Have you guys considered using AI to generate your dream toys?
Here's a quick imperfect Reptar using the Deliberate model on
https://aqualxx.github.io/stable-ui/recommended models: Deliberate, Dreamshaper, Dreamlike Diffusion.
tips: use k_dpmpp_2m as the sampler and around 20 steps. guidance should be left relatively low. negative prompts are useful, use them. for example if your image looks too cartoonish just add "cartoon" to the negative prompt. prompts can have weights so it will have a bias towards or against it. for example
>(prompt:1.2), anotherpromptwill produce something that leans towards the first prompt. going past 1.3 can result in strange and incoherent generated images. weights also work on negative prompts. prompts can be grouped and weighted (prompt, anotherprompt:1.2). You can create hybrids of characters by combining the prompt like this
>(Character:AnotherCharacter:YetAnotherCharacter)or
>[Character|AnotherCharacter:YetAnotherCharacter]I've found both formats work just as well. the first character in the hybrid prompt will always be the most apparent. the 2nd and all subsequent characters in the hybrid will be applied in steps/iterations eg. Step 1. Character Step 2. AnotherCharacter Step 3. YetAnotherCharacter and Step 4 will be Character again
to make the hybrid look more like subsequent characters you'll need to repeat it for each step sans the initial character.
>(Character:AnotherCharacter:YetAnotherCharacter:AnotherCharacter:YetAnotherCharacter)this is the only method I've discovered however it can make other prompt less apparent. You can try increasing the weight on subsequent prompts.
Good luck and have fun making your dream toys
