RDBT - Anima
Back

RDBT - Anima
0 1 1

Manga

2D

RDBT - Anima

RDBT [Anima]

Mid scale finetuned + guidance distilled. Higher aesthetic, less deformed images, still creative, 3~4x faster, without overfitted shiny plastic glossy unified AI slop style.

I use it as a starting point to stack more style LoRAs.

See this page for update log. Random experiment, random quality. New version != better version. Feel free to leave feedback.

See this page for original LoRA (update more frequently, probably).

Sharing merges using this model is not allowed. Known model thieves: NukeA.I (closed-weight merged model on tensorart),

This model is based on

Usage:

Settings:

CFG scale: 1~4. This model has been guidance distilled. You can disable CFG (CFG 1) and run the model 2x faster. Cover images are without CFG for demonstration.

Steps: 24. Guidance distillation != step distillation. If you need low steps (8~12). Try to add 0.2~0.5x turbo lora.

Prompt

Always specify style, or use a style LoRA. Otherwise, you will get random/mixed style. This model does not provide overfitted default style. This is a feature, not a bug.

Quality tags:

It's recommended to omit all the quality tags, or just keep the "masterpiece", if you're not confident. Omitting those redundant tokens allows LLM to pay more attention on other words.

Quality tags have been reinforced during distillation. Thus they don't have noticeable effects. Same as negative tags. If you use cfg, there is no need to dump "score_1, blurry, worst quality, jpeg artifacts, extra arms,... x100 words" in your negative prompt. Those things have been distilled out.

Training settings:

~10k images finetuning -> guidance distillation

All captions are NL from Google Gemini.

Optimizer: adamw, constant lr 0.00002.

LoRA rank/alpha 24.

Guidance distillation target CFG 4.

Block 0-2 and adaln linear layers are skipped.

This model is sourced from an external transfer (transfer address: https://civitai.com/models/2364703/rdbt-anima?modelVersionId=3017013 ）,if the original author has objections to this transfer, you can click，
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request

Saki爱Comfyui

Saki爱Comfyui

Manga

2D

Model Information

Active
Original author：
reakaakasky
Model Type：
LoRA
Basic Model：
anima
Resource Name：
models/loras/rdbt_anima_b1_v0.39.b.safetensors
MD5：
f6e81fe5e969f93d5deaebd2c077afde

RDBT [Anima]

Mid scale finetuned + guidance distilled. Higher aesthetic, less deformed images, still creative, 3~4x faster, without overfitted shiny plastic glossy unified AI slop style.

I use it as a starting point to stack more style LoRAs.

See this page for update log. Random experiment, random quality. New version != better version. Feel free to leave feedback.

See this page for original LoRA (update more frequently, probably).

Sharing merges using this model is not allowed. Known model thieves: NukeA.I (closed-weight merged model on tensorart),

This model is based on

Usage:

Settings:

CFG scale: 1~4. This model has been guidance distilled. You can disable CFG (CFG 1) and run the model 2x faster. Cover images are without CFG for demonstration.

Steps: 24. Guidance distillation != step distillation. If you need low steps (8~12). Try to add 0.2~0.5x turbo lora.

Prompt

Always specify style, or use a style LoRA. Otherwise, you will get random/mixed style. This model does not provide overfitted default style. This is a feature, not a bug.

Quality tags:

It's recommended to omit all the quality tags, or just keep the "masterpiece", if you're not confident. Omitting those redundant tokens allows LLM to pay more attention on other words.

Quality tags have been reinforced during distillation. Thus they don't have noticeable effects. Same as negative tags. If you use cfg, there is no need to dump "score_1, blurry, worst quality, jpeg artifacts, extra arms,... x100 words" in your negative prompt. Those things have been distilled out.

Training settings:

~10k images finetuning -> guidance distillation

All captions are NL from Google Gemini.

Optimizer: adamw, constant lr 0.00002.

LoRA rank/alpha 24.

Guidance distillation target CFG 4.

Block 0-2 and adaln linear layers are skipped.

This model is sourced from an external transfer (transfer address: https://civitai.com/models/2364703/rdbt-anima?modelVersionId=3017013 ）,if the original author has objections to this transfer, you can click，
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request