Incredible World – SD 1.5 – v4.0

Incredible World v.4 – this is not just another checkpoint. In this version I added trained Lora on a set of unique photographs which allowed me to create incredible, fantastic, unreal and fashionable girls. I hope that you get interesting generations at this checkpoint and that you will share them on the checkpoint page. There is no universal product, use those suggested in the preview, I, in turn, will also further explore the possibilities of this check-out and share the results on the project page.

Version 3 – Basic model + LORA (ArtMan, Russian Village, RussianClothes ,Expression2 ,epiCRealismHelper)(Recommended Clip Skip 2)

Recommended Settings for version 3:

Steps: 28, CFG scale: 3-7, Hires 16 step 0.4 denoise Upscaler: 4x-NMKD-Siax-200k

Negative for photorealistic: ((watermark,3d,cartoon,paint,draw,illustration,anime,render,3d render)),(canvas,border,frame),(muted colors:1.4), (extra fingers, deformed hands, polydactyl:1.3), ugly, (low detail, image artifacts, pixelation), (worst quality, low quality, poor quality, bad quality:1.35),(worst quality:2), (low quality:2), (normal quality:2), lowres,bad anatomy,bad hands,extra digit, fewer digits, cropped, username,blurry,bad feet,cropped,poorly drawn hands,poorly drawn face,mutation,deformed,jpeg artifacts,extra fingers,fewer digits,extra limbs,extra arms,extra legs,malformed limbs,fused fingers,too many fingers,long neck,cross-eyed,mutated hands,polar lowres,bad body,bad proportions,gross proportions,text,error,missing fingers,missing arms,missing legs,extra digit,extra hands, deformed_finger, artist logo, signature

or

(analog photo:1.2),({prompt}),(high quality, film grain), (natural colors, correct white balance, color correction, dehaze,clarity)
negative: ((3d,cartoon,paint,draw)),(muted colors:1.4). simple background, (extra fingers, deformed hands, polydactyl:1.3), ugly, (worst quality, low quality, poor quality, bad quality:1.35), artist logo, signature

Version 2 – Basic model + LORA (Expression v.2, RealHuman v.1, Prastor v.1, EpicCRealismHelper) (Recommended Clip Skip 2)

Version 1 – Basic model + LORA (ArtRetoch v.2, RealHuman v.1, Expresive_photo v.1)

Recommended Settings for version 2:

(photo RAW),({prompt}),masterpiece, top quality, best quality, official art,intricate detailed,highest detailed,depth of field,bokeh,realistic natural lighting,complex, multiple subjects, 4k HDR, dusty atmospheric haze, high quality textures of materials, volumetric textures, coating textures, metal textures
Negative prompt: ((watermark,3d,cartoon,paint,draw)),(canvas,border,frame),(muted colors:1.4). simple background, (extra fingers, deformed hands, polydactyl:1.3), ugly, (worst quality, low quality, poor quality, bad quality:1.35), artist logo, signature

Steps: 28, Sampler: DPM++ 2M SDE Karras, CFG scale: 3-7, Hires 12 step 0.5 denoise

Recommended Settings:

Negative prompt: ((3d,cartoon,paint,draw)),(muted colors:1.4). simple background, (extra fingers, deformed hands, polydactyl:1.3), ugly, (worst quality, low quality, poor quality, bad quality:1.35), artist logo, signature

Sampler: DPM++ 2M SDE Karras

The CFG scale must be at most 6 and the sampling steps must be at least from 15 and above. I recommend using hires and upscaler. I usually use Clip Skip 1-2.

Leave a Comment

Details
Model TypeCHECKPOINT
Base ModelSD 1.5
Model Hash39ECBD7D27
Model Versionv4.0
Trained Words
Picture of edward.cui

edward.cui