Pixar's RenderMan-Inspo Digital Art SD1.5版本MK1 (ID: 459769)

This checkpoint is the result of various merges of the following public SD 1.5 checkpoints:
C3 - v1.10
https://civitai.com/models/39044
Pixar Style Model
https://civitai.com/models/15773
RCNZ Cartoon 3d - v2.0
https://civitai.com/models/66347
Disney Pixar Cartoon type B
https://civitai.com/models/75650

As with all 1.5-based checkpoints, it is recommended to use a VAE as I did NOT bake one into the merges.

I've created this checkpoint to get more of that "Pixar-y" style while also attempting to make it as versatile as possible in generating various depictions of objects, backgrounds, animals, and people.

This checkpoint is capable of both SFW and NSFW generations; however, it does slightly lean biased toward the SFW end of the spectrum. Reinforce either the positive, negative, or both prompts to get some NSFW results if that is what you intend to generate.

Generations can work with brief or lengthy prompts, but I tend to get the best results using the following settings:

Positive prompt: (pixarstyle), masterpiece, best quality... (Whatever else you want to add here)
Negative prompt: easynegative, bad-artist, bad-artist-anime, (bad_prompt:0.8), (artist name, signature, watermark:1.4), (ugly:1.2), (worst quality, poor details:1.4), bad-hands-5, badhandv4, blurry

Image Resolution: 512x768 or 768x512

CFG Scale: 6.5 (6-7 also works fine)
Sampler: Euler A or DPM++ 2M Karras
Sampling Steps: 24 (Not recommended to go above 32 using suggested samplers)
Clip skip: 1 (2 also seems to produce decent results as well)

Hi-res Settings:
Upscaler: R-ESRGAN 4x+ or DAT x4 (Not recommended to use any Latent upscalers)
Upscale Factor: 1.5 - 2
Denoising strength: 0.3 - 0.4
Hires steps: 20 (Going higher than 24 seems to not yield any noticeable differences in quality)

Feel free to experiment around with the generation parameters to potentially get better generations and upload your results! I'm always eager to see what settings work with which prompting methods ?

?PRO TIP!
Sometimes it is best to use a different hi-res sampling method when generating images with this checkpoint and many others available on CivitAI. In order to enable this option on A1111 or Forge, go to Settings -> User Interface -> UI alternatives -> Enable option: "Hires fix: show hires checkpoint and sampler selection". While there, you should also enable "Hires fix: show hires prompt and negative prompt" to add a separate positive and negative prompt for the hires fix to focus on. Thank me later! ?

描述:

The merge where it all started!

训练词语: