| --- |
| license: openrail++ |
| task_categories: |
| - unconditional-image-generation |
| language: |
| - en |
| tags: |
| - art |
| pretty_name: n/a |
| size_categories: |
| - 1K<n<10K |
| --- |
| |
| * Ars AI Experimental Description Language (AAEDL) |
| * Protocol : AAEP-0001 |
| * Version : 1.0 |
| --- |
|
|
| # Description |
| * Merge 'pipeline' reconstructed from surviving notes. |
| * This document describes the conception of Ars Divina. |
| --- |
|
|
| # Runtime Environment |
| * "AUTOMATIC1111 WebUI" |
| * "SuperMerger" |
| --- |
|
|
| # Repository |
| * "Animagine XL 3.0" |
| * "Pony Diffusion v6 (Base)" |
| * "SDXL v1.0 Base" |
| * "Snow Pony" |
| * "IPS Mix" |
| * "Illustrious XL v0.1" |
| * "TitiMix Illustrious IL" |
| --- |
|
|
| * Enhancement Library |
| * "Anti Long Torso XL v0.1" |
| * "SDS_Contrast tool_XL v1.0" |
| * "SPO-SDXL_4k-p_10ep_LoRA_webui" |
| --- |
|
|
| # Merge Profiles |
| # MBW_PROFILE_01 |
| * model_a [0.0,0.0,0.0,0.0,0.0,0.0,0.1,0.3,0.4,0.4,0.4,0.4,0.4,0.3,0.1,0.0,0.0,0.0,0.0,0.0] |
| * model_b [1.0,1.0,1.0,1.0,1.0,1.0,0.9,0.7,0.6,0.6,0.6,0.6,0.6,0.7,0.9,1.0,1.0,1.0,1.0,1.0] |
| * elemental ["BASE:conditioner.embedders.0:0.25,BASE:conditioner.embedders.1:0.35,BASE::0"] |
|
|
| # MBW_PROFILE_02 |
| * model_a [0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.3,0.4,0.4,0.4,0.4,0.4,0.4,0.4,0.3,0.2,0.0,0.0,0.0] |
| * model_b [1.0,1.0,1.0,1.0,1.0,1.0,1.0,0.7,0.6,0.6,0.6,0.6,0.6,0.6,0.6,0.7,0.8,1.0,1.0,1.0] |
| * elemental [BASE:conditioner.embedders.0:0.20,BASE:conditioner.embedders.1:0.30,BASE::0"] |
|
|
| # MBW_PROFILE_03 |
| * model_a [0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.4,0.4,0.4,0.3,0.4,0.4,0.4,0.4,0.0,0.0,0.0,0.0,0.0] |
| * model_b [1.0,1.0,1.0,1.0,1.0,1.0,1.0,0.6,0.6,0.6,0.7,0.6,0.6,0.6,0.6,1.0,1.0,1.0,1.0,1.0] |
| * elemental ["BASE" = 0.00] |
| --- |
|
|
| # Workflow #1 |
| * Phase "Foundation Construction" |
| * Experiment: EXP_001 |
| * merge |
| * model_a: "Animagine XL" |
| * weight: 0.45 |
| * model_b: "Pony Diffusion v6" |
| * weight: 0.55 |
| * method: WeightSum.CosineB |
| |
| # Workflow #2 |
| * Experiment: EXP_002 |
| * merge |
| * model_a: "EXP_001" |
| * weight: 0.45 |
| * model_b: "SDXL v1.0 base" |
| * weight: 0.55 |
| * method: AddDifference.TrainDifference |
| |
| # Workflow #3 |
| * Experiment: EXP_003 |
| * merge |
| * model_a: "EXP_002" |
| * weight: 0.65 |
| * model_b: "Snow Pony" |
| * weight: 0.35 |
| * method: AddDifference.TrainDifference |
| |
| # Workflow #4 |
| * Experiment: EXP_004 |
| * merge |
| * model_a: "EXP_003" |
| * weight: 0.45 |
| * model_b: "EXP_002" |
| * weight: 0.55 |
| * method: AddDifference.TrainDifference |
|
|
| # Workflow #5 |
| * Experiment: EXP_005 |
| * merge |
| * model_a: "EXP_004" |
| * weight: 0.95 |
| * model_b: "IPS Mix" |
| * weight: 0.05 |
| * method: WeightSum.CosineA |
|
|
| # Workflow #6 |
| * Phase "Cross-Architecture Integration" |
| * Experiment: EXP_006 |
| * merge |
| * model_a: "EXP_005" |
| * weight: |
| * model_b: "Illustrious XL v0.1" |
| * weight: |
| * method: WeightSum.Normal, MBW |
| * profile: MBW_Profile_01 |
|
|
| # Workflow #7 |
| * Experiment: EXP_007 |
| * merge |
| * model_a: "EXP_006" |
| * weight: |
| * model_b: "EXP_005" |
| * weight: |
| * method: WeightSum.CosineA, MBW |
| * profile: MBW_Profile_01 |
| |
| # Workflow #8 |
| * Experiment: EXP_008 |
| * merge |
| * model_a: "EXP_007" |
| * weight: |
| * model_b: "TitiMix Illustrious IL" |
| * weight: |
| * method: AddDifference.TrainDifference, MBW |
| * profile: MBW_Profile_01 |
| |
| # Workflow #9 |
| * Experiment: EXP_009 |
| * merge |
| * model_a: "EXP_008" |
| * weight: |
| * model_b: "EXP_007" |
| * weight: |
| * method: WeightSum.CosineA |
| * profile: MBW_PROFILE_02 |
|
|
| # Workflow #10 |
| * phase: "LoRA Enhancement" |
| * Experiment: EXP_010 |
| * bake |
| * source: "EXP_009" |
| * lora: "Anti Long Torso XL v0.1" |
| * strength: 4.0 |
|
|
| # Workflow #11 |
| * Experiment: EXP_011 |
| * merge |
| * model_a: "EXP_010" |
| * weight: |
| * model_b: "EXP_009" |
| * weight: |
| * method: WeightSum.Normal, MBW |
| * profile: MBW_Profile_02 |
| |
| # Workflow #12 |
| * Experiment: EXP_012 |
| * bake |
| * source: "EXP_011" |
| * lora: "SDS_Contrast_Tool_XL v1.0" |
| * strength: 0.25 |
|
|
| # Workflow #13 |
| * Experiment: EXP_013 |
| * bake |
| * source: "EXP_012" |
| * lora: "SPO_SDXL_4k-p_10ep_LoRA_webui" |
| * strength: 1.0 |
| |
| # Workflow #14 |
| * phase: "Final Consolidation" |
| * Experiment: EXP_014 |
| * merge |
| * model_a: "EXP_013" |
| * weight: |
| * model_b: "EXP_011" |
| * weight: |
| * method: AddDifference.TrainDifference, MBW |
| * profile: MBW_Profile_02 |
|
|
| # Workflow #15 |
| * Experiment: EXP_015 |
| * merge |
| * model_a: "EXP_014" |
| * weight: |
| * model_b: "EXP_013" |
| * weight: |
| * method: WeightSum.Normal, MBW |
| * profile: MBW_Profile_03 |
| |
| # Published Checkpoint |
| * Name: "Ars Divina" |
| * Source: "EXP_015" |
| * Version: "7.2.4a" |
| * Laboratory: "Ars AI" |
| * Status: "Released" |
| * Author: "NeverWinter13" |
|
|
| # Evaluation |
| * after each experiment, inference was made via x/y/z plot with a static seed of "1111" |
| * qualitative reviews were made before release. |
|
|
| # FAQs |
| * Q: Why does it look confusing and time-consuming? |
| * A: It is. Plain and simple. |
|
|
| * Q: How long did it took? |
| * A: Fairly long, tbh. |
