NOTE THIS IS STILL BETA v0.4: + add some real face photos (mainly close-ups) + some more realistic 3d images + and some more photos from NijiStrangely, sdxl cannot create a single style for a model, it is required to have multiple styles for a model. Over-multiplication is the problem I'm having with the sdxl model. Initially I just wanted to create a Niji3d model for sdxl, but it only works when you don't add other keywords that affect the style like realistic, photo, realism, artist name, drawing style...etc. So I decided to change from Niji's 3D-only model to an even more diverse model.Negative prompt:(worst quality:1.5), (low quality:1.5), (normal quality:1.5), lowres, bad anatomy, bad hands, multiple eyebrow, (cropped), extra limb, missing limbs, deformed hands, long neck, long body, (bad hands), signature, username, artist name, conjoined fingers, deformed fingers, ugly eyes, imperfect eyes, skewed eyes, unnatural face, unnatural body, error, painting by bad-artistSDXL usage is completely different from all previous models: 1/ the number of cfg is greatly affected, can replace clip skip. if you want to create surreal images cfg 3-4 or 5 is suitable, cartoon or 3d character images should be clear, cfg should be 6-7, and real people photos can sometimes be up to 7-8, but only if you need high precision and detail to get cfg up to 9-10 or higher, I haven't tried it yet. 2/ the number of steps also has a much greater influence on the results, the image only starts to achieve high quality when it is 50 steps or more. Right now I don't know where its limit is, it's like the higher the step count, the better the image. Of course it can be different when changing styles, changing models, or even changing samplers. I still can't try it all. 3/ the way to get prompts will be different from other sd models, but I don’t know how, just feel that the prompts are too cumbersome and not as good as before and instead of writing in tag style, You can try to choose a descriptive writing style like Midjourney4/ SDXL it seems that the more data, the less keywords are needed in prompt and negative prompt. 5/ That's all for now, I'll keep updating as new information or changes become available, anyway it's new technology for all of us.If you like my content, support me at: <a target="_blank" rel="ugc" href="https://linktr.ee/Duc_Haiten">https://linktr.ee/Duc_Haiten</a>For those who don't have a powerful enough GPU, or are just too lazy to install enough complex software to run the model. <a target="_blank" rel="ugc" href="http://mage.space">mage.space</a> will be a great site for you, with the number of models increasing every month, including my model of course. Especially for only $15 per month you can have unlimited photo creation for yourself, no credit score, no time, absolutely truly unlimited photo creation in a month.

DucHaiten-Niji-SDXL

duchaitenNijiSDXL_v01.safetensors

v0.1

v0.4

v0.3.5

v0.3.0

Strangely, sdxl cannot create a single style for a model, it is required to have multiple styles for a model. Over-multiplication is the problem I'm having with the sdxl model. Initially I just wanted to create a Niji3d model for sdxl, but it only works when you don't add other keywords that affect the style like realistic, photo, realism, artist name, drawing style...etc. So I decided to change from Niji's 3D-only model to an even more diverse model.Negative prompt:(worst quality:1.5), (low quality:1.5), (normal quality:1.5), lowres, bad anatomy, bad hands, multiple eyebrow, (cropped), extra limb, missing limbs, deformed hands, long neck, long body, (bad hands), signature, username, artist name, conjoined fingers, deformed fingers, ugly eyes, imperfect eyes, skewed eyes, unnatural face, unnatural body, error, painting by bad-artistSDXL usage is completely different from all previous models: 1/ the number of cfg is greatly affected, can replace clip skip. if you want to create surreal images cfg 3-4 or 5 is suitable, cartoon or 3d character images should be clear, cfg should be 6-7, and real people photos can sometimes be up to 7-8, but only if you need high precision and detail to get cfg up to 9-10 or higher, I haven't tried it yet. 2/ the number of steps also has a much greater influence on the results, the image only starts to achieve high quality when it is 50 steps or more. Right now I don't know where its limit is, it's like the higher the step count, the better the image. Of course it can be different when changing styles, changing models, or even changing samplers. I still can't try it all. 3/ the way to get prompts will be different from other sd models, but I don’t know how, just feel that the prompts are too cumbersome and not as good as before and instead of writing in tag style, You can try to choose a descriptive writing style like Midjourney4/ SDXL it seems that the more data, the less keywords are needed in prompt and negative prompt.5/ That's all for now, I'll keep updating as new information or changes become available, anyway it's new technology for all of us.If you like my content, support me at: https://linktr.ee/Duc_Haiten

Files with SHA256: 378025e2c27cf49288a6f0bb29162520388c24ebbd75f3545dccecfb936e123b