Big Love Klein3 is also offered for $10 on Tensor Art. You can 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! A subscription includes the latest release and new ones within 30 days. Any problem? Contact me via chat.
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 Klein is a fine-tune of Flux 2 klein. It can do txt2img and also edit images with prompts. It only needs 4 steps, so is quite fast. Check the "About this version" box on the right for info on the individual variants. Some are up to 3x faster.
It is recommended to use it with ComfyUI. 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.
Settings are 4 steps (2 steps for upscaling), cfg1, Euler sampler, Beta scheduler, 832x1216, 1024x1536, 1280x1920 or 1536x2240 pixels
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. Check the "About this version" box on the right for info on the individual variants.
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.
For training a lora read here for Gwen1.
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 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, the Big Love Z-Image, Ernie & Qwen versions are licensed under Apache 2.0, and the Big Love Klein versions are licensed under the FLUX Non-Commercial License 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, Qwen, 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
Please note: The high-res Ultra versions of Big Love are pay-only and special license conditions apply. See description!
Big Love Ultra2 produces up to 80% more usable images than Ultra1. Much higher quality at 1360x1984 and caption-style prompts work more effectively. In fact, using caption-style prompts (e.g. blonde woman sitting with spread legs on wodden floor) produces more usable images than tags (e.g. woman, blonde, sitting, spread legs, wodden floor). Ultra2 was trained on 2200 hi-res images.
FAQ
Comments (92)
Hello, your ULtra1 version file has disappeared!
It says under About this version:
"Please note: The high-res Ultra versions of Big Love are pay-only and special license conditions apply. See description!"
Okay, I went over to TensorArt and picked up Photo4 and Ultra1... OMG, so good!! Well done! Having a blast playing around with them. Thanks!
Can male characters be generated?
Yes, no problem. You can also generate male & female in one image. 2 males too.
So we can't download ultra1 even early access time is up ? That is shame.
Yes, Ultra is a lot more work (because of hi-res images) and training costs more. Will also take a lot more work in future because as good as noone is training at 1536 resolution. 2048 resolution will be even more efforts & costs.
@SubtleShader But there is no option to buy it again, that is why x2 shame if you know what i mean. If i knew, you won't let download, i would take it with buzz.
@Reelai Read right below the showcase images. There are 2 options. Either on TensorArt or by tipping buzz on my profile.
So if I understand correctly (personally I don't care, I don't download this type of models)
BUT you make people pay in Buzz so that they can help the community unlock your model and once you have your little "buzz" you lock the model to really screw them over?
But what type of creator are you in reality? don't you have any shame?
Civitai should ban you for such practices!
I'm really sorry for the people who contributed to pay for in the end nothing at all.
The poor ones. Frankly at this stage, don't trust him, such practices suck.
I asked Civitai if it is OK. They said yes. Actually they added the "On-site Generation only" option for such occasions. No one is forced to buy anything.The people who do appreciate my work.They get the model in return. No one is screwed here. I wrote beforehand that it would not be free. There are countless other checkpoints and loras that I offer for free.
Training these models costs money. There are countless others models to choose if this one doesn't suit you.
I have no interest in generating on civitai.com. unfortunate nothing for local gen.
What? Ultra1 is no downloads after the preview period? WTF?
If you could read, it was day 1 in the description, it wouldnt be dowloadable here.
@hendryxward516 Nothing like that was written in the description. Just the standard text of the Early Access feature that I cannot influence. I wrote below it that Ultra1 was pay-only & won't be available for free.
Before all of that I asked Civitai if switching to generation-only after Early Acces war fine. They said it was ok. Otherwise I would not have started Early Access, activated "Generation-only" from the start and would have given payment instructions.
Native high resolution is the only option left to squeeze out everything, what is in sdxl. I think about better eyes and more nuanced faces. But only, if there is no body horror(long torso mostly)
@SubtleShader You wrote in the about section, it wont be downloadable here.
@hendryxward516 Ultra1 works fine with a lot of prompts. Still not the level of the low-res Photo versions, but it wiill be improved in upcoming versions with more training. It already compensates it with better details & creativity. Improving Ultra is much more work because I more or less have to train from scratch, it requires much more gpu power at higher resolution and it needs higher resolution images, which are not as easy to curate as low-res ones.
@SubtleShader Minimum 2000p on both sides?
@hendryxward516 Ultra1 is trained at 1536 resolution which is 1536x1536 or appox 1248x1824, so that is what it can do. 1360x1984 works less well so far, but will be much better in the next version. 1536x2240 will take more time. Certain prompts do better at higher image sizes than others.
At a later point I plan to train 2048 resolution, which will need even more work and gpu power. End goal is to be able to generate 2048x3072 in one pass in txt2img.
@SubtleShader 2048x should be enough for everything, but newer illustrious also trained on higher resolution and the result was not good(at least on a finetune, what i tried). Long torso was the minimum and sometimes horror faces.
@hendryxward516 Ultra1 does fine at 1280x1872. Works better with caption-style prompts though. SImple one or two word tags tend to produce more failures.
Jmo, but I've looked through both the Ultra1 and Photo4 pages several times during early access and never once noticed anywhere that they were going to be pay-only and was actually surprised when I saw it was on-site only. I'll often tip a little buzz for some models in early access, hoping they might get released early without reading/re-reading a wall of text for a model I've already used in the past, so I can see how some people might get upset that they chipped in for something without realizing they'd never get it without having to pay whatever the full early access fee was anyway.
I personally don't have a problem at all with creators deciding to sell their models for $5 or $10 if that's what they need to do. I'm not going to take a chance buying a model revision I might not like, but that's 100% the creators' prerogative. I just don't understand the purpose of offering early access for a model that is going to be pay-only. Just make it pay-only/on-site-only immediately. With CivitAI's tipping during early access system, it's borderline deceptive imo (not that you intended it to be). Maybe there's a way to disable tipping for early access? That way people could either buy it or tip directly to your account without mistakenly thinking they're helping a free model become available sooner.
Again, just my personal view point on the matter. Noticed a few other "special release" versions being sold/advertised here on CivitAI and it just leaves a bad taste in my mouth. I enjoy testing out models for my own hobby/amusement and not for profit, so outside of hardware and creating models for myself, there's nothing else I feel the need to spend money on. If others do, that's absolutely fine too. I'd just hate to see everything new/different at CivitAI eventually be pay-only, rather than artists deciding to either share or not share whatever they've created. Still appreciate your work, take care! <3
@AFD_0 I never had the tipping feature during early access enabled. I never ever used it. So no none tipped anything. All people who gave buzz received the checkpoint. No deception. There is no reason to get upset other than wanting to get something for free that was never meant to be free.
@SubtleShader In that case, yeah, there would be nothing deceptive going on, whether accidentally or intentionally. Although, I was still 100% expecting Ultra1 to be free and waited patiently for 2 weeks (or whatever) to try it out, so I'm gonna need to be reimbursed for my disappointment! ;D
@AFD_0 People just don't read the "About This Version" box. Check your Civitai chat for my message!
@SubtleShader I always read the "about this version" right before I start the download, usually making brief notes in the filename for recommended parameters, etc. Just an FYI, but Photo4's "about" still doesn't mention anything about that yet like Ultra1 does, though my eyes are now finally seeing the bright yellow text about it in the main description that I never noticed during early access. 8)
@AFD_0 Photo4 does not mention anything because it will be freely downloadable after EA
@SubtleShader My misunderstanding, sorry!
Even after all this discussion, and reading and re-reading the descriptions, I was still under the recent impression that both were early access and then switching to paid/on-site only. I do see where CivitAI states "This Checkpoint will be available for free in 6 days.." but thought that's what Ultra1 had said as well. Too many places to read different things, I guess. Swear that I completed all my English classes in college, but maybe I'm just getting senile now, lol.
@SubtleShader Thank you for your community contributions, but who in their right mind would pay for porn in 2025? I guess there are some willing whales with open wallets (as with any other addiction). Best of luck man!
@pokegirl003861 Porn is a $100 billion industry. Onlyf@ns is another $20-30 billion. People do pay for porn, especially for fetish or Girlfriend experience etc.
@Rapidfire_suk @pokegirl003861 Not to forget that Instagram is the biggest porn producer of all 😆 Surgical enhanced body porn.
He very clearly stated the Ultra models will not be available to buy with byzz after and only for 10K buzz or $5 on tensor during the Early Access period. People just don't read descriptions.
How does your img2img work in regards to the workflow, i uploaded the img2img workflow in comfy and the photo remains the same, i want to upload a photo and then n**ify it, do i need to input further prompts to get desired results? is there some setting i need to tweak be it the lora, kspampler etc?
The denoise value deternines how much the image remains or changes. For upscaling 0.3 works best. Higher values rework the image. But you are looking for inpainting. That lets you paint a mask for only changing the body, not the whole image. Google for such a workflow.
@SubtleShader It's unclear why I need to load an image into this img2img process? What does loading an image and providing a text hint accomplish if the image doesn't change?
@bertos2013623 img2img means you start with a image and modify it. I use it mainly for upscaling, but you can also use it to make an anime image photorealistic or vice versa. txt2img starts with a prompt and you receive an image.
@SubtleShader so is that to say modification can only be done using inpainting and that one cannot simply utilize denoise or any other feature to change the image or at least the clothed parts to become nude, and what if i want to bring out the many positions while uploading a image of my own? what the process then there?
@someonestole54105 Just try it and see yourself. Just saying that if you only want to change certain image parts, a mask is most effective. Inpainting offers a mask.
@SubtleShader ive been trying img2img but im not getting any changes, is there defacto settings i need to put in place to get changes, in the meantime i will try inpainting
@someonestole54105 The "Denoise" setting in img2img is what changes the image.
At 1, the model starts with the starting image, but everything is reworked. At 0, nothing changes at all. Values in between offer points in between. The model sometimes gets stuck on near identical images despite 1 Denoise. To break this behaviour, make small or big changes to the prompt, or switch to another starting image.
As SubtleShader said, you can use img2img at low denoise for upscaling.
(One scenario where you would have zero change every time is if your are in img2img inpaint, but forgot to apply a mask, or thought you were in regular img2img.
Inpaint works by only modifying what you have masked. The rest of the image is untouched)
Big Love Z
I plan to release a Big Love version of Z-Image-Turbo called ZT1 within the next days. Very likely on a new project page. I'm still training it on a subset of the Big Love dataset for better faces, nipples, pussies and skin texture.
Z-Image-Turbo is not ready for all things NSFW yet (still needs a lot more training), so don't expect a miracle. But I already saw some very nice results.
Any chance for Qwen?
@BuffaloSymphony Too big. I don't like these monster models.
@SubtleShader Yeah it's a beast.
@BuffaloSymphony finetuning qwen would be around 200k.
Could I suggest less images on tan lines? Old models have bikini tan lines a lot. Great models though
@BlessedBooba My Big Love training set has few tan lines. In the SDXL version of Big Love they mostly come from the models I merged before I started training.
It might pay to wait for the base model.
Fair warning to all who consider purchasing the Ultra2 model, it is completely incapable of producing decent results without the use of the dmd2 lora. For a model priced at 10k, with no mention of this in the description, I think this is unacceptable, therefore I cannot in good faith recommend purchasing this model. With all this being said, I will note that when using the dmd2 lora, the results are quite good, sometimes even excellent.
I actually mention in the description that Big Love is optimized for DMD2. I would not recommend using it without DMD2 (or at least Lightning). Gives you far superior results than using it without.
Quote from the description: "Pony1/Pony2/XL1 versions were optimized for the Lightning lora while the newer versions were optimized for the DMD2 lora. Both loras speed up rendering by a factor of 3-6x and let Big Love produce more photorealistic results." I now added " It is not recommended to use Big Love without DMD2 or at least Lightning. The results will be not be as good.". I hope that satisfies you.
Thanks a lot for warning about Big Love's exellence. I appreciate it.
Fair Warning... RUN TO BUY THIS!!! I love SS's models, I've been a fan since BigLove Pony2... These DMD versions are fantastic. $5 on TensorArt, well worth it! Thank you for all the great work!! I posted a couple images... Man, I fall in love with these girls every day!! :-D
Thanks for the even nicer warning. 😆
@SubtleShader The reality is that, this model is GREAT, it's like a wild horse a the first but very Great after a little warm up. ^^
Hi there,
Can we generate all kinds of woman and girls ? I mean most of the image either made by you or other users mostly are petite, young for Lust1. so I was wondering if this model just focus on that type of girls ? or no you can generate any kind of color, race, age, body. but users just stick to that types.
including some general images would be very good (either NSFW or SFW).
Yes, you can generate any age, color, race, body type, average face, beautiful face, ugly face, amateur, pro, SFW, NSFW or whatever. If you look through all Big Love versions you will find a lot of variety. Have fun prompting or grab a prompt and modify it.
Perfect, I just bought ultra 2 on tensorart, can you share a workflow optimized for it, the previous didn't work very well with it
It is always in the description
@SubtleShader But i got many images thats comes out blurry and bad quality with the workflow in description
@alexandrebrice33257 If you did not change the settings, I would suggest changing the prompt. You can also start with one of the prompts here any modify it got your needs.
Hi, I really need any good WF for ComfyUI for ULTRA2 model i just bought. I need for BEST Quality and second WF for fast. Can any one can, please ?
I use stable-diffusion-webui-reForge, which works with SD1.5 and SDXL models. You have to read the model instructions carefully to make it work.
Check the Big Love description. There is one there!
@SubtleShader feel like the workflow works well with ultra 1 but not ultra 2
@alexandrebrice33257 Ultra2 is normally more stable than Ultra1, but some Ultra1 prompts might not work as good with Ultra2 & need modifucation.
@SubtleShader How do you usually find your prompts? Do you just keep typing different versions until one works, or do you have another method?
@alexandrebrice33257 I mostly use wildcards. Or wildcards of existing prompts. I have dozens of them. I adjust prompts manually if the image is interesting but lacks somethings.
Could someone help me with the lora training config o tensorart?
Photo 4 is really good. Thanks again for a superb checkpoint.
Ultra2 >> Can we have some samples made without DMD2 4 step lora?
Where is the LORA and where is this checkpoint alone?
Always check the description. Everything is there. Get the checkpoint alone with the Download button at the top. DMD2 lora here: https://huggingface.co/tianweiy/DMD2/blob/main/dmd2_sdxl_4step_lora_fp16.safetensors
Big Love is optimized for DMD2. I don't recommend using it without it.
@SubtleShader So, this checkpoint is only usable with DMD2.. Is that what you mean?
@SubtleShader seems like you miss-read my comment. I asked about how can I see your samples that was made without DMD2 LORA to see the effect of this checkpoint without LORA. I don't want to use DMD2 LORA as it brings limitation. It is useful to render skin and genitalia of female sluts, but I need other concepts.
@dioxidin i meant what I said. I recommend using DMD2. Too annoying to not use it. Anatomy is worse with standard SDX, results are less photorealistic and takes 3-6x longer to generate. Have fun doing that yourself. Big Love is also trained on sexy females mostly. Better use another checkpiing if you want to generate something else.
Thanks for making it free.
Appreciate the hard work!
Thanks for your work! Prompt adherence is great! 👏
I am getting blurry/poorly detailed images using photo4 + DMD2 lora with LCM+Karras, CFG1, 8-12 steps settings, do you know why? 🫤 Images generated with similar settings using other DMD models (Gonzalomo, Pikon or IntoRealism) are much much clearer and detailed.
Cheers!
Use 8 steps & Exponential. Big Love has its own prompting style that is different from other models. Often removing surplus stuff from the prompt improves image quality a lot. There is a lot of trash in prompts that Big Love does not like.
@SubtleShader CAN YOU GIVE SOME EXAMPLES OF USELESS WORDS? thanks!
@JetterJack Here you go:
masterpiece, best quality, high quality, ultra detailed, highly detailed, intricate, intricate details, 8k, 4k, UHD, RAW, sharp focus, award winning, professional, perfect, flawless, stunning, magnificent, epic, majestic, breathtaking, incredible, amazing, fantastic, wonderful, excellent, superior, premium, top quality, high resolution, high-resolution photograph, high detail, elegant, pristine, vivid colors, vibrant colors, ethereal glow, atmosphere, vibe, mood, concept art, digital painting, digital art, trending on artstation, artstation, unreal engine, score_9, score_8
Exponential
on photo 4 how to get full nelson working. Can you show a prompt example.
I was so busy with Ultra2 that I forgot all about Photo4. I gotta say, each one of the Photo models has been a notable improvement over the last. Photo4 is fantastic.
Thanks. Photo4 is more consistent. Ultra2 has amazing details.
@SubtleShader Just like JibMixRealistic 17 and 18 and those are free. And fp32. Have a nice work guys! :-D
@bellissima3d JibMix cannot do the xxx stuff that Big Love does. It is also rather semi-realistic. Different niche.
Love it! Any tips on how to create gymnastic sex positions with Photo 4? I tried gymnastic straddle V-Split etc. but when in a sex scene (man carries woman in straddle split) that works like almost never. Also the limbs often get mixed up, and adding a lot of loras to fix it deludes the quality sometimes. And well.. control net is a mess with 2 people as well (but maybe Im not pro enough with it) Hints appreciated :)
That would most likely require specific training images.
The model cannot generalize effectively what it has never seen. So at present, you would need to generate separate pictures, make a crude collage in GIMP or Photoshop, and then use that collage in img2img. You would need to construct a detailed text prompt describing your envisioned scenario and then try to find the right level of Denoise and the right Seed to make something.
If the bodies are positioned roughly right you should be able to mask off the genital area and describe penetration in the prompt. It is a lot of work, and no guarantee of success. You could also find an existing picture that already has the type of scenario you have in your head, use taggui with JoyCaption to process it, creating a detailed prompt, then you can apply that image in some variation of ControlNet (Canny, LineArt, Scribble, Soft Edge) at sufficient strength and manipulate the prompt to match the way you want your characters to look.
All of this is fiddly, but whenever you try to make the model do something it has no concept of, you need to apply these kinds of measures and have a LOT of patience, creativity and will to experiment.
Yet another possible strategy is to gather pics (high resolution, good quality) that depict the position and train a LORA for that concept. You could use it in tandem with the other methods and you should be able to get decent results. But to get a picture of the same high quality as the checkpoint is unlikely.
Big Love is not trained on these positions. Would take 30-50 good images to train it. I doubt so many are easy to find.
@SubtleShader Yeah I agree, maybe a mix is needed of sfw and then sort of getting all the way. Somethimes it even works by accident but lol
You know the idea for AI is that you can craft things that are hard to do in RL or achieve, at least that's what I try to do.
Btw love your models, theyre the best.
@pornthulhu Yes, sometimes it works because the checkpoint saw a few of these images, but takes a lot of trial & error. Only works more reliable if it saw 50+ images 50 times, so was trained 2500-5000 times on such an image.