ReKemono -v9.0
-Enhanced style consistency, improved prompt feedback, added material library, added female-oriented content
OG Name LUNO FURRY illustrious SDXL
I created a keyword generation Bot, and I deployed it in the Prysai Discord channel. It supports comic style/Furry style. It is free. https://discord.gg/52q6zuxcr2
Prompt Word Agent Bot for REED | Civitai
My personal recommendations:
• Sampler: euler a or euler
• Steps: 15–40 (Best around 30 )
• CFG: 3~6 is ideal; for short prompts, it can go as low as 2, while longer prompts should not exceed 7
• Resolution: 720p or higher with hires fix enabled; upscale to 1080p+ using Remacri for the best results
Additionally, I highly recommend pairing it with this upscaler model 👉 Remacri
Negative Prompt
lowres,early,monochrome,greyscale,worst quality,bad_quality,normal quality,lowres,anatomical nonsense,bad anatomy,anatomical nonsense,watermark,simple background,transparent,bad_feet,bad_hands,logo,text,bad_anatomy,signature,face backlighting,(worst quality, bad quality:1.2),jpeg artifacts,censored,extra digit,ugly,deformed anatomy,bad proportions,
If you like my work, please post the images you created, thank you!
Description
v7.0! The first version after the model was renamed
Enhanced style consistency, improved prompt feedback, added material library, added female-oriented content
FAQ
Comments (12)
This is likely the best model for kemono and bara. Japanese/Chinese artist tags work better on here, than indigo XL. Even the Loras for those artists work better on here.
Thank you very much for giving such a high rating!
[Lucky Draw Announcement]
I hope Rekemono can win first place in the bidding results of the first week of February. Additionally, I will launch a one-month Seaart monthly subscription giveaway in the Prysai Discord community「https://discord.gg/52q6zuxcr2 」
where the winning members will receive an online membership provided by Seaart.ai! I would like to thank Seaart for their continuous support and recognition.
We did it! First place in a bid for the very first time ever! I'll be kicking off a four-week membership giveaway for it on February 2nd, North America time.「https://discord.gg/52q6zuxcr2 」Can't wait to see you on the Prysai Discord!
What is different from this and Indigo Furry? What training set is used? e621?
In fact, ReKemono and Indigo Furry are not closely related; it is just used for adapting NAi Lora. In the early stages, I added a personal style-related training set, and subsequently, I have been extracting character sets from popular models. Their data is universal for me, but it has been diluted. The main features of Rekemono are its unique style and tolerant adaptability; you can redraw any image in Rekemono, supporting most character codes.
@ReedVaughn Okay, thanks for information!
@rytlockdaddy All models listed as checkpoint merges do not have additional training data unlike a trained checkpoint. Were limited to using LoRAs in order to add new data if we watned to. pretty much all illustrious models contain the same training data(Noob AI & Chenkin, Just add more recent data). Whats changing with model merges is the base style, text encoder, and wieghts for the Unet blocks(this effects everything from anatomy to prompt adherence).
So think of trained checkpoints like an expansion pack, while merges are like texture packs/mods.
Indigo doesnt really share the same models merged in ReKemono.
Indigo uses converted Pony models + SDXL models which provides a lower quality result compared to ReKemono that uses the most current illustrious/NAI combinations as a base model.
If youre coming from Indigo XL, the only possible downside I can see, is that the base style of ReKemono is less semi-realistic & more 2D. But that doesnt mean this model is incapable of emulating the semi-realism of Indigo. Because this model is like a full spectrum in versatility.
@LatteLeopard Ah thanks for the explaination!
А меня очень радут что эта модель так часто обновляется, но теперь она акцентрируется на "кемоно"? Если так, то как получить лучший кемономими результат?
If I understand your question correctly, the answer is to add the names of the corresponding animal ears: "cat ears" "rabbit ears."
Great style flairs! 🙈👍