Aesthetic Quality Modifiers - Masterpiece
Training data is a subset of all my manually rated datasets with the quality/aesthetic modifiers, including only the masterpiece tagged images.
ℹ️ LoRA work best when applied to the base models on which they are trained. Please read the About This Version on the appropriate base models, trigger usage, and workflow/training information.
Recommended prompt structure:
Positive prompt (quality tags at the start of prompt):
masterpiece, best quality, very aesthetic, {{tags}}, {{natural language}}Description
Trained on NoobAI-XL (NAI-XL) V-Pred 1.0-Version
Recommended prompt structure:
Positive prompt (quality tags at the end of prompt):
{{tags}}
masterpiece, best quality, very aestheticWith the kohya_ss dev branch, v_paratemization ,zero_terminal_ssr enabled, and noise offet set to 0.
Included some newly rated images and small updates to match the noobai tagging:
by {artist}->artist:{artist}very aesthetic->very awa
Previews are generated in Forge with DynamicThresholding (CFG-Fix) Integrated enabled, settings:
dynthres_enabled: True, dynthres_mimic_scale: 7, dynthres_threshold_percentile: 1, dynthres_mimic_mode: Half Cosine Down, dynthres_mimic_scale_min: 1, dynthres_cfg_mode: Half Cosine Down, dynthres_cfg_scale_min: 3, dynthres_sched_val: 1, dynthres_separate_feature_channels: enable, dynthres_scaling_startpoint: ZERO, dynthres_variability_measure: STD, dynthres_interpolate_phi: 1FAQ
Comments (6)
I see 2 different models trained by you. Masterpiece and Complete, which do you prefer or recommend to use?
I think masterpiece might be better for character focused stuff and the complete is a full picture character and background included.
Hello, would like to recommend the Complete version, as it aims combine the quality and concept knowledge of all my rated datasets.
It has a much stronger effect due to the size and scale of the training, but has some issues to be resolved.
(~6000 images, any of which could create problems if mis-tagged or of low quality).
As an improvement to any generation, I might prefer this relatively much smaller (~300 image) Masterpieces LoRA at the moment.
It is a dataset of exclusively images I'd manually rated as the best possible after all.
@HaloSkull I think the masterpieces dataset would have a higher proportion of images with detailed backgrounds and compositions by comparison
Thank you for your reply and your nice work🥰!@motimalu
@motimalu I see, Thanks!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.