Added Lightning version of v3.3.5.1.5, in just 13 steps you can create stunning high-quality photos, but please note the following important parameters:
Some notes when using the Lightning model:
Pay special attention, when using hires fix on this model, set Denoising strength to 0, and Hire steps 8-10
Sampling method: Euler a (You can try other sampling methods, but most of them are unusable)
Sampling steps: 13 (If it is lower, it will lack important details, but if it is higher, redundant details will appear)
CFG Scale: 1,5 (Lower will not have enough accuracy or sharpness or vivid colors in the image, and anything higher will cause spots and color errors to appear)
Negative prompt: (This doesn't have to be the same as mine, perhaps yours might be even better.)
(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-artist
Some people have reported bugs in my v3 version, so this will be a bug fix version of v3
If you like my content, support me at:
Patreon in here:
https://linktr.ee/Duc_Haiten
If you want to use my models to create unlimited images at a small cost without having to install anything, mage.space will be a good choice, i have some models that are exclusively available on there.
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-artist
Steps: 50
Sampler: Euler a
CFG scale: 9
Size: 1024
Description
FAQ
Comments (13)
Question, what about refiners with these custom sdxl checkpoints. Just forget about the refiner?
you mean other people will take my model and continue to train and become their model?
@DucHaiten I'm sure you already know this, but SD XL was designed as a 2 part process with the model, and then refiner. https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/blob/main/sd_xl_refiner_1.0.safetensors
I have been running the stuff I generate with your model through the standard refiner, sometimes it seems to help and other times it makes them worse. What is your process?
@moxie1991 I think SAI said the refiner probably wouldn't be necessary any more once there were finetunes. But "sometimes it seems to help and other times it makes them worse," seems to be the consensus.
@moxie1991 I haven't tried much but most of the time it gives worse results
I get the feeling that nobody actually uses the SDXL refiner :)
@hourglassfcupslimwaist indeed, mostly experienced bad results
does anyone else getting error in google colab while running automatic 1111
I updated my webui but still got:
AttributeError: 'GeneralConditioner' object has no attribute 'process_tokens'
Anyone see the same thing?
I see that your finetuning has very good results, can I ask how long did you train on what GPU
Are these pruned and fp16? It seems quite big in comparison to the 2GB of SD1.5
SDXL is huge, this is the standard size
Can we get an fp16 version?... I tested this against some models that were fp16 (like counterfeitXl) and this one seems to be a bit slower (I have a GTX970, so every little detail seems to matter, haha...).
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.







