Assets will forever be the hardest part of game dev for me because I'm simply not a visual artist. Yet I refuse to use already-created assets readily available in many online stores. My assets, sprites, need to come about from my brain, even if they don't come out of my hands.
I'm still working with Stable Diffusion and the many available models but nothing comes close to usable. I still plan on bringing any AI generated assets into Photoshop, as I've been doing, and cleaning them up, but so far the sprite sheets I'm generating in Stable Diffusion are simply not it. I'm sure there's a lot of user error, but also the models just aren't sufficiently trained, or so I believe.
Here's I'm using RealVisXL_V4.0.safetensors with the Pixel_Art_Sprite_Sheet_space_candy_media Lora with the following prompt:
sprite sheet, silent comic, jumping, parody, blue background,
male focus, multiple views,
closed eyes <lora:Pixel_Art_Sprite_Sheet_space_candy_media:1>
Negative Prompt:
(worst quality:2), (low quality:2), (normal quality:2), (jpeg artifacts),
(blurry), (duplicate), (morbid), (mutilated), (out of frame),
(extra limbs), (bad anatomy), (disfigured), (deformed), (cross-eye),
(glitch), (oversaturated), (overexposed), (underexposed),
(bad proportions), (bad hands), (bad feet), (cloned face),
(long neck), (missing arms), (missing legs), (extra fingers),
(fused fingers), (poorly drawn hands), (poorly drawn face),
(mutation), (deformed eyes), watermark, text, logo, signature,
grainy, tiling, censored, nsfw, ugly, blurry eyes, noisy image,
bad lighting, unnatural skin, asymmetry
Other parameters:
Steps: 30, Sampler: DPM++ 2M, Schedule type: Karras, CFG scale: 7,
Seed: 1886189679, Size: 512x512, Model hash: 912c9dc74f,
Model: RealVisXL_V4.0, Denoising strength: 0.45, Tiling: True,
Hires upscale: 2, Hires steps: 15, Hires upscaler: Latent,
Lora hashes: "Pixel_Art_Sprite_Sheet_space_candy_media:
4ee6879e3489", Version: v1.10.1
And the final image:
There was a female cardinal in my backyard :)
Top comments (0)