REYAMix

Name: REYAMix - Beta
Rating: 0 (0 reviews)
Author: REYA_012

CHECKPOINT

Original

REYA_012

Updated: May 28, 2024 12:07 PM

No showcase images available, the model won't be visible to others.

Run791

My first try on a model merge (or mix or whatever it's called)!

Consists of Alouarn's Adventure Series model, one of my favourite model for creating a balanced but very good image and Stoke's Stoked Illustration that has the ability to generate amazing illustration, with a touch of Meina and Hassaku's anime aesthetics.

Generated image isn't really 2D, but not 2.5D or realistic either, somewhere in between? I don't know, you tell me XD.

WARNING!

CONTAINS NSFW!

(But rest assured that not all of the image generations are NSFW)

Shout-out to these creators and check out more of their amazing models:

Alouarn: Creator profile Alouarn | **********

Stoke: Creator profile Stoke | **********

Meina: Creator profile Meina | **********

Ikena (Hassaku creator): Creator profile Ikena | **********

I'm making this model's permission as close as the original.

If by any means this model is a problem, please contact REYA_012 on Tensor Art's Discord Server. Also, feel free to roast me if this model is really bad XD.

Thanks and have fun!

Version Detail

Uploaded

Nov 17, 2023 8:28 AM

Base Model

SD 1.5

Description

Beta known issue (so far): 1. Image generation is really focused on person. Background generation is mediocre, especially for the masterpiece tag. 2. Image generation with "abstract illustration" or with complex composition somewhat returns a dark background (needs more testing on this). Prompt such as masterpiece, best quality, etc. isn't necessary but always welcome to be used (especially if you want to generate godly art, though I don't know if this particular model can handle god prompt XD). Negative would be: (worst quality, low quality, normal quality:1.4) and (interlocked fingers). Sometimes (text), (watermark), (logo), jpeg artifact. You can try 1.5 instead of 1.4. You can also try the available negative embeddings and compare which has the best result! > Sampling is either Euler A or DPM++2M SDE Karras. > Upscaler is either R-ESRGAN4x+Anime6B with 0.4 denoising strength (can be lower) or Latent Nearest Exact with 0.5 denoising strength (can be higher). > No VAE is okay. Using ft-mse 840000 is okay. (Haven't tried other VAEs yet).