IMG to IMG
Positive: redheaded girl, photo, photorealistic
Negative: anime,drawing, painting, sketch, lowres, blurry, error, jpeg artifacts, low quality, worst quality, messy, 3d, (oversaturated:1), cartoon, sims, (airbrushed :1),(cartoon:1.1), deformed,blender
0.6 Denoising Strength
ControlNet with the Reddit logo as source, methods Canny or Scribble (depending on the SD model used)
For each model, I changed the prompt and the CFG slighty (CFGs from 10 to 20). DDIM, 60 steps, 512x512 upscaled by an external software (Topaz Gigapixel)
Prompt: variations of "red background with a 3d white happy robot with a white antenna in the foreground"
Negative prompt for the photorealistic models: octane, trending on ArtStation, hyper detailed illustration, render, concept art, corona render
Stable diffusion + automatic 11 11
Checkpoint model: photogeninfinity
prompt: aerial drone photography, wide angle nature scene, <lora:add detail:1 >
negative: deepnegative external version (ng_deepnegative_v1_75t
sampling method: DPM ++ SDE Karras
drag drop logo (size 1080x1920)
Preprocessor lineart_standard (From white bg & Black line)
model: control_v11p_sd15_lineart
X type: Prompt S/R
X values: nature scene, scocer field , baseball stadium, golf course, racing track , hiking trail
Y type: nothing