Pony to TAME txt2Img


Updated:

Workflow Preview

Showcases

You can change the base model in use (Pony Realistic Diffusion v3) with whatever PDXL/SDXL model you like, but make sure to switch out the header and footer prompts, as those are specific to this model. The light and shadow lora is broadly useful at improving realistic lighting and textures, but is aimed at real and semi-real pony models.

If you switch PRDv3 for a non-realism model, you'll need to scour the "Theme" box and take all but the strict setting info to the "additional & corrective terms" box over in 2nd pass

Nodes Detail

50 Nodes
Primitive Node Types
13
  • CLIPSetLastLayer1
  • VAELoader1
  • PreviewImage2
  • SaveImage1
  • CLIPTextEncode5
  • UpscaleModelLoader1
  • CheckpointLoaderSimple2
Custom Node Types
37
  • PrimitiveNode4
  • CR SDXL Aspect Ratio1
  • BetaSamplingScheduler1
  • Reroute8
  • SAMLoader1
  • CR Seed1
  • MaskToImage1
  • MediaPipeFaceMeshDetectorProvider //Inspire1
  • SamplerCustom4
  • FaceDetailerPipe1
  • VAEDecodeTiled_TiledDiffusion1
  • FromBasicPipe_v21
  • pipe-util-to-basic-pipe1
  • Fast Groups Bypasser (rgthree)1
  • UltimateSDUpscale1
  • SplitSigmas1
  • KSamplerSelect1
  • LoraLoaderModelOnly5
  • ToDetailerPipe1
  • CR Switch Model and CLIP1
Comments