* Version 5 Update:
Ether Blu Mix 5 looks to further enhance the vibrant, painterly look with my first attempt at LORA mixing. There's also a lot more Counterfeit 3 allowing for more compositional variety.
EBM 4 did a lot of things right; as a base it created detailed, vibrant images reminiscent of a classic anime style thanks to YOZORA. Though, personally, it seemed too biased at depicting characters at a somewhat close distance, typically creating waist-high portraits. Adding more weight to Counterfeit 3 should hopefully add more compositional variety, while slightly reducing YOZORA and Color Box's weights to maintain EBM 4's style. Basil Mix was added as a supplementary realistic model in tandem with AyoniMix. It's a personal favorite that plays nice with anime (booru tag) style models.
EBM 5 is my first attempt at LORA mixing. I was initially hesitant at merging LORA's to checkpoints as they might introduce unwanted bias, since LORA's tend to be trained on small datasets. After plenty of testing, as well as learning from Luna's documentation of Kenshi and Aurora, I applied a light mix of LORA's to help enhance the vibrant, painterly style, while also improving background expression.
Your prompts might react differently as the base of the model is no longer just AOM3, so please experiment and have fun!
📘 Model Merging Info - v5
I have no training or real knowledge of machine learning. I merged the model using the information provided on WarriorMama777's AOM model page. I very arbitrarily used the UBM weights given to merge and create AOM3 and its derivatives and modified the values with trial and error until I was visually happy with it.
I utilized the SuperMerger extension for AUTO1111 by hako-mikan.
â–¼ Merging
MODEL A: AOM3.safetensors
MODEL B: Counterfeit-V3.0_fp16.safetensors
Add Sum merge - Don't Use MBW
Alpha: 0.5
Model: AOM3-Counter3Sum-05
â–¼
MODEL A: AOM3-Counter3Sum-05
MODEL B: nightSkyYOZORAStyle_yozoraV1PurnedFp16
Add Sum merge - Don't Use MBW
Alpha: 0.4
Model: AOM3-Counter3Sum-05-YozoraSum-04
â–¼
MODEL A: AOM3-Counter3Sum-05-YozoraSum-04
MODEL B: colorBoxModel_colorBOX.safetensors
MODEL C: AOM3-Counter3Sum-05
Add Difference merge - Don't Use MBW
Alpha: 0.2
Model: AOM3-Counter3Sum-05-YozoraSum-04-ColorBoxDiff-02
â–¼
MODEL A: AOM3-Counter3Sum-05-YozoraSum-04-ColorBoxDiff-02
MODEL B: pastelmix-better-vae-fp16.safetensors
MODEL C: Basil_mix_fixed.safetensors
Weight Sum merge - Use MBW
Alpha: 1
0,0.2,0.2,0.2,0.2,0.2,0,0,0,0,0.2,0.1,0.2,0.2,0.2,0.2,0.2,0.1,0.1,0.1,0.2,0.2,0.1,0.2,0.2,0.2
Beta: 1
0,0.2,0.2,0.2,0.2,0.2,0,0,0,0,0.2,0.1,0.2,0.2,0.2,0.2,0.2,0.1,0.1,0.1,0.2,0.2,0.1,0.2,0.2,0.2
Model: AOM3-Counter3Sum-05-YozoraSum-04-ColorBoxDiff-02-PastelSumS1-02-BasilSumS1-02
â–¼
MODEL A: AOM3-Counter3Sum-05-YozoraSum-04-ColorBoxDiff-02-PastelSumS1-02-BasilSumS1-02
MODEL B: ayonimix_V5.safetensors
Weight Sum merge - Use MBW
Alpha: 1
0,0.2,0.2,0.2,0.2,0.2,0,0,0,0,0.2,0.1,0.2,0.2,0.2,0.2,0.2,0.1,0.1,0.1,0.2,0.2,0.1,0.2,0.2,0.2
Final Model: EtherBluMix-5
â–¼ LORA Mix
xlimo768: 0.25
dpep4-768-000004: 0.3
sttabi_v1.4-04: 0.05
Jordan_3: 0.2
lightingv2: 0.4
Models Used:
*AOM3.safetensors
D124FC18F0232D7F0A2A70358CDB1288AF9E1EE8596200F50F0936BE59514F6D
*Counterfeit-V3.0_fp16.safetensors
CBFBA64E662370F59D4AA2AA69BF16749FCE93846CCCE20506AEE5DF01169859
*pastelmix-better-vae-fp16.safetensors
D01A68AE76F97506363F387F5F28BB564AD9E20924844FD5945E600B72D39E79
*colorBoxModel_colorBOX.safetensors
93A20525F51798309AF22AC6E8285E61CF563A88F1EE00AA9F544CA939E63181
*nightSkyYOZORAStyle_yozoraV1PurnedFp16
E7BF829CFFD9EA034DA3B59FA13421F52AA4E4ED74FCBBEEFBF9CFB2F97C329E
*ayonimix_V5.safetensors
B46B04432B070FD88D69161B62C43EDCD4340DFC27FFB2E5FE7EDA64950C7E34
*Basil_mix_fixed.safetensors
0FF127093F5BE455057742C40CEF578407B6933F240EE8DC5ED0F3061196FB38