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
Big Love ZT3 offers finer details, more sharpness, improved anatomy and more creativity. A de-turbo version of Big Love ZT3 is provided for training (instructions here).
FAQ
Comments (106)
Has the gguf version of Klein been released yet?
It is the first file on the list where it says "Pruned Model nf4 (6.04 GB) ... GGUF". It is a Q5_K_M GGUF.
Sorry, but Civitai offers no GGUF terminology why you have to choose something like nf4.The "Pruned/Full" stuff is also a leftover from SD1.5 times.
Question: I bought the klien version and im wondering if I'll still have access to any updates if available
I usually don't do minor updates of checkpoints. Make a backup of your download. You can contact me via Civitai chat if you need something. Always happy to help.
you will have to pay again if they release a v2. not sure why this guy can't give straight answers.
Civitai's Early Access is separate for each new release. You do not automatically get a new Early Access version if you bought an old one. So yes, Klein2 will require extra pay. If you use my subscription option on Whop, you get all new releases within a month. Civitai does not offer subscriptions.
@SubtleShader and here we are... just about a month later and the version i bought is obsolete... don't know if I like that
First of, thanks for making these great models, they really generate beautiful images. I really like the lust1 version and was wondering if you had any plans of creating/releasing a lust2 version (merge with Photo 5, for instance). Also, since Forge Neo now supports Klein, do you still advice against using the Klein version in Forge?
Lust1 was not that popular, so no version 2. Feel free to do your own merge.
I have to look at Klein1 in Forge Neo again. It rendered a lot faster than Comfy, so I suspect that it uses a shortcut that reduces quality. Hopefully a update will fix it. I prefer Forge over Comfy, but if there is such a big quality difference, I have no other choice.
@SubtleShader Thanks for the quick reply and info. Too bad there will be no lust2, but maybe one day I'll try to merge myself when I learn about merging models. I also prefer Forge to ComfyUI (less of a headache, lol), but I indeed see a quality difference between the two. Maybe I should try to find a good and easy/simple workflow for ComfyUI that is as user friendly as Forge Neo (one that includes an ADetailer like face fixer, upscaler and lora nodes).
@Guidebo Check out https://github.com/mcmonkeyprojects/SwarmUI Never tried it, but puts a UI on top of Comfy.
@SubtleShader Hey, thanks a lot for that info. I've seen SwarmUI being mentioned a few times, maybe I'll give it a shot. Should you also happen to know about a good CofyUI workflow like I mentioned before, I'd be interested. Thanks again for going above and beyond, you might be the most helpful creator I know 👍
@Guidebo Check out the Comfy workflows in the Big Love description. Klein1 workflows are embedded in the showcase images.
Is there something I’m missing on Klein 1? I paid 1000buzz to use it on Civitai, the unstable resources flag went away but I’m still not able to get anything to generate.
I will send you the 1k buzz back (and the 1k of the other people back too). It looks like Civitai does not manage to make finetunes of some models like klein & Qwen work on the generator. I did not know that otherwise I would not have wasted my own buzz to get Klein1 there.
@SubtleShader That’s too bad, looks like a great model. Thanks for being willing to reimburse
@FooDingus00 I now refunded the 1k buzz to all who paid for Klein1 generator access.
Plastic Barbie world
(ZIT3)
That is what Z is. Some people like it. Less plastic than Pony and Pony is/was a mass favorite.
Why don't you make some decent art first? :-) Or make a checkpoint yourself.
@LuckyJack1408 Yep, easy to criticize, hard to create. Often the most critical people are the least creative. So they have to put other people down to feel better themselves. So obvious to others, but not to themselves.
@SubtleShader I see this all the time here; when you go to the site there's hardly anything there, or it's just junk. I'm currently using ZT3, and it rocks!
@LuckyJack1408 Wow. Love your images! Let the barbies rock! 🤣
@SubtleShader 🤣😋
@LuckyJack1408 You don't have to be a cook to realize that the food isn't delicious.
@SubtleShader I didn't mean to humiliate you at all. I wrote my opinion, and that's what comments are for, isn't it?
@210881175 Thanks for clarifying. Much appreciated. Too many haters commenting lately. No time for that shit. I'm too busy improving checkpoints. I started blocking disrespectful people and cyberstalkers. Sorry, if I'm or was too harsh. Easy to misread Intentions with only a few words. Will hopefully improve soon with trouble makers blocked.
@SubtleShader @luckyjack1408 I LOVE BARBIE!! :-D
Hello, Does it work with the character's lora?
If you train a character lora on ZT3, it works with it. I'm still uploading the training version of ZT3. Will be available later today.
@SubtleShader is it available?
@manukillthequeen446 Civitai took Big Love offline when I uploaded it. They said they will check it out and adjust their security system. Till then you can write me to get it.
Big Love ZT3 Lora Training
If you want to use your lora with Big Love ZT3, don't train on Z (Base/Turbo) or another Z checkpoint. Train on Big Love ZT3. Z is not as flexible as SDXL in this regard.
OneTrainer Instructions
Download the BigLoveZT3_deturbo.safetensors version of Big Love ZT3 to train a lora on Big Love ZT3 with One-Trainer. Choose the z-Image DeTurbo LoRA preset in OneTrainer. On the model tab click the ... button next to Override Transformer / GGUF and choose the downloaded BigLoveZT3_deturbo.safetensors. Add your concept and click Start Training.
Please note: OneTrainer has 512 resolution set on the training tab. If you have 24 GB VRAM, you may want to set resolution to 1024. If you want to train 1024 resolution with 16 GB VRAM, use the z-Image DeTurbo LoRA 8 GB preset and set resolution to 1024. In both cases it will take longer to train.
AI-Toolkit Instructions
Choose the Z-Image De-Turbo (De-Distilled) option in AI-Toolkit. Then in the "Name or Path" text box replace ostris/Z-Image-De-Turbo with SubtleShader/Big-Love-ZT3-De-Turbo-Diffusers
General formula for steps:
Image count * 100 epochs / batch size
So, 25 images * 100 epochs / batch size 2 = 1250 steps. Save every 10 epochs (every 125 steps here) and compare the last few saved versions.
With a batch size above 1 calculate a new learning rate as
lr = base_lr * sqrt(batch size)
For example: 0.0001 * sqrt(batch size 2) = 0.00014
Please note: Do not activate Cache Text Embeddings if you want to use a batch size above 1. Otherwise it will error out. Deactivate sampling as it takes too long & does not show you how it will look with ZT3.
hello, First off z3 is amazing, somebody really knows their art. Just wondering as far as the lora training for z3...is it possible that you can make a video tutorial whenever you have free time. More of a visual learner. If so, thanks a lot in advance
@yujirofitt Thanks. Watch https://youtu.be/Kmve1_jiDpQ?si=cORMQ7kZO-xr1oaD
Do what Ostris says in the video except for what I mentioned above.
I've been waiting EAGERLY for this update! Thank you my friend! I'm DLing the files now and will play around with it tonight!! :-D
@virustotal See my chat message
Anyone know if this can be done with OneTrainer. I tried this training concept before but it never worked.
edit: nevermind. it works now with OneTrainer.
@Kierkegaard420 Thanks for confirming!
Just want to confirm. Downloaded the new ZT3 model, trained locally on a 3060 with the instructions above (took about 4 hours with 37 images) , so far it is one of the best ZIT + character generation tools I have used.
trained a lora and got nice results, definitely an improvement over zt2.
@virustotal Nice to hear! De-turbo is certainly better than using the adapter. It is similar to base. Unfortunatelly not possible to convert turbo to base AFAIK, so de-turbo is the closest.
@Kierkegaard420 Can i ask you, what did you put in the "Override Transformer/GGUF" textbox? leave it blank? What did you put in the VAE override? Also blank? When you selected the Z-Image Turbo config were there any other changes you had to make in order for it to work(not talking about personal choices like optimizers/steps)? Thanks!
Even better, it would be amazing if you shared your config because I keep getting an error when I try to run it.
@Tezozomoctli I don't remember any override options. I use the default UI settings and only make a few changes. What exactly does.not work? Copy and paste the error into a LLM and it will usually tell you what the problem is. My config is optimized for 32 GB VRAM. Not sure if helpful.
Hi ya, I bought the model off of Civit, so it's just the safetensors file, not a zip file, How do I get that to work with AIToolkit please? Or where can I get ZT3_deturbo_diffusers folder? Thanks.
@Dynelements Download "Full Model bf16 (9.14 GB)" above. It is a zip file that you need to unzip.
Is training on Big Love Z for your own lora a recommendation you would apply to Klein 9B version as well?
@brnfd24434343d Yep, general stuff. For SDXL too. Klein 9B reacts to training similar to SDXL.
@SubtleShader Brilliant thanks, not sure how I missed that.. But, it now just says 'unauthorised' when I try and download it.
@Dynelements See my private message.
@SubtleShader Everytime Onetrainer says "NotImplementedError: Loading of single file Z-Image models not supported. Use the diffusers model instead. Optionally, transformer-only safetensor files can be loaded by overriding the transformer." Even though I made sure to unzip the files and put them in the diffuser folder. I pasted the entire error into gemini/chatgpt and they both told me to make numerous edits in your model_index file and other config files, telling me to change the architectures and transformers (Qwen2Model/Qwen2Tokenizer/Qwen3ForCausalLM). After constantly following gemini's instructions and copying and pasting new errors again and again, it eventually even told me to delete your text_encoder/transformer folders entirely and replace them with the default ones from the official z-image huggingface page. There's no way that could be right, so I just stopped. Im gonna reinstall Onetrainer again. I'm probably just making a silly syntax error somewhere.
@Tezozomoctli OK, I see. OneTrainer wants the de-turbo safetensors file while AI-Toolkit requries the diffusers version of it. I'm re-uploading both now. Should be ready in 2 hours. See my new OneTrainer instructions above.
Hi @SubtleShader Love your work, I have downloaded the Safetensor Files for ZT3, but it seems that the zipped folder is no longer available for download unless I am looking in the wrong place. Your support would be much appreciated.
@BTA74389 The zip file made Big Love go offline because Civitai's secutity system went paranoid when uploading it again. I will upload again once they gave me a green light. Will send it to you via Chat.
@SubtleShader Hi, one quick related question: After training a LoRA on the DeTurbo diffusers version in AI-Toolkit, what BigLove ZT3 version(s) should (or could) that LoRA ideally be used with for generation (DeTurbo diffuser/safetensor or also the regular Turbo ones)?
@clairevalentine646 Both. Your choice.
@SubtleShader I'll say it again: I really appreciate your art AND your tech support! Thank you my friend! :-D
@SubtleShader I can't see Full Model bf16 (9.14 GB), both files SafeTensor
@_RUST_ Read my updated instructions at the top.
@SubtleShader Select the BigLoveZT3_deturbo.safetensors model?
@_RUST_ If you use OneTrainer ... Are my instructions so unclear?
@SubtleShader I'm using it AI-Toolkit. And I don't see Big-Love-ZT3-De-Turbo-Diffusers where to download it. There is no link
@_RUST_ It's at the top of this comment thread :-D
@_RUST_ AI-Toolkit will download it for you if you enter that. If you do not trust AI-Toolkit use https://huggingface.co/SubtleShader/Big-Love-ZT3-De-Turbo-Diffusers/tree/main
Hiya, just paid 10k for ZT3 Diffusers but I can't download it, just redirects to a json with 3 lines saying "unauthorized".
Can you help please 🙏
Sorry to hear. Civitai having problems, it seems. I wrote you via chat.
For others: Please open a chat with me if you have the same problem. Will give you an alternative download.
@SubtleShader Yep, received and downloading now.....Thank you👍
That's a great model. But I wonder if there's any sampler or scheduler that can make skin texture a little bit less SDXL like, as in the raw base Z-image.
Which of my models do refer to? Ultra4? I actually like Photo/Ultra skin texture better than Z. Besides Photo/Ultra are optimized for DMD2, so more comparable to Z turbo than Z base.
Ok! I'll try to use Ultra as a second pass on Z-image! Thank you!
for the SDXL models, why not just bake in the DMD Lora?
There are around 8-10 reasons. I mentioned them in another comment a few months ago. One reason: DMD2 has a non-commercial license, SDXL has not. Another reason: Not possible to train anymore if baked in.
@SubtleShader ah, i see.. i don't commercialize my models, but for the extreme early access some people love to max out as much as possible it makes sense you wouldn't merge it. would you upload a DMD merge after the early access ends? i can merge the models myself to make it easier to use, but some people have no idea about the benefits of DMD for image quality and accuracy so they use these models without DMD thinking they can get just as good of a result
@SLACK69 So far everyone managed to use the DMD2 lora.. Many people use tons of loras.
also DMD2 is not always preferable. In fact I use it very selectively only
@kunde2 Yes, you can also reduce its strength for more or less softness/sharpness. Much more versatile if not merged.
A few things: First off, I love the update ZT3!! Nice look, follows prompts, great faces and lighting!
Secondly, THANK YOU for the continued hard work and ongoing development!
Thirdly, also THANK YOU for the tech support and integration with AI-Toolkit, etc. You've got out of your way to help us goofs create some great art and I for one appreciate it!
I've posted a couple images for fun as well. I created a Lora with BT3 also and used it in the images I posted as well.
Keep it up my friend!
Big Love Outage
Big Love was offline for 28h due to a security glitch caused by reuploading the ZT3 AI-Toolkit training file. Civitai seems to have increased security lately.
Sorry for the inconvenience. Please contact me for that file. I will not upload it again till Civitai gave me a OK.
Future Big Love Plans
What started with SDXL has grown into something much bigger and much more complex. Expanding across multiple model architectures means more possibilities, but also more obstacles, more experiments, and more late nights figuring out why something refuses to work.
Where others might reach for Thor's hammer, I prefer Leonardo's chisel. Great work not always comes from destruction and reconstruction. It comes from patient refinement, from finding what a model already does well and coaxing it further.
Anyway, here are some short comments about my plans:
Photo6/Ultra5:
In addition to new concepts I plan to improve old ones with lora merging. Will also train more amateur style for improved realism.
Hyper1:
Based on Ultra but trained on higher 2048px resolution
Klein2:
Enhanced default lighting and finally train image pairs to make editing more "naughty".
Mini1:
Klein 4B gives a lot of trouble. Seems BFL made it very hard to train. Still learning what works & what not. Will still take some time.
ZT4:
Will try to finetune the de-turbo version for better pussies & more realism. Keep your fingers crossed that it will work!
ZB1:
Unfortunately Z-Turbo is not compatible to Z-Base. So I might try to come up with a Z-Base compatible "turbo" version.
Chroma1:
I gave up on the idea. Too slow, clumsy poses, low fine detail quality. Does not inspire.
The road ahead is long, but the passion driving my work is longer. What keeps the Big Love project moving forward is a genuine commitment to pushing quality further with every release. Stay tuned and as always thanks for the support!
Loving the Klein1 model, excellent work. I'm a bit of a newb and can't seem to get LoRAs working with it, would you be able to share an example of a workflow including LoRAs?
Add a Load Lora node (double click background and type "Load Lora") and put it right after the Load Diffusion Model node. Connect output & input of both. Hold down the Shift key to drag all output connections from the Load Diffusion Model node to the output of the Load Lora node. That's it!
Hey is there a way to generate more realistic bodies, like fat legs, fat rolls. but not straight up obese bodies? All I get are supermodel bodies or obese bodies.
What model do you refer to? I hope you don't mean Z with its slim barbies. So far I only trained fat asses into SDXL and fat tits with the Insta1 version. But the knowledge is in there. Try fat, plump, thick, obsese, curvy or any of the other 20 synonyms. Something will work.
Sorry I cannot help more. Not into it.
@SubtleShader thanks for the answer, I was refering to the klein one, will keep playing with it then.
It's a pity, using any zimage models and any workflow I have no problems, but it's worth switching to using ZT3 there are distortions. I don't understand what the problem is.
Use the Z-Turbo workflow in the Big Love description. What distortions do you mean?
@SubtleShader Yes, I used the workflow from Big Love description. Here is an example - https://mega.nz/file/OCQXkKKR#DdaXaoEfXMYlbGB7WC4dVtc20kfzRC40JEbo2dqPIIE
@GogerFF You use the ZT3 De-turbo version (Full Model) instead of ZT3 Turbo. De-Turbo needs 20-50 steps and cfg5. I only provide it for training with OneTrainer. Please download Pruned Model bf16 or Pruned Model fp8. They work with 8 steps and cfg1.
@SubtleShader I'm sorry))). I was inattentive. Thanks! Спасибо!
@GogerFF No problem. You are welcome. Civitai has a bad way of naming model versions. Also does not let me rearrange them. If an image is blurry, it means that it needs more steps.
I am getting lots of SD1.5 era body horror in T2I on Forge Neo but I2I is absolutely awesome. I am not sure why that is. I tried recreating your samples but they're all so strangely prompted, not natural language. Do I need to do that too? Would prefer not going back to Pony style tagging.
You mean Klein1? You can prompt anything you like. Qwen is quite flexible. Klein1 does not have so great NSFW anatomy yet. Was not trained enough. Should get better in future.
@SubtleShader I do mean Klein1. Yes thanks. I did not know that. Looking forward to the improved version.
@SubtleShader I've actually installed comfy now and copied the workflow from your images. i still get massive body horror. will probably wait until a new version is out. keep it up man!
@wfux Klein1 is about image editing not txt2img. Image editing is more stable. Try it. Use one of my other models to do txt2img.
I'm using ZT3 with forge neo and getting good results, but when I upscale with image to image the image quality is pixilated. I'm not sure what im doing wrong?
Yep, I use it with Forge Neo too. Sounds like you do something wrong. If you upload a screenshot of your settings, I could take a look.
@SubtleShader Thanks, I sorted it out I was indeed doing something wrong.
@The_Last_Curtains_ Always good to mention briefly what the problem was in case somebody has the same problem in the future and comes here looking for solutions. Up to you of course!
There's no link to download Klein9B version? From where can I have it?
Read below the showcase images.
If you get creative with your search you can DL it for free on HF.
Is the klein model not available at all for running locally?
Read below the showcase images.
Any ideas how to create a missionary pose?
Prompt 1girl, 1boy, sex, penis, missionary position