Big update of DucHaitenAIart, v4 is able to receive more diverse, more detailed prompts with gorgeous colors and more realistic shadows. The image has the breath of 3D anime, but the material is much more realistic.
If you like my content, support me at:
https://linktr.ee/Duc_Haiten
for those who don't have a powerful enough GPU, or are just too lazy to install enough complex software to run the model. mage.space 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.
*****
All sample images only use text to image, no editing, no image to image, no restore face no highres fix no extras.
DPM++ 2S a Karras cfg 11
negative prompt:
illustration, painting, cartoons, sketch, (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, ((monochrome)), ((grayscale)), collapsed eyeshadow, multiple eyeblows, vaginas in breasts, (cropped), oversaturated, extra limb, missing limbs, deformed hands, long neck, long body, imperfect, (bad hands), signature, watermark, username, artist name, conjoined fingers, deformed fingers, ugly eyes, imperfect eyes, skewed eyes, unnatural face, unnatural body, error
Description
Big update DucHaitenAIart_v2.0
DucHaitenAIart_v2.0 is an extended version of v1.2, v2.0 improves everything to be better than v1.2, can do things 1.2 cannot do, more diverse and more detailed
FAQ
Comments (25)
safer safetensors format please...
New version still showing to have a trojan in it.
the file has been verified by the site, and I also upload it to safetensor so people can download it if they don't feel secure ckpt
@duchaiten Understood. The safetensor version is not showing the issue. Thank you for the reply. You do awesome work, and it is appreciated.
FWIW, I downloaded the .ckpt version and scanned it on my system. It does not show any issues.
Love the previews, but it seems, my prompts don't get me very close, even if I use the excat same word. Do I have to use a specific VAE?
hmm i don't use VAE or anything like that, no higher fix, clip skip 1, 512x760 but you can try 768x960 for better picture
@duchaiten I came pretty close, adding a missing Bracket "((Lying on her back...". I guess some of those got "lost in translation". I'll experiment a bit with the weights of the lights.
@ragnarook ok, have fun with it and post some results here, it will help me a lot, you can also try some of my other models, DucHaitenDarkside is not very good, today i will try to give it out its new version
@ragnarook can you provide the prompt you used for generation because I cant seem to get anything close to the sample image of the serpent goddess
@bumbumbum2323 Cinematic film still, full body, of ((Norse Female serpent goddess)), (Hel), pale skin, (Lying on her back in a pit of snakes)), legs spread, toned abs, large breasts, seductive look, exposed pussy, large ass, ((tentacle sex)), ((detailed facial features)), alluring blue eyes, (wet, detailed skin), (covered in snakes), blonde hair, magic, (dark cave:1.2), (cold colors), damp, moist, intricate details, shallow depth of field, [volumetric fog], cinematic lighting, reflections, photographed on a Canon EOS R5, 50mm lens, F/2.8, HDR, 8k resolution, cinematic film still from Vikings
Negative prompt: bright, (monochrome:1.3), (oversaturated:1.3), bad hands, lowers, 3d render, cartoon, long body, ((blurry)), duplicate, ((duplicate body parts)), (disfigured), (poorly drawn), (extra limbs), fused fingers, extra fingers, (twisted), contorted, conjoined, ((missing limbs)), logo, signature, text, words, low res, boring, mutated, artifacts, bad art, gross, ugly, poor quality, low quality
Size: 512x768, Seed: 2318004813, Steps: 20, Sampler: DPM++ 2S a Karras, CFG scale: 11, Model hash: 61d24597
@duchaiten just noticed (but have no clue how important this is): The Model hash in the preview (61d24597) is different from the download. Is this a new version of your awesome model?
@bumbumbum2323
Not perfect, but starting "((Lying ... in a pit of snakes))" with enough brackets to fit the rest of the phrase (there's one missing in the preview-text) and the picture begins to resemble the preview. Emphasize aspects like the light or tentacles, a bit more and you might get somewhere...
Cinematic film still, full body, of ((Norse Female serpent goddess)), (Hel), pale skin, ((Lying on her back in a pit of snakes)), ((legs spread)), toned abs, large breasts, seductive look, exposed pussy, large ass, ((tentacle sex)), ((detailed facial features)), alluring blue eyes, (wet, detailed skin), ((covered in snakes)), blonde hair, magic, (dark cave:1.2), (cold colors), damp, moist, intricate details, shallow depth of field, [volumetric fog], cinematic lighting, reflections, photographed on a Canon EOS R5, 50mm lens, F/2.8, HDR, 8k resolution, cinematic film still from Vikings
Negative prompt: bright, (monochrome:1.3), (oversaturated:1.3), bad hands, lowers, 3d render, cartoon, long body, ((blurry)), duplicate, ((duplicate body parts)), (disfigured), (poorly drawn), (extra limbs), fused fingers, extra fingers, (twisted), contorted, conjoined, ((missing limbs)), logo, signature, text, words, low res, boring, mutated, artifacts, bad art, gross, ugly, poor quality, low quality
Steps: 28, Sampler: DPM++ 2S a Karras, CFG scale: 11, Seed: 3653582821, Size: 640x960, Model hash: 0ef25ae93a, Model: duchaitenaiart_V20, ENSD: 31337
@ragnarook or maybe you are confusing ckpt and safetensor
@ragnarook ah i remember, it's still the model v2.0 safetensor but it's the model i converted from ckpt to safe and put it in webui on huggingface, it has different code but it's all v2.0 no change
Why can’t I exactly recreate your gens? Are you using any vae or embeddings?
no
i use skip clip 1
I get closer to the pregens with "DPM++ 2M Karras", or "DDIM" (instead of "DPM++ 2S a Karras" that is mentioned in the pregen-descriptions), but I can't recreate them either. Perhaps the current Automatic1111 (which I currently use) does something to the prompts?
@duchaiten Do you perhaps use something different, than Automatic1111's web-ui? Say Invoke UI or the good old command line?
@ragnarook maybe it's because i use the built in webui on huggingface made by a friend, with the gpu provided by huggingface for free
@duchaiten
Hardware might make a difference, too, but I think it's because something has changed with Automatic1111. When I go to the Settings/Compatibility and turn on "[X]Use old karras scheduler sigmas (0.1 to 10).", I get extremely close to your pregen-picture (Setting the "Eta Noise Seed Delta" to "3137" helps as well).
Could you perhaps take a look at the settings (don't change them, just see, if it's there), and see if you even have "Use old Karras..." and "Eta Noise Seed Delta" and whether they're turned on/which value they have?
@ragnarook because my machine has a pretty weak gpu so i used colab and huggingface to test, and it has pre-set parameters from other people so i can't see the settings and edit them
@duchaiten
I took a look at the Hugging Face Colab (had no clue about it before) and tried the default values for "Eta Noise Seed Delta", which is "0" (the tutorial I used to set up my machine used 31337).
For newer versions of Automatic1111: "Eta Noise Seed Delta 0" in combination with "[X]Use old karras scheduler sigmas (0.1 to 10)" works perfectly to re-create the previews.
Thank you very much! And thanks for your patience!!!
(This bugged me with quite a lot of examples I found on the interne).
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.