Of course not. Thank you for your contributions to this community!
Love it โฅ๏ธ that's certainly better than my attempt ๐
As proof that it's better at animal hybrids then humans sitting:
Tests
First 4 attempts at a hybrid. (Rodent with a cat's head inspired by one of the submissions here by @elvith@discuss.tchncs.de )
First 4 attempts at a man on a bench. (These are the most coherent humans the model has created for me so far ๐
)
You can interpret these results however you want.
My thought process was, which two animals would make a completely nonsensical combination? My first attempt trying out SD 3, I believe it's better at animal hybrids than humans sitting on a bench TBH.
Prompt: fantastical hybrid creature combining an elephant and a butterfly. The creature has the body of a small, rounded elephant with iridescent scales. Its head features a long trunk and large eyes, with vibrant butterfly wings extending from the sides. The legs are slender with intricate patterns, and the tail ends in colorful, feather-like appendages. Place this creature in a lush, green forest with colorful flowers and sunlight filtering through the trees, creating a magical and enchanting atmosphere. Photorealistic and cinematic
Model: SD 3
Cool test, thanks for sharing!
Base image:
professional diver deep in the dark depths, (complete darkness), scary rock formations, lower below is a small cottage in the distance Negative prompt: sunlight, bright blue water, skylight Steps: 25, Sampler: DPM++ 2M SDE, Schedule type: Exponential, CFG scale: 7, Seed: 260230308, Size: 1152x896, Model hash: 1718b5bb2d, Model: albedo_base_xl, Version: v1.10.0
Removed diver with A1111 Cleaner
Inpainted submarine:
Deep-submergence vehicle submarine, many cables, air bubbles, deep in the dark depths, deep ocean photography, 8k photo, (gloomy lighting), (facing right) Steps: 25, Sampler: Euler a, Schedule type: Automatic, CFG scale: 7, Seed: 4116327859, Size: 1152x896, Model hash: 1718b5bb2d, Model: albedo_base_xl, Denoising strength: 1, Mask blur: 4, Inpaint area: Only masked, Masked area padding: 120, Masked content: latent noise, Version: v1.10.0