
I have tried more than 50 different sdxl models, this model is the best, very flexible, a real masterpiece!
Thanks. Very kind of you! And thanks for the buzz!
very good. thanks alot
hey is the VAE already baked in??
Yes. It is in as good as all SDXL models. Otherwise would need some extra steps in A1111/Forge that would confuse most users.
@SubtleShader thank you
I've been testing this for a while and I'm very impressed. The results from this merge are better than any of the core checkpoints on their own.
I find that XL1 version to be the better of the two.
The pony model lost a lot of it's "ponyness" as it gained realism, and cannot do some of the crazier stuff pony typically excels at, so it's basically just a less realistic version of XL1 in most cases. (it can do a few things XL cant)
I tested with Lightning and DMD2. Lightning 8-step is better, but I get the best results running the without these enhancements at all.
I get the best results (realism and detail) at more traditional settings:
DPM++ 2M SDE, Karras, 30 steps, CFG 4-5
NMKD Superscale at 1.5 upscale, 3 hires steps, denoise around 3 - 3.5
Getting fantastic results.
I was having a bit of an issue with the pictures being zoomed out to far. This model does an exceptionally good job at a distance (Upscale and ADetailer required), but it was over-doing wide when I specified medium full or half body shots.
Adding 50mm to the positives and wide angle, full body shot to the negatives fixed that.
"The whole is bigger than the sum of it's parts" truly applies here
I use dmd2 and its very good, but now hires fix gives enormous shit result. Whatever sampler or scheduler, steps, cgf i use, everything is liked an overcooked shit, the girls look like they were sunbathing 5 hours in 60 C degrees.
I recommend using img2img instead oh hires fix. Better quality & you can produce multiple variations with a seed. If you want to use hires fix, use 4 hires fix steps or less. This avoids overcooking, but also means lower quality. In img2img you can use as many steps as you like. I use 12 steps usually resulting in sharper images with much higher quality.
@SubtleShader Thanx
@SubtleShader 12 steps with with or without the dmd2 lora?
@Hieheihei 12 steps without DMD2 would be too soft.
This is the best SDXL model on Civit. Any chance for a sequel to the XL model?
Check the LoRAs from the same creator as this checkpoint. They are very good
Did not manage to make XL1 substancially better yet. Sorry.
don't know why I held off downloading this. It's likely the best current NSFW model. Fantastic.
Why the fuck is this not on Civit on-site generator yet? It's clearly better than any other Pony realism model around. What does it take?
Someone told me they have to keep the amount of on-site generated NSFW images down to avoid repercussions. But you can use TensorArt to render images with it. Go here: https://tensor.art/models/805531042304810793?source_id=nzu0r1_nkEa0o_QubHvz8BUm
Ask them. Civitai will eventually start to become irrelevant if they don't start to allow the use of popular checkpoints like big lust and big love xl.
@steegot456 Agreed, how do we go about asking them? Seems that voting and downloads should be a clear indicator for them.
@ponyboi They did not even add Lustify to the on-site generator, although it has almost 100k downloads. Asking won't help.
They dont want realistic hardcore models on Civgen.
They recently said they’re working on the hardware to make all checkpoints available, but wouldn’t hold my breath. Also, to be fair, you kinda need options and resources not available on civit to make this shine. Like you can’t do LCM exponential on civit for instance
@DrAIvyn Yep, the Civitai on-site generator is very limited. I like TensortArt better. But best is Forge locally.
Amazing model but what do I have to prompt to get a blowjob? I've tried "fellatio, blowjob, cock in mouth, sucking on penis... But all I can get is a chick gripping penis with hand and sticking her tongue out
Yep, can sometimes be difficult. Maybe I'll do a lora at some point. This prompt seems to work rather reliable:
1girl, hetero, 1boy, oral, fellatio, penis
You can try to add other things to it.
@SubtleShader yeah I looked for POV bj LoRAs but most were for anime and not realism the one I tried made the weirdest long curved penis I've ever generated lol.
@MyaSturbate Anime loras can also work & nevertheless produce photorealism with a photorealistic checkpoint.
Hi Subtleshader!
Dont you know an ultrarealistic photoshop workflow?:)
I use Cyberrealistic Pony because of compability with loras, not your model. I would use it of course, but i need those lora characters.
I use forge and refiner, highres, adetalier are just not good or not for realism.
You can try the Pony loras with Big Love XL1. If they don't work, try Big Love Pony2. img2img is best for upscaling photorealistic images in Forge. After upscaling I sometimes also upscale face, breasts & pussy separately by 2x, scale down by 50% in Photoshop and add them back to the image. Works much better than ADetailer.
@SubtleShader But how it will be realistic with only upscaling? So no automated workflow for Photoshop.
Too much hassle for fap material, i think. Thanx for your answer.
@lorenzitay956 You can use Big Love during upscaling. You don't need the loras for upscaling.
@SubtleShader Thanx, but i search for total other things. I need an automated workflow, which gives realism.
I am browsing now free photoshop actions(workflows), but they are mostly artsy, vintage etc effects. There should be a realistic workflow for AI generation:) Its fast and easy to use it in photoshop, but somebody has to create it first of course, and i have no idea how to do it.
@lorenzitay956 If you want automation, use ComfyUI, not Photoshop.
@SubtleShader Its not comfy for me, but photoshop is.
this one, is just simply amazing.
One question is there a prompt to add cum to image (like on boobs and such)
You can simply use cum or cum on breasts
This model is amazing, you did a great job. Will you make an inpainting version for it?
There is no inpainting version necessary. You can also use it for inpainting.
@SubtleShader Do you have any suggestions for inpainting settings?
@LannaPanna Same settings. Higher denoise.
Thanks for making this, Love it so far! One of the best ever
I almost baited by dmd2 lora. First thought was, how realistic. So i managed to make a setup, which was good.
I could bear with detail loss on eyes, hair, facial expressions, because, there are general detailers for them. But, what the fuck is going on with clothes?
I always prompt skintight, shiny clothes: swimsuits, leggings, yoga pants.
My usual prompt is sexy shiny skintitght swimsuit, and first i didnt notice, that they were loose and not so shiny.
I switched to yoga pants, leggings and i was shocked. From shiny, skintight leggings, i got matte, loose casual pants.
I know, that low cfg is is low cfg, but when sex poses works, why clothes are this much shit?
So its like the other shit ones, lcm, hyper, turbo, lightning. Too much loss of details to worth it.
Sorry for the long comment.
Dont you plan other version?
You can reduce the DMD2 strength & set cfg to 1.5 to 2. But sometimes makes the image softer.
Yes, there will certainly be a XL2 at some point.
I used it with euler a, but with lcm is the same(higher cfg).
Next model will be SDXL or Pony or maybe Illustrious?
@neerronen757 Euler a is not so great for DMD2. You need to use the LCM sampler with Exponential scheduler. Another option is to use DMD2 in txt2img and to not use DMD2 in img2img for upscaling.
I won't do less photorealistic with my next checkpoint, so no Pony or Illustrious. Pony is already a part of XL1. Illustrious is not photorealistic enough yet to be included.
So Big Love XL2 will be a bit more photorealistic, but as XL1 is already quite good, it is hard to do better.
Thanks for this model, it's impressive!! Can we expect a Flux version or would that be dreaming too much? Please!!! :)
Yes, I will try to do a Flux merge at some point. Flux is likely not as matured concerning NSFW as SDXL, I guess.
@SubtleShader This is going to be tricky, flux is idiotic in regards to NSFW.
Что вы цепляетесь за Flux, громоздкая не кому не нужная ерунда
Details
Files
bigLove_xl1.safetensors
Mirrors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bigLove_xl1.safetensors
bgL_xl1.safetensors
bigLove_xl1.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.