OrangeMixs
"OrangeMixs" shares various Merge models that can be used with StableDiffusionWebui:Automatic1111 and others.
Maintain a repository for the following purposes.
- to provide easy access to models commonly used in the Japanese community.The Wisdom of the Anonsπ
- As a place to upload my merge models when I feel like it.
Hero image prompts(AOM3B2):https://majinai.art/ja/i/jhw20Z_
UPDATE NOTE / How to read this README
How to read this README
- Read the ToC as release notes.
Sections are in descending order. The order within the section is ascending. It is written like SNS. - UPDATE NOTE
- View the repository history when you need to check the full history.
UPDATE NOTE
- 2023-02-27: Add AOM3A1B
- 2023-03-10: Model name fix I found that I abbreviated the model name too much, so that when users see illustrations using OrangeMixs models on the web, they cannot reach them in their searches. To make the specification more search engine friendly, I renamed it to "ModelName + (orangemixs)".
- 2023-03-11: Change model name : () to _ Changed to _ because an error occurs when using () in the Cloud environment(e.g.:paperspace). "ModelName + _orangemixs"
- 2023-04-01: Added description of AOM3A1 cursed by Dreamlike
- 2023-06-27: Added AOM3B2. Removed Terms of Service.
- 2023-11-25: Add VividOrangeMix (nonlabel, NSFW, Hard)
- 2023-06-27: Added AOM3B2. Removed Terms of Service.
- 2023-11-25: Add VividOrangeMix (nonlabel, NSFW, Hard)
- 2024-01-07: Fix repo & Done upload VividOrangeMixs
Gradio
We support a Gradio Web UI to run OrangeMixs:
Table of Contents
- OrangeMixs
- UPDATE NOTE / How to read this README
- Gradio
- Table of Contents
- Reference
- Licence
Terms of use- Disclaimer
- How to download
- Model Detail & Merge Recipes
Reference
+/hdg/ Stable Diffusion Models Cookbook - https://rentry.org/hdgrecipes#g-anons-unnamed-mix-e93c3bf7 Model names are named after Cookbook precedentsπ
Licence
This model is open access and available to all, with a CreativeML OpenRAIL-M license further specifying rights and usage. The CreativeML OpenRAIL License specifies:
- You can't use the model to deliberately produce nor share illegal or harmful outputs or content
- The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license
- You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully) Please read the full license here οΌhttps://huggingface.co./spaces/CompVis/stable-diffusion-license
Terms of use
- Clearly indicate where modifications have been made.
If you used it for merging, please state what steps you took to do so.
Removed terms of use. 2023-06-28
Freedom. If you share your recipes, Marge swamp will be fun.
Disclaimer
READ MORE: Disclaimer
The user has complete control over whether or not to generate NSFW content, and the user's decision to enjoy either SFW or NSFW is entirely up to the user.The learning model does not contain any obscene visual content that can be viewed with a single click.The posting of the Learning Model is not intended to display obscene material in a public place. In publishing examples of the generation of copyrighted characters, I consider the following cases to be exceptional cases in which unauthorised use is permitted. "when the use is for private use or research purposes; when the work is used as material for merchandising (however, this does not apply when the main use of the work is to be merchandised); when the work is used in criticism, commentary or news reporting; when the work is used as a parody or derivative work to demonstrate originality." In these cases, use against the will of the copyright holder or use for unjustified gain should still be avoided, and if a complaint is lodged by the copyright holder, it is guaranteed that the publication will be stopped as soon as possible. I would also like to note that I am aware of the fact that many of the merged models use NAI, which is learned from Danbooru and other sites that could be interpreted as illegal, and whose model data itself is also a leak, and that this should be watched carefully. I believe that the best we can do is to expand the possibilities of GenerativeAI while protecting the works of illustrators and artists.How to download
Batch Download
β Deprecated: Orange has grown too huge. Doing this will kill your storage.
- install Git
- create a folder of your choice and right click β "Git bash here" and open a gitbash on the folder's directory.
- run the following commands in order.
git lfs install
git clone https://huggingface.co./WarriorMama777/OrangeMixs
- complete
Batch Download (Advanced)
Advanced: (When you want to download only selected directories, not the entire repository.)
Toggle: How to Batch Download (Advanced)
- Run the command
git clone --filter=tree:0 --no-checkout https://huggingface.co./WarriorMama777/OrangeMixs
to clone the huggingface repository. By adding the--filter=tree:0
and--no-checkout
options, you can download only the file names without their contents.
git clone --filter=tree:0 --no-checkout https://huggingface.co./WarriorMama777/OrangeMixs
- Move to the cloned directory with the command
cd OrangeMixs
.
cd OrangeMixs
- Enable sparse-checkout mode with the command
git sparse-checkout init --cone
. By adding the--cone
option, you can achieve faster performance.
git sparse-checkout init --cone
- Specify the directory you want to get with the command
git sparse-checkout add <directory name>
. For example, if you want to get only theModels/AbyssOrangeMix3
directory, entergit sparse-checkout add Models/AbyssOrangeMix3
.
git sparse-checkout add Models/AbyssOrangeMix3
- Download the contents of the specified directory with the command
git checkout main
.
git checkout main
This completes how to clone only a specific directory. If you want to add other directories, run git sparse-checkout add <directory name>
again.
Select and download
- Go to the Files and vaersions tab.
- select the model you want to download
- download
- complete
Model Detail & Merge Recipes
VividOrangeMix (VOM)
Prompt: https://majinai.art/ja/i/VZ9dNoI
Civitai: https://civitai.com/models/196585?modelVersionId=221033
2023-11-25
VividOrangeMix
βΌAbout
"VividOrangeMix is a StableDiffusion model created for fans seeking vivid, flat, anime-style illustrations. With rich, bold colors and flat shading, it embodies the style seen in anime and manga.β
One of the versions of OrangeMixs, AbyssOrangeMix1~3 (AOM), has improved the anatomical accuracy of the human body by merging photorealistic models, but I was dissatisfied with the too-realistic shapes and shadows.
VividOrangeMix is a model that has been adjusted to solve this problem.
βΌSample Gallery Default LoRA
VividOrangeMix_NSFW / Hard
βΌAbout VividOrangeMix NSFW/Hard is, as before, a model that Merges elements of NAI and Gape by U-Net Blocks Weight method. As of AOM3, elements of these models should be included, but when I simply merged other models, the elements of the old merge seem to gradually fade away. Also, by merging U-Net Blocks Weight, it is now possible to merge without affecting the design to some extent, but some changes are unavoidable, so I decided to upload it separately as before. .
βΌSample Gallery
Instructions
βΌTool
βΌVividOrangeMix
STEP: 1 | Base model create
STEP: 2 | Model merge
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3B4 | Animelike_2D_Pruend_fp16 | sum @ 0.3 | VividOrangeMix |
βΌVividOrangeMix_NSFW
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
VividOrangeMix | NAI full | NAI sfw | Add Difference @ 1.0 | 0,0.25,0.25,0.25,0.25,0.25,0,0,0,0,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.2,0.25,0.25,0.25,0.25,0,0 | VividOrangeMix_NSFW |
βΌVividOrangeMix_Hard
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
VividOrangeMix_NSFW | gape60 | NAI full | Add Difference @ 1.0 | 0.0,0.25,0.25,0.25,0.25,0.25,0.0,0.0,0.0,0.0,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.25,0.0,0.0 | VividOrangeMix_Hard |
AbyssOrangeMix3 (AOM3)
ββEveryone has different βABYSSβ!
βΌAbout
The main model, "AOM3 (AbyssOrangeMix3)", is a purely upgraded model that improves on the problems of the previous version, "AOM2". "AOM3" can generate illustrations with very realistic textures and can generate a wide variety of content. There are also three variant models based on the AOM3 that have been adjusted to a unique illustration style. These models will help you to express your ideas more clearly.
βΌLinks
- [β NSFW] Civitai: AbyssOrangeMix3 (AOM3) | Stable Diffusion Checkpoint | https://civitai.com/models/9942/abyssorangemix3-aom3
About
Features: high-quality, realistic textured illustrations can be generated.
There are two major changes from AOM2.
1: Models for NSFW such as _nsfw and _hard have been improved: the models after nsfw in AOM2 generated creepy realistic faces, muscles and ribs when using Hires.fix, even though they were animated characters. These have all been improved in AOM3.
e.g.: explanatory diagram by MEME : GO TO MEME ZONEβ
2: sfw/nsfw merged into one model. Originally, nsfw models were separated because adding NSFW content (models like NAI and gape) would change the face and cause the aforementioned problems. Now that those have been improved, the models can be packed into one.
In addition, thanks to excellent extensions such as ModelToolkit, the model file size could be reduced (1.98 GB per model).
More feature
In addition, these U-Net Blocks Weight Merge models take numerous steps but are carefully merged to ensure that mutual content is not overwritten.
(Of course, all models allow full control over adult content.)
- π When generating illustrations for the general public: write "nsfw" in the negative prompt field
- π
When generating adult illustrations: "nsfw" in the positive prompt field-> It can be generated without putting it in. If you include it, the atmosphere will be more NSFW.
Variations / Sample Gallery
π§Editingπ§
AOM3
(Actually, this gallery doesn't make much sense since AOM3 is mainly an improvement of the NSFW part π ...But we can confirm that the picture is not much different from AOM2sfw.)
AOM3A1
βOnly this model (AOM3A1) includes ChilloutMix. The curse of the DreamLike license. In other words, only AOM3A1 is not available for commercial use. I recommend AOM3A1B instead.β GO TO MEME ZONEβ
Features: Anime like illustrations with flat paint. Cute enough as it is, but I really like to apply LoRA of anime characters to this model to generate high quality anime illustrations like a frame from a theatre version.
βΌA1
Β©
(1)Β©Yurucamp: Inuyama Aoi, (2)Β©The Quintessential Quintuplets: Nakano Yotsuba, (3)Β©Sailor Moon: Mizuno Ami/SailorMercuryAOM3A2
π§Editingπ§ Features: Oil paintings like style artistic illustrations and stylish background depictions. In fact, this is mostly due to the work of Counterfeit 2.5, but the textures are more realistic thanks to the U-Net Blocks Weight Merge.
AOM3A3
π§Editingπ§ Features: Midpoint of artistic and kawaii. the model has been tuned to combine realistic textures, a artistic style that also feels like an oil colour style, and a cute anime-style face. Can be used to create a wide range of illustrations.
AOM3A1B
AOM3A1B added. This model is my latest favorite. I recommend it for its moderate realism, moderate brush touch, and moderate LoRA conformity.
The model was merged by mistakenly selecting 'Add sum' when 'Add differences' should have been selected in the AOM3A3AOM3A2 recipe. It was an unintended merge, but we share it because the illustrations produced are consistently good results.
The model was merged by mistakenly selecting 'Add sum' when 'Add differences' should have been selected in the AOM3A3AOM3A2 recipe. It was an unintended merge, but we share it because the illustrations produced are consistently good results.
In my review, this is an illustration style somewhere between AOM3A1 and A3.
βΌA1B
- Meisho Doto (umamusume): https://civitai.com/models/11980/meisho-doto-umamusume
- Train and Girl: JR East E235 series / train interior
Β©
Β©umamusume: Meisho Doto, Β©Girls und Panzer: Nishizumi Miho,Β©IDOLM@STER: Sagisawa FumikaAOM3B2
my newest toy.
Just AOM3A1B + BreakdomainM21: 0.4
So this model is somewhat of a troll model.
I would like to create an improved DiffLoRAKit_v2 based on this.
Upload for access for research etc. 2023-06-27
Sample image prompts
- Maid
- Yotsuba: https://majinai.art/ja/i/f-O4wau
- Inuko in cafe: https://majinai.art/ja/i/Cj-Ar9C
- bathroom: https://majinai.art/ja/i/XiSj5K6
AOM3B3
2023-09-25
This is a derivative model of AOM3B2. I merged some nice models and also merged some LoRAs to further adjust the color and painting style.
βInstructions:
βΌTool Supermerger
βΌModel Merge AOM3B2+Mixprov4+BreakdomainAnime triple sum : 0.3, 0.3 | mode:normal
οΌ
βΌLoRA Merge loraH(DiffLoRA)_FaceShadowTweaker_v1_dim4:-2,nijipretty_20230624235607:0.1,MatureFemale_epoch8:0.1,colorful_V1_lbw:0.5
AOM3B4
βΌInstructions:
USE: https://github.com/hako-mikan/sd-webui-supermerger/
STEP: 1 | Model merge
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3B2 | Mixprov4 | BreakdomainAnime | triple sum @ 0.3, 0.3, mode:normal | temp01 |
STEP: 2 | LoRA Merge
Color fix
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
temp01 | colorful_V1_lbw | sum @ 0.45 | AOM3B4 |
βGO TO VividOrangeMix Instructionsβ
AOM3B3
2023-09-25
This is a derivative model of AOM3B2. I merged some nice models and also merged some LoRAs to further adjust the color and painting style.
βInstructions:
βΌTool Supermerger
βΌModel Merge AOM3B2+Mixprov4+BreakdomainAnime triple sum : 0.3, 0.3 | mode:normal
οΌ
βΌLoRA Merge loraH(DiffLoRA)_FaceShadowTweaker_v1_dim4:-2,nijipretty_20230624235607:0.1,MatureFemale_epoch8:0.1,colorful_V1_lbw:0.5
AOM3B4
βΌInstructions:
USE: https://github.com/hako-mikan/sd-webui-supermerger/
STEP: 1 | Model merge
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3B2 | Mixprov4 | BreakdomainAnime | triple sum @ 0.3, 0.3, mode:normal | temp01 |
STEP: 2 | LoRA Merge
Color fix
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
temp01 | colorful_V1_lbw | sum @ 0.45 | AOM3B4 |
βGO TO VividOrangeMix Instructionsβ
Description for enthusiast
AOM3 was created with a focus on improving the nsfw version of AOM2, as mentioned above.The AOM3 is a merge of the following two models into AOM2sfw using U-Net Blocks Weight Merge, while extracting only the NSFW content part.
(1) NAI: trained in Danbooru
(2)gape: Finetune model of NAI trained on Danbooru's very hardcore NSFW content.
In other words, if you are looking for something like AOM3sfw, it is AOM2sfw.The AOM3 was merged with the NSFW model while removing only the layers that have a negative impact on the face and body. However, the faces and compositions are not an exact match to AOM2sfw.AOM2sfw is sometimes superior when generating SFW content. I recommend choosing according to the intended use of the illustration.See below for a comparison between AOM2sfw and AOM3.
βΌA summary of the AOM3 work is as follows
- investigated the impact of the NAI and gape layers as AOM2 _nsfw onwards is crap.
- cut face layer: OUT04 because I want realistic faces to stop β Failed. No change.
- gapeNAI layer investigationο½
a. (IN05-08 (especially IN07) | Change the illustration significantly. Noise is applied, natural colours are lost, shadows die, and we can see that the IN deep layer is a layer of light and shade.
b. OUT03-05(?) | likely to be sexual section/NSFW layer.Cutting here will kill the NSFW.
c. OUT03,OUT04ο½NSFW effects are in(?). e.g.: spoken hearts, trembling, motion lines, etc...
d. OUT05ο½This is really an NSFW switch. All the "NSFW atmosphere" is in here. Facial expressions, Heavy breaths, etc...
e. OUT10-11ο½Paint layer. Does not affect detail, but does have an extensive impact. - (mass production of rubbish from here...)
- cut IN05-08 and merge NAIgape with flat parameters β avoided creepy muscles and real faces. Also, merging NSFW models stronger has less impact.
- so, cut IN05-08, OUT10-11 and merge NAI+gape with all others 0.5.
- β AOM3
AOM3 roughly looks like this
βΌHow to use
- Prompts
- Negative prompts is As simple as possible is good.
(worst quality, low quality:1.4) - Using "3D" as a negative will result in a rough sketch style at the "sketch" level. Use with caution as it is a very strong prompt.
- How to avoid Real Face
(realistic, lip, nose, tooth, rouge, lipstick, eyeshadow:1.0), (abs, muscular, rib:1.0), - How to avoid Bokeh
(depth of field, bokeh, blurry:1.4) - How to remove mosaic:
(censored, mosaic censoring, bar censor, convenient censoring, pointless censoring:1.0),
- How to remove blush:
(blush, embarrassed, nose blush, light blush, full-face blush:1.4),
- How to remove NSFW effects:
(trembling, motion lines, motion blur, emphasis lines:1.2),
- π°Basic negative prompts sample for Anime girl β
- v1
nsfw, (worst quality, low quality:1.4), (realistic, lip, nose, tooth, rouge, lipstick, eyeshadow:1.0), (dusty sunbeams:1.0),, (abs, muscular, rib:1.0), (depth of field, bokeh, blurry:1.4),(motion lines, motion blur:1.4), (greyscale, monochrome:1.0), text, title, logo, signature
- v2
nsfw, (worst quality, low quality:1.4), (lip, nose, tooth, rouge, lipstick, eyeshadow:1.4), (blush:1.2), (jpeg artifacts:1.4), (depth of field, bokeh, blurry, film grain, chromatic aberration, lens flare:1.0), (1boy, abs, muscular, rib:1.0), greyscale, monochrome, dusty sunbeams, trembling, motion lines, motion blur, emphasis lines, text, title, logo, signature,
- v1
- Negative prompts is As simple as possible is good.
- Sampler:
βDPM++ SDE Karrasβ is goodTake your pick - Steps:
- DPM++ SDE Karras: Test: 12ο½ ,illustration: 20ο½
- DPM++ 2M Karras: Test: 20ο½ ,illustration: 28ο½
- Clipskip: 1 or 2
- CFG: 8 (6ο½12)
- Upscaler :
- Detailed illust β Latenet (nearest-exact)
Denoise strength: 0.5 (0.5~0.6) - Simple upscale: Swin IR, ESRGAN, Remacri etcβ¦
Denoise strength: Can be set low. (0.35~0.6)
- Detailed illust β Latenet (nearest-exact)
π©βπ³Model details / Recipe
βΌHash(SHA256) βΌHash(SHA256)
AOM3.safetensors
D124FC18F0232D7F0A2A70358CDB1288AF9E1EE8596200F50F0936BE59514F6DAOM3A1.safetensors
F303D108122DDD43A34C160BD46DBB08CB0E088E979ACDA0BF168A7A1F5820E0AOM3A2.safetensors
553398964F9277A104DA840A930794AC5634FC442E6791E5D7E72B82B3BB88C3AOM3A3.safetensors
EB4099BA9CD5E69AB526FCA22A2E967F286F8512D9509B735C892FA6468767CFAOM3A1B.safetensors 5493A0EC491F5961DBDC1C861404088A6AE9BD4007F6A3A7C5DEE8789CDC1361
AOM3B2.safetensors F553E7BDE46CFE9B3EF1F31998703A640AF7C047B65883996E44AC7156F8C1DB
AOM3A1B.safetensors 5493A0EC491F5961DBDC1C861404088A6AE9BD4007F6A3A7C5DEE8789CDC1361
AOM3B2.safetensors F553E7BDE46CFE9B3EF1F31998703A640AF7C047B65883996E44AC7156F8C1DB
βΌUse Models
- AOM2sfw
γ038ba203d8ba3c8af24f14e01fbb870c85bbb8d4b6d9520804828f4193d12ce9γ - AnythingV3.0 huggingface pruned
[2700c435]γ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eγ - NovelAI animefull-final-pruned
[925997e9]γ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8γ - NovelAI sfw
[1d4a34af]γ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaγ - Gape60
[25396b85]γ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5γ - BasilMix
γbbf07e3a1c3482c138d096f7dcdb4581a2aa573b74a68ba0906c7b657942f1c2γ - chilloutmix_fp16.safetensors
γ4b3bf0860b7f372481d0b6ac306fed43b0635caf8aa788e28b32377675ce7630γ - Counterfeit-V2.5_fp16.safetensors
γ71e703a0fca0e284dd9868bca3ce63c64084db1f0d68835f0a31e1f4e5b7cca6γ - kenshi_01_fp16.safetensors
γ3b3982f3aaeaa8af3639a19001067905e146179b6cddf2e3b34a474a0acae7faγ
βΌAOM3
βInstructions: βInstructions:
Tool: SuperMerger
USE: https://github.com/hako-mikan/sd-webui-supermerger/
Tool: SuperMerger
USE: https://github.com/hako-mikan/sd-webui-supermerger/
(This extension is really great. It turns a month's work into an hour. Thank you)
STEP: 1 | BWM : NAI - NAIsfw & gape - NAI
CUT: IN05-IN08, OUT10-11
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM2sfw | NAI full | NAI sfw | Add Difference @ 1.0 | 0,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0 | temp01 |
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
--- | --- | --- | --- | --- | --- |
AOM2sfw | NAI full | NAI sfw | Add Difference @ 1.0 | 0,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0 | temp01 |
CUT: IN05-IN08, OUT10-11
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
temp01 | gape60 | NAI full | Add Difference @ 1.0 | 0,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0 | AOM3 |
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
--- | --- | --- | --- | --- | --- |
temp01 | gape60 | NAI full | Add Difference @ 1.0 | 0,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0 | AOM3 |
βΌAOM3A1
βInstructions:
Tool: SuperMerger βInstructions:
Tool: SuperMerger
STEP: 1 | Change the base photorealistic model of AOM3 from BasilMix to Chilloutmix.
Change the photorealistic model from BasilMix to Chilloutmix and proceed to gapeNAI merge.
STEP: 2 |
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | SUM @ 0.5 | Counterfeit2.5 | Kenshi | Counterfeit+Kenshi | |
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
--- | --- | --- | --- | --- | --- |
1 | SUM @ 0.5 | Counterfeit2.5 | Kenshi | Counterfeit+Kenshi |
STEP: 3 |
CUT: BASE0, IN00-IN08οΌ0, IN10οΌ0.1, OUT03-04-05οΌ0, OUT08οΌ0.2
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit+Kenshi | Add SUM @ 1.0 | 0,0,0,0,0,0,0,0,0,0.3,0.1,0.3,0.3,0.3,0.2,0.1,0,0,0,0.3,0.3,0.2,0.3,0.4,0.5 | AOM3A1 |
βΌAOM3A1 βOnly this model (AOM3A1) includes ChilloutMix (=The curse of DreamLike).Commercial use is not available.
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit+Kenshi | Add SUM @ 1.0 | 0,0,0,0,0,0,0,0,0,0.3,0.1,0.3,0.3,0.3,0.2,0.1,0,0,0,0.3,0.3,0.2,0.3,0.4,0.5 | AOM3A1 |
βΌAOM3A1 βOnly this model (AOM3A1) includes ChilloutMix (=The curse of DreamLike).Commercial use is not available.
βΌAOM3A2
β? β?
CUT: BASE0, IN05:0.3γIN06-IN08οΌ0, IN10οΌ0.1, OUT03οΌ0, OUT04οΌ0.3, OUT05οΌ0, OUT08οΌ0.2
βInstructions: βInstructions:
Tool: SuperMerger Tool: SuperMerger
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit2.5 | nai | Add Difference @ 1.0 | 0,1,1,1,1,1,0.3,0,0,0,1,0.1,1,1,1,1,1,0,1,0,1,1,0.2,1,1,1 | AOM3A2 |
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
--- | --- | --- | --- | --- | --- |
AOM3 | Counterfeit2.5 | nai | Add Difference @ 1.0 | 0,1,1,1,1,1,0.3,0,0,0,1,0.1,1,1,1,1,1,0,1,0,1,1,0.2,1,1,1 | AOM3A2 |
βAOM3A3 βAOM3A3
CUT : BASE0, IN05-IN08οΌ0, IN10οΌ0.1, OUT03οΌ0.5, OUT04-05οΌ0.1, OUT08οΌ0.2
Tool: SuperMerger
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit2.5 | nai | Add Difference @ 1.0 | 0,0.6,0.6,0.6,0.6,0.6,0,0,0,0,0.6,0.1,0.6,0.6,0.6,0.6,0.6,0.5,0.1,0.1,0.6,0.6,0.2,0.6,0.6,0.6 | AOM3A3 |
βΌAOM3A1B
βInstructions:
Tool: SuperMerge
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit2.5 | Add Sum @ 1.0 | 0,1,1,1,1,1,0.3,0,0,0,1,0.1,1,1,1,1,1,0,1,0,1,1,0.2,1,1,1 | AOM3A1B |
βΌAOM3B2
βInstructions:
Tool: Checkpoint Merger
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3A1B | Breakdomain m21_fp16 | Add Sum | 0.4 | AOM3B2 |
Tool: SuperMerger
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit2.5 | nai | Add Difference @ 1.0 | 0,0.6,0.6,0.6,0.6,0.6,0,0,0,0,0.6,0.1,0.6,0.6,0.6,0.6,0.6,0.5,0.1,0.1,0.6,0.6,0.2,0.6,0.6,0.6 | AOM3A3 |
βΌAOM3A1B
βInstructions:
Tool: SuperMerge
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3 | Counterfeit2.5 | Add Sum @ 1.0 | 0,1,1,1,1,1,0.3,0,0,0,1,0.1,1,1,1,1,1,0,1,0,1,1,0.2,1,1,1 | AOM3A1B |
βΌAOM3B2
βInstructions:
Tool: Checkpoint Merger
Model: A | Model: B | Model: C | Interpolation Method | Weight | Merge Name |
---|---|---|---|---|---|
AOM3A1B | Breakdomain m21_fp16 | Add Sum | 0.4 | AOM3B2 |
AbyssOrangeMix2 (AOM2)
ββCreating the next generation of illustration with βAbyssβ!
Prompt: https://majinai.art/ja/i/nxpKRpw
βΌAbout
AbyssOrangeMix2 (AOM2) is an AI model capable of generating high-quality, highly realistic illustrations. It can generate elaborate and detailed illustrations that cannot be drawn by hand. It can also be used for a variety of purposes, making it extremely useful for design and artwork. Furthermore, it provides an unparalleled new means of expression. It can generate illustrations in a variety of genres to meet a wide range of needs. I encourage you to use "Abyss" to make your designs and artwork richer and of higher quality.
β»nvidia joke.βΌDescription for engineers/enthusiasts
The merged model was formulated using an extension such as sdweb-merge-block-weighted-gui, which merges models at separate rates for each of the 25 U-Net blocks (input, intermediate, and output). The validation of many Anons has shown that such a recipe can generate a painting style that is anatomically realistic enough to feel the finger skeleton, but still maintains an anime-style face.
The changes from AbyssOrangeMix are as follows.
- the model used for U-Net Blocks Weight Merge was changed from Instagram+F222 to BasilMix. (https://huggingface.co./nuigurumi)
This is an excellent merge model that can generate decent human bodies while maintaining the facial layers of the Instagram model. Thanks!!! This has improved the dullness of the color and given a more Japanese skin tone (or more precisely, the moisturized white skin that the Japanese would ideally like). Also, the unnatural bokeh that sometimes occurred in the previous version may have been eliminated (needs to be verified).
2.Added IN deep layers (IN06-11) to the layer merging from the realistic model (BasilMix).
It is said that the IN deep layer (IN06-11) is the layer that determines composition, etc., but perhaps light, reflections, skin texture, etc., may also be involved. It is like "Global Illumination", "Ray tracing" and "Ambient Occlusion" in 3DCG.
β»This does not fundamentally improve the fingers. Therefore, More research needs to be done to improve the fingers (e.g. 'bad_prompt'). About 30-50% chance of generating correct fingers(?). Abyss is deep.
βΌSample Gallery
The prompts for generating these images were all generated using ChatGPT. I simply asked "Pirates sailing the oceans" to tell me what the prompts were.
However, to make sure the AI understood the specifications, I used the template for AI questions (Question template for AI prompt generation(v1.2) ).
Please review the following.
https://seesaawiki.jp/nai_ch/d/AI%a4%f2%b3%e8%cd%d1%a4%b7%a4%bf%a5%d7%a5%ed%a5%f3%a5%d7%a5%c8%c0%b8%c0%ae
The images thus generated, strangely enough, look like MidJourney or Nijijourney illustrations. Perhaps they are passing user prompts through GPT or something else before passing them on to the image AIπ€
βΌREAD MOREπΌ
βΌAll prompts to generate sample images
βΌHow to use
- VAE: orangemix.vae.pt
Prompts can be long or short
As simple as possible is good. Do not add excessive detail prompts. Start with just this negative propmt.
(worst quality, low quality:1.4)- Sampler: βDPM++ SDE Karrasβ is good
- Steps: forTest: 12ο½ ,illustration: 20ο½
- Clipskip: 1 or 2
- Upscaler : Latenet (nearest-exact)
- CFG Scale : 5 or 6 (4ο½8)
- Denoise strength: 0.5 (0.45~0.6)
If you use 0.7ο½, the picture will change too much.
If below 0.45, Block noise occurs.
πModel List
- AbyssOrangeMix2_sfwο½BasilMix U-Net Blocks Weight Merge
- AbyssOrangeMix2_nsfwο½+ NAI-NAISFW 0.3 Merge
- AbyssOrangeMix2_hardο½+ Gape 0.3 Merge
- AbyssOrangeMix2_nsfwο½+ NAI-NAISFW 0.3 Merge
β»Changed suffix of models.
_base β_sfw: _base was changed to_sfw.
_night β_nsfw: Merged models up to NAI-NAI SFW were changed from _night to_nsfw.
_half and non suffix β_hard: Gape merged models were given the suffix _hard.gape was reduced to 0.3 because it affects character modeling.
βΌHow to choice models
- _sfw : SFWπ
- _nsfw : SFW ο½ Soft NSFWπ₯°
- _hard : SFW ο½ hard NSFWπ
βΌHash
- AbyssOrangeMix2_sfw.ckpt
γf75b19923f2a4a0e70f564476178eedd94e76e2c94f8fd8f80c548742b5b51b9γ - AbyssOrangeMix2_sfw.safetensors
γ038ba203d8ba3c8af24f14e01fbb870c85bbb8d4b6d9520804828f4193d12ce9γ - AbyssOrangeMix2_nsfw.safetensors
γ0873291ac5419eaa7a18726e8841ce0f15f701ace29e0183c47efad2018900a4γ - AbyssOrangeMix_hard.safetensors
γ0fc198c4908e98d7aae2a76bd78fa004e9c21cb0be7582e36008b4941169f18eγ
βΌUse Models
- AnythingV3.0 huggingface pruned
[2700c435]γ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eγ - NovelAI animefull-final-pruned
[925997e9]γ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8γ - NovelAI sfw
[1d4a34af]γ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaγ - Gape60
[25396b85]γ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5γ - BasilMix
γbbf07e3a1c3482c138d096f7dcdb4581a2aa573b74a68ba0906c7b657942f1c2γ
AbyssOrangeMix2_sfw (AOM2s)
βΌInstructions:
STEP: 1ο½Block Merge
Model: A | Model: B | Weight | Base alpha | Merge Name |
---|---|---|---|---|
AnythingV3.0 | BasilMix | 1,0.9,0.7,0.5,0.3,0.1,1,1,1,1,1,1,0,0,0,0,0,0,0,0.1,0.3,0.5,0.7,0.9,1 | 0 | AbyssOrangeMix2_sfw |
AbyssOrangeMix2_nsfw (AOM2n)
βΌ?
JUST AbyssOrangeMix2_sfw+ (NAI-NAISFW) 0.3.
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AbyssOrangeMix_base | NovelAI animefull | NovelAI sfw | AbyssOrangeMix2_nsfw |
AbyssOrangeMix2_hard (AOM2h)
βΌ? +Gape0.3 version AbyssOrangeMix2_nsfw.
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AbyssOrangeMix2_nsfw | Gape60 | NovelAI animefull | AbyssOrangeMix2_hard |
EerieOrangeMix (EOM)
EerieOrangeMix is the generic name for a U-Net Blocks Weight Merge Models based on Elysium(Anime V2).
Since there are infinite possibilities for U-Net Blocks Weight Merging, I plan to treat all Elysium-based models as a lineage of this model.
β»This does not fundamentally improve the fingers. Therefore, More research needs to be done to improve the fingers (e.g. 'bad_prompt').
EerieOrangeMix (EOM1)
βΌ?
This merge model is simply a U-Net Blocks Weight Merge of ElysiumAnime V2 with the AbyssOrangeMix method.
The AnythingModel is good at cute girls anyway, and no matter how hard I try, it doesn't seem to be good at women in their late 20s and beyond. Therefore, I created a U-Net Blocks Weight Merge model based on my personal favorite ElysiumAnime V2 model. ElyOrangeMix was originally my favorite, so this is an enhanced version of that.
πModel List
- EerieOrangeMix_baseο½Instagram+F222 U-Net Blocks Weight Merge
- EerieOrangeMix_nightο½+ NAI-NAISFW Merge
- EerieOrangeMix_halfο½+ Gape0.5 Merge
- EerieOrangeMixο½+ Gape1.0 Merge
- EerieOrangeMix_nightο½+ NAI-NAISFW Merge
βΌ How to choice models
- _base : SFWπ
- _Night : SFW ο½ Soft NSFWπ₯°
- _half : SFW ο½ NSFWπ
- unlabeled : SFW ο½ HARDCORE ο½π€― ex)AbyssOrangeMix, BloodOrangeMix...etc
βΌHash
- EerieOrangeMix.safetensors
- EerieOrangeMix_half.safetensors
- EerieOrangeMix_night.safetensors
- EerieOrangeMix_base.ckpt
βΌUse Models
[] = WebUI Hash,γγ= SHA256
- Elysium Anime V2 []γ5c4787ce1386500ee05dbb9d27c17273c7a78493535f2603321f40f6e0796851γ
- NovelAI animefull-final-pruned [925997e9]γ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8γ
- NovelAI sfw [1d4a34af]γ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaγ
- Gape60 [25396b85]γ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5γ
- instagram-latest-plus-clip-v6e1_50000.safetensors [] γ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5γ
- f222 [] γ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fγ
- sd1.5_pruned [] γe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053γ
βΌ Sample Gallery
MoreπΌ
βΌ How to use
- VAE: orangemix.vae.pt
- As simple as possible is good. Do not add excessive detail prompts. Start with just this. (worst quality, low quality:1.4)
- Sampler: βDPM++ SDE Karrasβ is good
- Steps: forTest: 20ο½24 ,illustration: 24ο½50
- Clipskip: 1
- USE βupscale latent spaceβ
- Denoise strength: 0.45 (0.4~0.5)
If you use 0.7ο½, the picture will change too much.
βΌPrompts
πWhen generating cute girls, try this negative prompt first. It avoids low quality, prevents blurring, avoids dull colors, and dictates Anime-like cute face modeling.
nsfw, (worst quality, low quality:1.3), (depth of field, blurry:1.2), (greyscale, monochrome:1.1), 3D face, nose, cropped, lowres, text, jpeg artifacts, signature, watermark, username, blurry, artist name, trademark, watermark, title, (tan, muscular, loli, petite, child, infant, toddlers, chibi, sd character:1.1), multiple view, Reference sheet,
EerieOrangeMix_base (EOM1b)
βΌ?
Details are omitted since it is the same as AbyssOrangeMix.
βΌInstructions:
STEP: 1ο½Creation of photorealistic model for Merge
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 1.0 | instagram-latest-plus-clip-v6e1_50000 | f222 | sd1.5_pruned | Insta_F222 |
STEP: 2ο½Block Merge
Merge InstaF222
Model: A | Model: B | Weight | Base alpha | Merge Name |
---|---|---|---|---|
Elysium Anime V2 | Insta_F222 | 1,0.9,0.7,0.5,0.3,0.1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.1,0.3,0.5,0.7,0.9,1 | 0 | Temp1 |
EerieOrangeMix_Night (EOM1n)
βΌ?
JUST EerieOrangeMix_base+ (NAI-NAISFW) 0.3.
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | EerieOrangeMix_base | NovelAI animefull | NovelAI sfw | EerieOrangeMix_Night |
EerieOrangeMix_half (EOM1h)
βΌ? +Gape0.5 version EerieOrangeMix.
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.5 | EerieOrangeMix_Night | NovelAI animefull | NovelAI sfw | EerieOrangeMix_half |
EerieOrangeMix (EOM1)
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 1.0 | EerieOrangeMix_Night | Gape60 | NovelAI animefull | EerieOrangeMix |
EerieOrangeMix2 (EOM2)
βΌ?
The model was created by adding the hierarchy responsible for detailing and painting ElysiumV1 to EerieOrangeMix_base, then merging NAI and Gape.
πModel List
- EerieOrangeMix2_baseο½Instagram+F222+ElysiumV1 U-Net Blocks Weight Merge
- EerieOrangeMix2_nightο½+ NAI-NAISFW Merge
- EerieOrangeMix2_halfο½+ Gape0.5 Merge
- EerieOrangeMix2ο½+ Gape1.0 Merge
- EerieOrangeMix2_nightο½+ NAI-NAISFW Merge
βΌ How to choice models
- _base : SFWπ
- _Night : SFW ο½ Soft NSFWπ₯°
- _half : SFW ο½ NSFWπ
- unlabeled : SFW ο½ HARDCORE ο½π€― ex)AbyssOrangeMix, BloodOrangeMix...etc
βΌHash
- EerieOrangeMix2.safetensors
- EerieOrangeMix2_half.safetensors
- EerieOrangeMix2_night.safetensors
- EerieOrangeMix2_base.ckpt
βΌUse Models
[] = webuHash,γγ= SHA256
- Elysium Anime V2 []γ5c4787ce1386500ee05dbb9d27c17273c7a78493535f2603321f40f6e0796851γ
- NovelAI animefull-final-pruned [925997e9]γ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8γ
- NovelAI sfw [1d4a34af]γ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaγ
- Gape60 [25396b85]γ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5γ
- instagram-latest-plus-clip-v6e1_50000.safetensors [] γ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5γ
- f222 [] γ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fγ
- sd1.5_pruned [] γe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053γ
- ElysiumV1 γabbb28cb5e70d3e0a635f241b8d61cefe42eb8f1be91fd1168bc3e52b0f09ae4γ
EerieOrangeMix2_base (EOM2b)
βΌ?
βΌInstructions
STEP: 1ο½Block Merge
Merge ElysiumV1
The generated results do not change much with or without this process, but I wanted to incorporate Elysium's depiction, so I merged it.
Model: A | Model: B | Weight | Base alpha | Merge Name |
---|---|---|---|---|
EerieOrangeMix_base | ElysiumV1 | 1,0.9,0.7,0.5,0.3,0.1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.1,0.3,0.5,0.7,0.9,1 | 0 | EerieOrangeMix2_base |
EerieOrangeMix2_night (EOM2n)
βΌ?
JUST EerieOrangeMix2_base+ (NAI-NAISFW) 0.3.
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | EerieOrangeMix_base | NovelAI animefull | NovelAI sfw | EerieOrangeMix2_Night |
EerieOrangeMix2_half (EOM2h)
βΌ? +Gape0.5 version EerieOrangeMix2.
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.5 | EerieOrangeMix_Night | NovelAI animefull | NovelAI sfw | EerieOrangeMix2_half |
EerieOrangeMix2 (EOM2)
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 1.0 | EerieOrangeMix_Night | Gape60 | NovelAI animefull | EerieOrangeMix2 |
Models Comparison
β»The difference is slight but probably looks like this. β warm color, β natural color, β animated colorAbyssOrangeMix (AOM)
ββHow can you guys take on such a deep swamp and get results?
Is it something like "Made in Abyss"?
By Anon, 115th thread
βΌ?
The merged model was formulated using an extension such as sdweb-merge-block-weighted-gui, which merges models at separate rates for each of the 25 U-Net blocks (input, intermediate, and output). The validation of many Anons has shown that such a recipe can generate a painting style that is anatomically realistic enough to feel the finger skeleton, but still maintains an anime-style face.
β»This model is the result of a great deal of testing and experimentation by many Anonsπ€
β»This model can be very difficult to handle. I am not 100% confident in my ability to use this model. It is peaky and for experts.
β»This does not fundamentally improve the fingers, and I recommend using bad_prompt, etc. (Embedding) in combination.
βΌSample Gallery
(1)
((masterpiece)), best quality, perfect anatomy, (1girl, solo focus:1.4), pov, looking at viewer, flower trim,(perspective, sideway, From directly above ,lying on water, open hand, palm, :1.3),(Accurate five-fingered hands, Reach out, hand focus, foot focus, Sole, heel, ball of the thumb:1.2), (outdoor, sunlight:1.2),(shiny skin:1.3),,(masterpiece, white border, outside border, frame:1.3),
, (motherhood, aged up, mature female, medium breasts:1.2), (curvy:1.1), (single side braid:1.2), (long hair with queue and braid, disheveled hair, hair scrunchie, tareme:1.2), (light Ivory hair:1.2), looking at viewer,, Calm, Slight smile,
,(anemic, dark, lake, river,puddle, Meadow, rock, stone, moss, cliff, white flower, stalactite, Godray, ruins, ancient, eternal, deep ,mystic background,sunlight,plant,lily,white flowers, Abyss, :1.2), (orange fruits, citrus fruit, citrus fruit bearing tree:1.4), volumetric lighting,good lighting,, masterpiece, best quality, highly detailed,extremely detailed cg unity 8k wallpaper,illustration,((beautiful detailed face)), best quality, (((hyper-detailed ))), high resolution illustration ,high quality, highres, sidelighting, ((illustrationbest)),highres,illustration, absurdres, hyper-detailed, intricate detail, perfect, high detailed eyes,perfect lighting, (extremely detailed CG:1.2),
Negative prompt: (bad_prompt_version2:1), distant view, lip, Pregnant, maternity, pointy ears, realistic, tan, muscular, greyscale, monochrome, lineart, 2koma, 3koma, 4koma, manga, 3D, 3Dcubism, pablo picasso, disney, marvel, mutanted breasts, mutanted nipple, cropped, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, lowres, trademark, watermark, title, text, deformed, bad anatomy, disfigured, mutated, extra limbs, ugly, missing limb, floating limbs, disconnected limbs, out of frame, mutated hands and fingers, poorly drawn hands, malformed hands, poorly drawn face, poorly drawn asymmetrical eyes, (blurry:1.4), duplicate (loli, petite, child, infant, toddlers, chibi, sd character, teen age:1.4), tsurime, helmet hair, evil smile, smug_face, naughty smile, multiple view, Reference sheet, (worst quality, low quality:1.4),
Steps: 24, Sampler: DPM++ SDE Karras, CFG scale: 10, Seed: 1159970659, Size: 1536x768, Model hash: cc44dbff, Model: AbyssOrangeMix, Variation seed: 93902374, Variation seed strength: 0.45, Denoising strength: 0.45, ENSD: 31337
(2)
street, 130mm f1.4 lens, ,(shiny skin:1.3),, (teen age, school uniform:1.2), (glasses, black hair, medium hair with queue and braid, disheveled hair, hair scrunchie, tareme:1.2), looking at viewer,, Calm, Slight smile,
Negative prompt: (bad_prompt_version2:1), distant view, lip, Pregnant, maternity, pointy ears, realistic, tan, muscular, greyscale, monochrome, lineart, 2koma, 3koma, 4koma, manga, 3D, 3Dcubism, pablo picasso, disney, marvel, mutanted breasts, mutanted nipple, cropped, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, lowres, trademark, watermark, title, text, deformed, bad anatomy, disfigured, mutated, extra limbs, ugly, missing limb, floating limbs, disconnected limbs, out of frame, mutated hands and fingers, poorly drawn hands, malformed hands, poorly drawn face, poorly drawn asymmetrical eyes, (blurry:1.4), duplicate (loli, petite, child, infant, toddlers, chibi, sd character, teen age:1.4), tsurime, helmet hair, evil smile, smug_face, naughty smile, multiple view, Reference sheet, (worst quality, low quality:1.4),
Steps: 24, Sampler: DPM++ SDE Karras, CFG scale: 10, Seed: 1140782193, Size: 1024x1536, Model hash: cc44dbff, Model: AbyssOrangeMix, Denoising strength: 0.45, ENSD: 31337, First pass size: 512x768, Model sha256: 6bb3a5a3b1eadd32, VAE sha256: f921fb3f29891d2a, Options: xformers medvram gtx_16x0
Used embeddings: bad_prompt_version2 [afea]
βΌHow to use
- VAE: orangemix.vae.pt
Prompts can be long or short
As simple as possible is good. Do not add excessive detail prompts. Start with just this. (worst quality, low quality:1.4)- Sampler: βDPM++ SDE Karrasβ is good
- Steps: forTest: 20ο½24 ,illustration: 24ο½50
- Clipskip: 1
- USE βupscale latent spaceβ
- Denoise strength: 0.45 (0.4~0.5) If you use 0.7ο½, the picture will change too much.
βΌPrompts
πWhen generating cute girls, try this negative prompt first. It avoids low quality, prevents blurring, avoids dull colors, and dictates Anime-like cute face modeling.
nsfw, (worst quality, low quality:1.3), (depth of field, blurry:1.2), (greyscale, monochrome:1.1), 3D face, nose, cropped, lowres, text, jpeg artifacts, signature, watermark, username, blurry, artist name, trademark, watermark, title, (tan, muscular, loli, petite, child, infant, toddlers, chibi, sd character:1.1), multiple view, Reference sheet,
πModel List
- AbyssOrangeMix_baseο½Instagram Merge
- AbyssOrangeMix_Nightο½+ NAI-NAISFW Merge
- AbyssOrangeMix_halfο½+ Gape0.5 Merge
- AbyssOrangeMixο½+ Gape1.0 Merge
- AbyssOrangeMix_Nightο½+ NAI-NAISFW Merge
βΌ How to choice models
- _base : SFWπ
- _Night : SFW ο½ Soft NSFWπ₯°
- _half : SFW ο½ NSFWπ
- unlabeled : SFW ο½ HARDCORE ο½π€― ex)AbyssOrangeMix, BloodOrangeMix...etc
βΌHash (SHA256)
- AbyssOrangeMix.safetensors
6bb3a5a3b1eadd32dfbc8f0987559c48cb4177aee7582baa6d6a25181929b345 - AbyssOrangeMix_half.safetensors
468d1b5038c4fbd354113842e606fe0557b4e0e16cbaca67706b29bcf51dc402 - AbyssOrangeMix_Night.safetensors
167cd104699dd98df22f4dfd3c7a2c7171df550852181e454e71e5bff61d56a6 - AbyssOrangeMix_base.ckpt
bbd2621f3ec4fad707f75fc032a2c2602c296180a53ed3d9897d8ca7a01dd6ed
βΌUse Models
- AnythingV3.0 huggingface pruned [2700c435]γ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eγ
- NovelAI animefull-final-pruned [925997e9]γ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8γ
- NovelAI sfw [1d4a34af]γ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaγ
- Gape60 [25396b85]γ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5γ
- instagram-latest-plus-clip-v6e1_50000.safetensors [] γ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5γ
- f222 [] γ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fγ
- sd1.5_pruned [] γe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053γ
AbyssOrangeMix_base (AOMb)
βΌ?
The basic trick for this merged model is to incorporate a model that has learned more than 1m Instagram photos (mostly Japanese) or a photorealistic model like f222. The choice of base model here depends on the person. I chose AnythingV3 for versatility.
βΌInstructions:
STEP: 1ο½Creation of photorealistic model for Merge
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 1.0 | instagram-latest-plus-clip-v6e1_50000 | f222 | sd1.5_pruned | Insta_F222 |
STEP: 2ο½Block Merge
Model: A | Model: B | Weight | Base alpha | Merge Name |
---|---|---|---|---|
AnythingV3.0 | Insta_F222 | 1,0.9,0.7,0.5,0.3,0.1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.1,0.3,0.5,0.7,0.9,1 | 0 | AbyssOrangeMix_base |
AbyssOrangeMix_Night (AOMn)
βΌ?
JUST AbyssOrangeMix_base+ (NAI-NAISFW) 0.3.
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AbyssOrangeMix_base | NovelAI animefull | NovelAI sfw | AbyssOrangeMix_Night |
AbyssOrangeMix_half (AOMh)
βΌ? +Gape0.5 version AbyssOrangeMix.
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.5 | AbyssOrangeMix_Night | Gape60 | NovelAI animefull | AbyssOrangeMix_half |
AbyssOrangeMix (AOM)
βΌInstructions:
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 1.0 | AbyssOrangeMix_Night | Gape60 | NovelAI animefull | AbyssOrangeMix |
ElyOrangeMix (ELOM)
βΌ?
Elysium_Anime_V2 + NAI + Gape.
This is a merge model that improves on the Elysium_Anime_V2, where NSFW representation is not good.
It can produce SFW, NSFW, and any other type of artwork, while retaining the Elysium's three-dimensional, thickly painted style.
βΌ How to choice models
- _base : SFWπ
- _Night : SFW ο½ Soft NSFWπ₯°
- _half : SFW ο½ NSFWπ
- unlabeled : SFW ο½ HARDCORE ο½π€― ex)AbyssOrangeMix, BloodOrangeMix...etc
βΌHow to use
- VAE: orangemix.vae.pt
βΌHash (SHA256)
- ElyOrangeMix [6b508e59]
- ElyOrangeMix_half [6b508e59]
- ElyNightOrangeMix[6b508e59]
ElyOrangeMix (ELOM)
βΌUse Models
- Elysium_Anime_V2 [6b508e59]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
- Gape60 [25396b85]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | Elysium_Anime_V2 | NovelAI animefull | NovelAI sfw | tempmix-part1 [] |
2 | Add Difference @ 1.0 | tempmix-part1 | Gape60 | NovelAI animefull | ElyOrangeMix [6b508e59] |
ElyOrangeMix_half (ELOMh)
βΌ?
+Gape0.5 version ElyOrangeMix.
βΌUse Models
- Elysium_Anime_V2 [6b508e59]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
- Gape60 [25396b85]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | Elysium_Anime_V2 | NovelAI animefull | NovelAI sfw | tempmix-part1 [] |
2 | Add Difference @ 0.5 | tempmix-part1 | Gape60 | NovelAI animefull | ElyOrangeMix_half [6b508e59] |
ElyNightOrangeMix (ELOMn)
βΌ?
It is a merged model that just did Elysium_Anime_V2+ (NAI-NAISFW) 0.3.
βΌUse Models
- Elysium_Anime_V2 [6b508e59]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | Elysium_Anime_V2 | NovelAI animefull | NovelAI sfw | ElyNightOrangeMix |
BloodOrangeMix (BOM)
βΌ?
Anything+NAI+Gape.
This is a merge model that improves on the AnythingV3, where NSFW representation is not good.
It can produce SFW, NSFW, and any other type of artwork, while retaining the flat, beautifully painted style of AnythingV3.
Stable. Popular in the Japanese community.
βΌModelList & [] = WebUI Hash,γγ= SHA256
- BloodNightOrangeMix.ckpt
[ffa7b160]γf8aff727ba3da0358815b1766ed232fd1ef9682ad165067cac76e576d19689e0γ - BloodOrangeMix_half.ckpt
[ffa7b160]γb2168aaa59fa91229b8add21f140ac9271773fe88a387276f3f0c7d70f726a83γ - BloodOrangeMix.ckpt
[ffa7b160] γ25cece3fe303ea8e3ad40c3dca788406dbd921bcf3aa8e3d1c7c5ac81f208a4fγ - BloodOrangeMix.safetensors
γ79a1edf6af43c75ee1e00a884a09213a28ee743b2e913de978cb1f6faa1b320dγ
βΌ How to choice models
- _base : SFWπ
- _Night : SFW ο½ Soft NSFWπ₯°
- _half : SFW ο½ NSFWπ
- unlabeled : SFW ο½ HARDCORE ο½π€― ex)AbyssOrangeMix, BloodOrangeMix...etc
βΌHow to use
- VAE: orangemix.vae.pt
BloodOrangeMix (BOM)
βΌUse Models
- AnythingV3.0 huggingface pruned [2700c435]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
- Gape60 [25396b85]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AnythingV3.0 | NovelAI animefull | NovelAI sfw | tempmix-part1 [] |
2 | Add Difference @ 1.0 | tempmix-part1 | Gape60 | NovelAI animefull | BloodOrangeMix [ffa7b160] |
BloodOrangeMix_half (BOMh)
βΌ? Anything+Nai+Gape0.5 +Gape0.5 version BloodOrangeMix. NSFW expression will be softer and have less impact on the Anything style painting style.
βΌUse Models
- AnythingV3.0 huggingface pruned [2700c435]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
- Gape60 [25396b85]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AnythingV3.0 | NovelAI animefull | NovelAI sfw | tempmix-part1 [] |
2 | Add Difference @ 0.5 | tempmix-part1 | Gape60 | NovelAI animefull | BloodOrangeMix_half [ffa7b160] |
BloodNightOrangeMix (BOMn)
βΌ?
It is a merged model that just did AnythingV3+ (NAI-NAISFW) 0.3.
βΌUse Models
- AnythingV3.0 huggingface pruned [2700c435]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
βΌInstructions
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.3 | AnythingV3.0 | NovelAI animefull | NovelAI sfw | BloodNightOrangeMix |
ElderOrangeMix
β»I found this model to be very prone to body collapse. Not recommended.
βΌ?
anything and everything mix ver.1.5+Gape+Nai(AnEve.G.N0.3)
This is a merged model with improved NSFW representation of anything and everything mix ver.1.5.
βΌHash [3a46a1e0]
βΌUse Models
- anything and everything mix ver.1.5 [5265dcf6]
- NovelAI animefull-final-pruned [925997e9]
- NovelAI sfw [1d4a34af]
- Gape60 [25396b85]
βΌInstructions:**
Step | Interpolation Method | Primary Model | Secondary Model | Tertiary Model | Merge Name |
---|---|---|---|---|---|
1 | Add Difference @ 0.5 | anything and everything mix ver.1.5 | Gape60 | NovelAI full | tempmix-part1 [] |
2 | Add Difference @ 0.3 | tempmix-part1 | NovelAI full | NovelAI sfw | ElderOrangeMix [3a46a1e0] |
Troubleshooting
- blurred Images & clearly low quality output
If the generated images are blurred or only clearly low quality output is produced, it is possible that the vae, etc. are not loaded properly. Try reloading the model/vae or restarting the WebUI/OS.
FAQ and Tips (πMEME ZONEπ¦)
Trash zone.
βΌNoooo, not work. This guy is Scammer
STEP1: BUY HUGE PC
βΌNoooo, can't generate image like samples.This models is hype.
β
π’
βΌNoooo, This models have troy virus. don't download.
All models in this repository are secure. It is most likely that anti-virus software has detected them erroneously.
However, the models with the .ckpt extension have the potential danger of executing arbitrary code.
A safe model that is free from these dangers is the model with the .safetensors extension.
βΌAOM3A1?
R.I.P.
βΌNoooo^()&*%#NG0u!!!!!!!!ηΈΊγ βηΉ§?ηΈΊε»?ηΈΊο½€ηΉο½ΌηΈΊο½¨ηΈΊε»?ηΈΊεΆο½ηΉο½ΌηΈΊο½―ι©δΈ»ο½ο½¦ιε΅?繧エηΉζΊγηΈΊ? (γAOM3A2 and A3 are overlearning and Trash. delete!γ)
βΌNoooo, Too many models. Tell me which one to choose.
- Downloads last month
- 1,273