Z-Image version is Available!
Nova Asian XL
Nova Asian XL is an asian photo-realistic pony SDXL model with the looks of generic sd15 asian face
It is also ponified version of PhotoAsian on SD15
All images on example were created with diffusers with custom png tags
Rules
You cannot use the generated images for commercial use if it's not edited (or just turning it to black and white)
You can share images without any restriction if you don't monetize it
Advertising this model to outside is always welcome
Recommend Settings
Sampler: Euler a
Steps: 20~40
Clip Skip: 1-2
Denoising Strength: 0.65 - 0.8
(Pony)
CFG Scales: 5~7
Prompt: score_9, score_8_up, score_7_up, realistic, photo, detailed face and eyes, detailed skin, (cute:1.1), BREAK
Negative Prompts: score_6, score_5, score_4, score_6_up, score_5_up, score_4_up, source_anime, source_furry, source_pony, source_cartoon, blurry, jpeg artifacts, normal quality, worst quality, low quality, missing fingers, extra digits, fewer digits, bad eye
Add "nsfw, nude" to Negative Prompt if you want SFW image
(Illustrious)
CFG Scales: 3 - 5
Prompt: masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, absurdres, scenery, photorealistic, {Prompt}, BREAK, depth of field, photorealistic details
Negative Prompts: anime, 3d, rendered, modern, old, oldest, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digits, fewer digits, very displeasing, (worst quality, bad quality:1.2), bad anatomy, sketch, jpeg artifacts, signature, watermark, username, simple background, conjoined, bad ai-generated
Description
Better prompt adherence and facial structures
FAQ
Comments (8)
V4 seems to be a big improvement; nice work!
Not being obedient enough.
I suggest not using Adetailer or any texture embeddings such as AissistXL
Also, I noticed that you use LoRAs all 1.0 alpha
I suggest set all of them to at most 0.7 and that way you can get great results
Adding "dynamic pose, dynamic angle" to prompt would also improve the overall looks
You need to change model License. dif from description
Thanks for reminding.
I didn’t put the restriction of selling images is because if I put the restriction, they will ask me more about selling images rather than the other valid questions. So comparing the amount of people to ask, I choose not to
How would you even go about sending a subpoena to an AI artist?
What would that subpoena entail, and how would you go about discovering someone used your model? (Especially with a creation that utilizes IPadapters, custom Unet attentions, ControlNets, LORAs, etc.) 🤔
Ok, so I just finished doing some research. Turns out you can't enforce licensing because Illustrious is open source. So you might want to change that on all your Illustrious based models. Furthermore, by law, you're allowed to make money from the outputs; simply, you aren't allowed to copyright your outputs.
I don't know if the Illustrious makers can sue you, but if you're imposing your own "licensing", I'd imagine they certainly could.
@Intosica For subpoena issue, I cannot know about it
For the licensing about images, it's not restriction: more like recommendation
The rule: few edits needed to monetize, is to make sure to not flood the entire of market with only ai-generated images
Paying respect to the source and all of artists that contributed in some way (or forced to), It's better to have these rules
I know that illustrious have its own unavoidable rule but the problem is that the original model structure of illustrious model is not major part of the model
There are lots of models I merged to make them and have pretty much different construction for encoder as well
In conclusion, you can just monetize it if you want but don't do it too much