@Wick_J4 I used entirely different prompts, It doesn't matter. Regardless of whatever I input, I get the same output.

bro, I see the keywords "Renee Olstead,light blonde, light curls," in your photo. looks like you didn't remove them, and lora cutemix4 is the asian woman

@veai The image is supposed to just be the Renee olstead bit onwards (that's my standard control prompt I test all new models with), what's happening is the model inserts the other stuff (light blonde etc) beforehand. It's like that prompt is baked into the model as a default and it always applies it for some reason.

@veai download the model yourself and run it and you'll see what I mean before downvoting (I'm trying to make the modelmaker aware of a bug) Doesn't matter what prompt I enter it automatically inserts the following into the generation beforehand: -

8k, raw photo, best quality, masterpiece:1.2, realistic, photo-realistic:1.37, high detailed background, 1girl in a blue shirt, shy expression, cute, cityscape, night, rain, wet, professional lighting, photon mapping, radiosity, physically-based rendering, , beautiful breasts, medium breast, look at viewer, high detail raw color photo, near shot, of girl wearing ancient chinese dress, wearing thin white shirt, clothed, standing in the middle of the battlefield, dragon tattoos on breast, elegant, beautiful face, blue eyes, girl, 30 yeas old, light blonde, light curls, cybernetic impalnts:1.2, augmentations, detailed skin, supple skin pores, silicon, metal, intricately detailed, fine details, hyperdetailed, bloom:0.4, raytracing, subsurface scattering, diffused lighting, side lighting:0.5, shallow depth of field, photographed on a canon eos r5, 85mm lens, f/2.8, sharp focus, smooth, cinematic film still <lora:mix4:0.8

I test a lot of models and I haven't encountered a bug like this before.

@Mirabilis i downloaded and tried it a few days ago and have had no problems except for a few minor bugs like hand and eye sometimes still not good.
I see you are copying and pasting without reading. the first part you have the words "in a blue shirt, shy expression, cute, cityscape, night, rain, wet," but the next you add "girl wearing ancient chinese dress, wearing thin white shirt, clothed, standing in the middle of the battlefield, dragon tattoos on breast"

Is there anyone who wears a blue shirt but also wears old Chinese clothes? standing in the city at night when it rains, standing in the middle of the battlefield?

In general, I try different prompts for different image creation needs. I don't copy and paste so maybe I don't understand why I don't see the error as you say. even with just a few simple keywords like "1girl, beautiful" I don't get the same error as you

@veai  the uploaded model can't even generate its own preview images. This is the prompt for the brunette in the Bikini. By rights you should be able to replicate it.

8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3,<lora:mix4:0.4>,mix4,(8k, RAW photo, best quality, masterpiece:1.2), (realistic, photo-realistic:1.37),high detailed background,1girl, random outfit ,shy expression,cute,cityscape, night, rain, wet, professional lighting, photon mapping, radiosity, physically-based rendering,,beautiful breasts,((big breast)), look at viewer,

Negative prompt: (low quality, worst quality:1.4), (bad_prompt_version2:0.8), (monochrome:1.1), (lowres:1.1), (loli:1.2), (child:1.2), (teenage:1.2), (aged down:1.2), blurry, censored, (worst quality, low quality:1.2), username, watermark, signature, text, (low quality, worst quality:1.4), (bad_prompt_version2:0.8), (monochrome:1.1), (lowres:1.1), (loli:1.2), (child:1.2), (teenage:1.2), (aged down:1.2),disfigured, missing limbs, extra limbs, extra legs, extra arms, blurred, missing fingers, discolored nipples, skintight, fat, obese, sfw, censored, (((bodysuit))), ,missing fingers, amputee, bodybuilder, muscular, child face, young face, (grayscale), (monochrome:1.3), nude, female nipples,umbrella

Size: 512x744, Seed: 2536036692, Model: MixReal, Steps: 51, hashes: [object Object], Sampler: Euler a, CFG scale: 4.5, Model hash: c7c8ea413f

@Wick_J4  Did a bit of testing with another member of the civitai Discord. The issue is to do with the Model Keywords. If you have model keywords enabled like I do (to access any model styles etc) then the model front-loads all of the keywords into the prompt. What is notable is that the last one <lora:mix4:0.8 is missing '>' at the end which may be the reason why it is frontloading them versus waiting to be triggered. Generally, model keywords shouldn't activate unless triggered in the prompt.

Change ")" at <lora:mix4:0.8>. I think op should remove lora as trigger word because people may have different name for same lora

@Mirabilis That's why I told you to read the prompt carefully

@veai Again it's nothing to do with the prompt, it's to do with the model keywords automatically triggering if you have model keyword enabled in A1111 (https://github.com/mix1009/model-keyword). That shouldn't happen as a default mode. Plenty of models have keywords (such as particular styles), but they only trigger if specifically mentioned in the prompt (Modelshoot, Analog, etc etc). With this model, if that extension is active all the keywords trigger automatically at the beginning of the prompt. That shouldn't happen.

https://i.imgur.com/W4b2fSp.png

^ Those are the trigger words

8k, raw photo, best quality, masterpiece:1.2, realistic, photo-realistic:1.37, high detailed background, 1girl in a blue shirt, shy expression, cute, cityscape, night, rain, wet, professional lighting, photon mapping, radiosity, physically-based rendering, , beautiful breasts, medium breast, look at viewer, high detail raw color photo, near shot, of girl wearing ancient chinese dress, wearing thin white shirt, clothed, standing in the middle of the battlefield, dragon tattoos on breast, elegant, beautiful face, blue eyes, girl, 30 yeas old, light blonde, light curls, cybernetic impalnts:1.2, augmentations, detailed skin, supple skin pores, silicon, metal, intricately detailed, fine details, hyperdetailed, bloom:0.4, raytracing, subsurface scattering, diffused lighting, side lighting:0.5, shallow depth of field, photographed on a canon eos r5, 85mm lens, f/2.8, sharp focus, smooth, cinematic film still <lora:mix4:0.8

^ this is what the model inserts. Same thing.