Model Database's logo

WarriorMama777
/
OrangeMixs

Text-to-Image Diffusers
StableDiffusionPipeline stable-diffusion
Model card Files Community
96
Edit model card

OrangeMixs

"OrangeMixs" shares various Merge models that can be used with StableDiffusionWebui:Automatic1111 and others.  

Maintain a repository for the following purposes.

  1. to provide easy access to models commonly used in the Japanese community.The Wisdom of the Anons๐Ÿ’Ž
  2. 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

  1. Read the ToC as release notes.
    Sections are in descending order. The order within the section is ascending. It is written like SNS.
  2. UPDATE NOTE
  3. 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. Remove terms of service.

Gradio

We support a Gradio Web UI to run OrangeMixs: Open In Spaces

Table of Contents

Reference

+/hdg/ Stable Diffusion Models Cookbook - https://rentry.org/hdgrecipes#g-anons-unnamed-mix-e93c3bf7 Model names are named after Cookbook precedents๐ŸŠ

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:

  1. You can't use the model to deliberately produce nor share illegal or harmful outputs or content
  2. 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
  3. 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 ๏ผšhttps://Model Database.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 publish the recipe, 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

โš Deprecated: Orange has grown too huge. Doing this will kill your storage.

  1. install Git
  2. create a folder of your choice and right click โ†’ "Git bash here" and open a gitbash on the folder's directory.
  3. run the following commands in order.
git lfs install
git clone https://Model Database.co/WarriorMama777/OrangeMixs
  1. complete

Batch Download (Advanced)

Advanced: (When you want to download only selected directories, not the entire repository.)  

Toggle: How to Batch Download (Advanced)
  1. Run the command git clone --filter=tree:0 --no-checkout https://Model Database.co/WarriorMama777/OrangeMixs to clone the Model Database 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://Model Database.co/WarriorMama777/OrangeMixs
  1. Move to the cloned directory with the command cd OrangeMixs.
cd OrangeMixs
  1. 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
  1. Specify the directory you want to get with the command git sparse-checkout add <directory name>. For example, if you want to get only the Models/AbyssOrangeMix3 directory, enter git sparse-checkout add Models/AbyssOrangeMix3.
git sparse-checkout add Models/AbyssOrangeMix3
  1. 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

  1. Go to the Files and vaersions tab.
  2. select the model you want to download
  3. download
  4. complete

Model Detail & Merge Recipes

AbyssOrangeMix3 (AOM3)

โ€•โ€•Everyone has different โ€œABYSSโ€!

โ–ผAbout

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.

โ–ผLinks

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 ZONEโ†“

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.)

  • ๐Ÿ” When generating illustrations for the general public: write "nsfw" in the negative prompt field
  • ๐Ÿ”ž 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

๐ŸšงEditing๐Ÿšง

AOM3

โ–ผAOM3

(Actually, this gallery doesn't make much sense since AOM3 is mainly an improvement of the NSFW part ๐Ÿ˜‚ ...But we can confirm that the picture is not much different from AOM2sfw.)

AOM3A1

โ›”Only 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.โ›” GO TO MEME ZONEโ†“

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.

โ–ผA1

ยฉ (1)ยฉYurucamp: Inuyama Aoi, (2)ยฉThe Quintessential Quintuplets: Nakano Yotsuba, (3)ยฉSailor Moon: Mizuno Ami/SailorMercury

AOM3A2

๐ŸšงEditing๐Ÿšง 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

๐ŸšงEditing๐Ÿšง 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 AOM3A3 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.

โ–ผA1B


ยฉ ยฉumamusume: Meisho Doto, ยฉGirls und Panzer: Nishizumi Miho,ยฉIDOLM@STER: Sagisawa Fumika

AOM3B2

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

AOM3B2_orangemixs_sampleGallery

Sample image prompts
  1. Maid
  2. Yotsuba: https://majinai.art/ja/i/f-O4wau
  3. Inuko in cafe: https://majinai.art/ja/i/Cj-Ar9C
  4. bathroom: https://majinai.art/ja/i/XiSj5K6

 

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.

โ–ผA summary of the AOM3 work is as follows

  1. investigated the impact of the NAI and gape layers as AOM2 _nsfw onwards is crap.
  2. cut face layer: OUT04 because I want realistic faces to stop โ†’ Failed. No change.
  3. gapeNAI layer investigation๏ฝœ
    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,OUT04๏ฝœNSFW effects are in(?). e.g.: spoken hearts, trembling, motion lines, etc...
    d. OUT05๏ฝœThis is really an NSFW switch. All the "NSFW atmosphere" is in here. Facial expressions, Heavy breaths, etc...
    e. OUT10-11๏ฝœPaint layer. Does not affect detail, but does have an extensive impact.
  4. (mass production of rubbish from here...)
  5. cut IN05-08 and merge NAIgape with flat parameters โ†’ avoided creepy muscles and real faces. Also, merging NSFW models stronger has less impact.
  6. so, cut IN05-08, OUT10-11 and merge NAI+gape with all others 0.5.
  7. โ†’ AOM3
    AOM3 roughly looks like this

โ–ผHow 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),
    • ๐Ÿ”ฐBasic negative prompts sample for Anime girl โ†“
      • 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,
  • Sampler: โ€œDPM++ SDE Karrasโ€ is good Take your pick
  • Steps:
    • DPM++ SDE Karras: Test: 12๏ฝž ,illustration: 20๏ฝž
    • DPM++ 2M Karras: Test: 20๏ฝž ,illustration: 28๏ฝž
  • Clipskip: 1 or 2
  • CFG: 8 (6๏ฝž12)
  • Upscaler :
    • Detailed illust โ†’ Latenet (nearest-exact)
      Denoise strength: 0.5 (0.5~0.6)
    • Simple upscale: Swin IR, ESRGAN, Remacri etcโ€ฆ
      Denoise strength: Can be set low. (0.35~0.6)

๐Ÿ‘ฉโ€๐ŸณModel details / Recipe

โ–ผHash

  • AOM3.safetensors
    D124FC18F0232D7F0A2A70358CDB1288AF9E1EE8596200F50F0936BE59514F6D
  • AOM3A1.safetensors
    F303D108122DDD43A34C160BD46DBB08CB0E088E979ACDA0BF168A7A1F5820E0
  • AOM3A2.safetensors
    553398964F9277A104DA840A930794AC5634FC442E6791E5D7E72B82B3BB88C3
  • AOM3A3.safetensors
    EB4099BA9CD5E69AB526FCA22A2E967F286F8512D9509B735C892FA6468767CF

โ–ผUse Models

  1. AOM2sfw
    ใ€Œ038ba203d8ba3c8af24f14e01fbb870c85bbb8d4b6d9520804828f4193d12ce9ใ€
  2. AnythingV3.0 Model Database pruned
    [2700c435]ใ€Œ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eใ€
  3. NovelAI animefull-final-pruned
    [925997e9]ใ€Œ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8ใ€
  4. NovelAI sfw
    [1d4a34af]ใ€Œ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaใ€
  5. Gape60
    [25396b85]ใ€Œ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5ใ€
  6. BasilMix
    ใ€Œbbf07e3a1c3482c138d096f7dcdb4581a2aa573b74a68ba0906c7b657942f1c2ใ€
  7. chilloutmix_fp16.safetensors
    ใ€Œ4b3bf0860b7f372481d0b6ac306fed43b0635caf8aa788e28b32377675ce7630ใ€
  8. Counterfeit-V2.5_fp16.safetensors
    ใ€Œ71e703a0fca0e284dd9868bca3ce63c64084db1f0d68835f0a31e1f4e5b7cca6ใ€
  9. kenshi_01_fp16.safetensors
    ใ€Œ3b3982f3aaeaa8af3639a19001067905e146179b6cddf2e3b34a474a0acae7faใ€

โ–ผAOM3

โ–ผInstructions:

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

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

โ–ผAOM3A1 โ›”Only this model (AOM3A1) includes ChilloutMix (=The curse of DreamLike).Commercial use is not available.

โ–ผInstructions:

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: 3 |

CUT: BASE0, IN00-IN08๏ผš0, IN10๏ผš0.1, OUT03-04-05๏ผš0, OUT08๏ผš0.2

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

โ–ผAOM3A2

โ–ผ?

CUT: BASE0, IN05:0.3ใ€IN06-IN08๏ผš0, IN10๏ผš0.1, OUT03๏ผš0, OUT04๏ผš0.3, OUT05๏ผš0, OUT08๏ผš0.2

โ–ผInstructions:

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 AOM3A2

โ–ผAOM3A3

โ–ผ?

CUT : BASE0, IN05-IN08๏ผš0, IN10๏ผš0.1, OUT03๏ผš0.5, OUT04-05๏ผš0.1, OUT08๏ผš0.2

Model: A Model: B Model: C Interpolation Method Weight Merge Name
AOM3 Counterfeit2.5 Add SUM @ 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

 

AbyssOrangeMix2 (AOM2)

โ€•โ€•Creating the next generation of illustration with โ€œAbyssโ€!

โ€HeroImage_AbyssOrangeMix2_Designed_01_comp001โ€

Prompt: https://majinai.art/ja/i/nxpKRpw

โ–ผAbout

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.

โ€UBM_ON_OFF_4_comp001.webpโ€ โ€ปnvidia joke.

โ–ผDescription 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.

  1. the model used for U-Net Blocks Weight Merge was changed from Instagram+F222 to BasilMix. (https://Model Database.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.

โ€AbyssOrangeMix2_comparison_comp001โ€

โ€ปThis 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.

โ–ผSample 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 AI๐Ÿค”

โ€SampleGallerBoardDesign_AbyssOrangeMix2_03_comp001โ€
โ–ผREAD MORE๐Ÿ–ผ โ€SampleGallerBoardDesign_AbyssOrangeMix2_03_comp001โ€

โ–ผAll prompts to generate sample images

  1. Gaming Girl
  2. Fantasy
  3. Rainy Day
  4. Kemomimi Girl
  5. Supermarket
  6. Lunch Time
  7. Womens in the Garden
  8. Pirate
  9. Japanese Girl
  10. Sweets Time
  11. Glasses Girl

โ–ผHow 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: โ€œDPM++ SDE Karrasโ€ is good
  • Steps: forTest: 12๏ฝž ,illustration: 20๏ฝž
  • Clipskip: 1 or 2
  • Upscaler : Latenet (nearest-exact)
  • CFG Scale : 5 or 6 (4๏ฝž8)
  • Denoise strength: 0.5 (0.45~0.6)
    If you use 0.7๏ฝž, the picture will change too much.
    If below 0.45, Block noise occurs.

๐Ÿ—’Model List

  • AbyssOrangeMix2_sfw๏ฝœBasilMix U-Net Blocks Weight Merge
    • AbyssOrangeMix2_nsfw๏ฝœ+ NAI-NAISFW 0.3 Merge
      • AbyssOrangeMix2_hard๏ฝœ+ Gape 0.3 Merge

โ€ปChanged suffix of models.
_base โ†’_sfw: _base was changed to_sfw. _night โ†’_nsfw: Merged models up to NAI-NAI SFW were changed from _night to_nsfw. _half and non suffix โ†’_hard: Gape merged models were given the suffix _hard.gape was reduced to 0.3 because it affects character modeling.

โ–ผHow to choice models

  • _sfw : SFW๐Ÿ˜‰
  • _nsfw : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _hard : SFW ๏ฝž hard NSFW๐Ÿ‘„

โ–ผHash

  • AbyssOrangeMix2_sfw.ckpt
    ใ€Œf75b19923f2a4a0e70f564476178eedd94e76e2c94f8fd8f80c548742b5b51b9ใ€
  • AbyssOrangeMix2_sfw.safetensors
    ใ€Œ038ba203d8ba3c8af24f14e01fbb870c85bbb8d4b6d9520804828f4193d12ce9ใ€
  • AbyssOrangeMix2_nsfw.safetensors
    ใ€Œ0873291ac5419eaa7a18726e8841ce0f15f701ace29e0183c47efad2018900a4ใ€
  • AbyssOrangeMix_hard.safetensors
    ใ€Œ0fc198c4908e98d7aae2a76bd78fa004e9c21cb0be7582e36008b4941169f18eใ€

โ–ผUse Models

  1. AnythingV3.0 Model Database pruned
    [2700c435]ใ€Œ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eใ€
  2. NovelAI animefull-final-pruned
    [925997e9]ใ€Œ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8ใ€
  3. NovelAI sfw
    [1d4a34af]ใ€Œ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaใ€
  4. Gape60
    [25396b85]ใ€Œ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5ใ€
  5. BasilMix
    ใ€Œbbf07e3a1c3482c138d096f7dcdb4581a2aa573b74a68ba0906c7b657942f1c2ใ€

AbyssOrangeMix2_sfw (AOM2s)

โ–ผInstructions:

STEP: 1๏ฝœBlock 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)

โ–ผ?

JUST AbyssOrangeMix2_sfw+ (NAI-NAISFW) 0.3.

โ–ผInstructions:

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)

โ–ผ? +Gape0.3 version AbyssOrangeMix2_nsfw.

โ–ผInstructions

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.

โ€ปThis does not fundamentally improve the fingers. Therefore, More research needs to be done to improve the fingers (e.g. 'bad_prompt').

โ€HeroImage_EerieOrangeMix_Designed_comp001โ€

 

EerieOrangeMix (EOM1)

โ–ผ?

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.

๐Ÿ—’Model List

  • EerieOrangeMix_base๏ฝœInstagram+F222 U-Net Blocks Weight Merge
    • EerieOrangeMix_night๏ฝœ+ NAI-NAISFW Merge
      • EerieOrangeMix_half๏ฝœ+ Gape0.5 Merge
      • EerieOrangeMix๏ฝœ+ Gape1.0 Merge

โ–ผ How to choice models

  • _base : SFW๐Ÿ˜‰
  • _Night : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _half : SFW ๏ฝž NSFW๐Ÿ‘„
  • unlabeled : SFW ๏ฝž HARDCORE ๏ฝž๐Ÿคฏ ex)AbyssOrangeMix, BloodOrangeMix...etc

โ–ผHash

  • EerieOrangeMix.safetensors
  • EerieOrangeMix_half.safetensors
  • EerieOrangeMix_night.safetensors
  • EerieOrangeMix_base.ckpt

โ–ผUse Models

[] = WebUI Hash,ใ€Œใ€= SHA256

  1. Elysium Anime V2 []ใ€Œ5c4787ce1386500ee05dbb9d27c17273c7a78493535f2603321f40f6e0796851ใ€
  2. NovelAI animefull-final-pruned [925997e9]ใ€Œ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8ใ€
  3. NovelAI sfw [1d4a34af]ใ€Œ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaใ€
  4. Gape60 [25396b85]ใ€Œ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5ใ€
  5. instagram-latest-plus-clip-v6e1_50000.safetensors [] ใ€Œ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5ใ€
  6. f222 [] ใ€Œ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fใ€
  7. sd1.5_pruned [] ใ€Œe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053ใ€

โ–ผ Sample Gallery

โ€2022-12-30_MotorbikeGIrlAsa3_comp001โ€
More๐Ÿ–ผ โ€2022-12-30_SampleGallery5โ€

โ–ผ 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: โ€œDPM++ SDE Karrasโ€ is good
  • Steps: forTest: 20๏ฝž24 ,illustration: 24๏ฝž50
  • Clipskip: 1
  • USE โ€œupscale latent spaceโ€
  • Denoise strength: 0.45 (0.4~0.5)
    If you use 0.7๏ฝž, the picture will change too much.

โ–ผPrompts

๐Ÿ–ŒWhen 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)

โ–ผ?
Details are omitted since it is the same as AbyssOrangeMix.

โ–ผInstructions:

STEP: 1๏ฝœCreation 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: 2๏ฝœBlock 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)

โ–ผ?

JUST EerieOrangeMix_base+ (NAI-NAISFW) 0.3.

โ–ผInstructions

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)

โ–ผ? +Gape0.5 version EerieOrangeMix.

โ–ผInstructions:

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)

โ–ผInstructions:

Step Interpolation Method Primary Model Secondary Model Tertiary Model Merge Name
1 Add Difference @ 1.0 EerieOrangeMix_Night Gape60 NovelAI animefull EerieOrangeMix

EerieOrangeMix2 (EOM2)

โ–ผ?

The model was created by adding the hierarchy responsible for detailing and painting ElysiumV1 to EerieOrangeMix_base, then merging NAI and Gape.

๐Ÿ—’Model List

  • EerieOrangeMix2_base๏ฝœInstagram+F222+ElysiumV1 U-Net Blocks Weight Merge
    • EerieOrangeMix2_night๏ฝœ+ NAI-NAISFW Merge
      • EerieOrangeMix2_half๏ฝœ+ Gape0.5 Merge
      • EerieOrangeMix2๏ฝœ+ Gape1.0 Merge

โ–ผ How to choice models

  • _base : SFW๐Ÿ˜‰
  • _Night : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _half : SFW ๏ฝž NSFW๐Ÿ‘„
  • unlabeled : SFW ๏ฝž HARDCORE ๏ฝž๐Ÿคฏ ex)AbyssOrangeMix, BloodOrangeMix...etc

โ–ผHash

  • EerieOrangeMix2.safetensors
  • EerieOrangeMix2_half.safetensors
  • EerieOrangeMix2_night.safetensors
  • EerieOrangeMix2_base.ckpt

โ–ผUse Models

[] = webuHash,ใ€Œใ€= SHA256

  1. Elysium Anime V2 []ใ€Œ5c4787ce1386500ee05dbb9d27c17273c7a78493535f2603321f40f6e0796851ใ€
  2. NovelAI animefull-final-pruned [925997e9]ใ€Œ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8ใ€
  3. NovelAI sfw [1d4a34af]ใ€Œ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaใ€
  4. Gape60 [25396b85]ใ€Œ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5ใ€
  5. instagram-latest-plus-clip-v6e1_50000.safetensors [] ใ€Œ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5ใ€
  6. f222 [] ใ€Œ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fใ€
  7. sd1.5_pruned [] ใ€Œe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053ใ€
  8. ElysiumV1 ใ€Œabbb28cb5e70d3e0a635f241b8d61cefe42eb8f1be91fd1168bc3e52b0f09ae4ใ€

EerieOrangeMix2_base (EOM2b)

โ–ผ?

โ–ผInstructions

STEP: 1๏ฝœBlock 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)

โ–ผ?

JUST EerieOrangeMix2_base+ (NAI-NAISFW) 0.3.

โ–ผInstructions

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)

โ–ผ? +Gape0.5 version EerieOrangeMix2.

โ–ผInstructions

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)

โ–ผInstructions:

Step Interpolation Method Primary Model Secondary Model Tertiary Model Merge Name
1 Add Difference @ 1.0 EerieOrangeMix_Night Gape60 NovelAI animefull EerieOrangeMix2

Models Comparison

MotorbikeGIrlAsa_Eerie_Abyss_Comparison_comp001 โ€Eerie_Abyss_Comparison_02_comp001โ€ โ€Eerie_Comparison_01_comp001โ€ โ€ปThe difference is slight but probably looks like this. โ† warm color, โ†‘ natural color, โ†’ animated color

AbyssOrangeMix (AOM)

โ€•โ€•How can you guys take on such a deep swamp and get results?
Is it something like "Made in Abyss"?
By Anon, 115th thread

โ–ผ?

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.

โ€ปThis model is the result of a great deal of testing and experimentation by many Anons๐Ÿค— โ€ปThis 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.
โ€ปThis does not fundamentally improve the fingers, and I recommend using bad_prompt, etc. (Embedding) in combination.

โ–ผSample 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]

โ–ผHow 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: โ€œDPM++ SDE Karrasโ€ is good
  • Steps: forTest: 20๏ฝž24 ,illustration: 24๏ฝž50
  • Clipskip: 1
  • USE โ€œupscale latent spaceโ€
  • Denoise strength: 0.45 (0.4~0.5) If you use 0.7๏ฝž, the picture will change too much.

โ–ผPrompts

๐Ÿ–ŒWhen 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,

๐Ÿ—’Model List

  • AbyssOrangeMix_base๏ฝœInstagram Merge
    • AbyssOrangeMix_Night๏ฝœ+ NAI-NAISFW Merge
      • AbyssOrangeMix_half๏ฝœ+ Gape0.5 Merge
      • AbyssOrangeMix๏ฝœ+ Gape1.0 Merge

โ–ผ How to choice models

  • _base : SFW๐Ÿ˜‰
  • _Night : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _half : SFW ๏ฝž NSFW๐Ÿ‘„
  • unlabeled : SFW ๏ฝž HARDCORE ๏ฝž๐Ÿคฏ ex)AbyssOrangeMix, BloodOrangeMix...etc

โ–ผHash (SHA256)

  • AbyssOrangeMix.safetensors
    6bb3a5a3b1eadd32dfbc8f0987559c48cb4177aee7582baa6d6a25181929b345
  • AbyssOrangeMix_half.safetensors
    468d1b5038c4fbd354113842e606fe0557b4e0e16cbaca67706b29bcf51dc402
  • AbyssOrangeMix_Night.safetensors
    167cd104699dd98df22f4dfd3c7a2c7171df550852181e454e71e5bff61d56a6
  • AbyssOrangeMix_base.ckpt
    bbd2621f3ec4fad707f75fc032a2c2602c296180a53ed3d9897d8ca7a01dd6ed

โ–ผUse Models

  1. AnythingV3.0 Model Database pruned [2700c435]ใ€Œ543bcbc21294831c6245cd74c8a7707761e28812c690f946cb81fef930d54b5eใ€
  2. NovelAI animefull-final-pruned [925997e9]ใ€Œ89d59c3dde4c56c6d5c41da34cc55ce479d93b4007046980934b14db71bdb2a8ใ€
  3. NovelAI sfw [1d4a34af]ใ€Œ22fa233c2dfd7748d534be603345cb9abf994a23244dfdfc1013f4f90322fecaใ€
  4. Gape60 [25396b85]ใ€Œ893cca5903ccd0519876f58f4bc188dd8fcc5beb8a69c1a3f1a5fe314bb573f5ใ€
  5. instagram-latest-plus-clip-v6e1_50000.safetensors [] ใ€Œ8f1d325b194570754c6bd06cf1e90aa9219a7e732eb3d488fb52157e9451a2a5ใ€
  6. f222 [] ใ€Œ9e2c6ceff3f6d6f65c6fb0e10d8e69d772871813be647fd2ea5d06e00db33c1fใ€
  7. sd1.5_pruned [] ใ€Œe1441589a6f3c5a53f5f54d0975a18a7feb7cdf0b0dee276dfc3331ae376a053ใ€

AbyssOrangeMix_base (AOMb)

โ–ผ?

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.

โ–ผInstructions:

STEP: 1๏ฝœCreation 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: 2๏ฝœBlock 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)

โ–ผ?

JUST AbyssOrangeMix_base+ (NAI-NAISFW) 0.3.

โ–ผInstructions:

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)

โ–ผ? +Gape0.5 version AbyssOrangeMix.

โ–ผInstructions:

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)

โ–ผInstructions:

Step Interpolation Method Primary Model Secondary Model Tertiary Model Merge Name
1 Add Difference @ 1.0 AbyssOrangeMix_Night Gape60 NovelAI animefull AbyssOrangeMix

ElyOrangeMix (ELOM)

โ–ผ?
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.

โ–ผ How to choice models

  • _base : SFW๐Ÿ˜‰
  • _Night : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _half : SFW ๏ฝž NSFW๐Ÿ‘„
  • unlabeled : SFW ๏ฝž HARDCORE ๏ฝž๐Ÿคฏ ex)AbyssOrangeMix, BloodOrangeMix...etc

โ–ผHow to use

  • VAE: orangemix.vae.pt

โ–ผHash (SHA256)

  • ElyOrangeMix [6b508e59]
  • ElyOrangeMix_half [6b508e59]
  • ElyNightOrangeMix[6b508e59]

ElyOrangeMix (ELOM)

โ–ผUse Models

  1. Elysium_Anime_V2 [6b508e59]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]
  4. Gape60 [25396b85]

โ–ผInstructions

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)

โ–ผ?

+Gape0.5 version ElyOrangeMix.

โ–ผUse Models

  1. Elysium_Anime_V2 [6b508e59]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]
  4. Gape60 [25396b85]

โ–ผInstructions

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)

โ–ผ?

It is a merged model that just did Elysium_Anime_V2+ (NAI-NAISFW) 0.3.

โ–ผUse Models

  1. Elysium_Anime_V2 [6b508e59]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]

โ–ผInstructions

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)

โ–ผ? 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.

โ–ผModelList & [] = WebUI Hash,ใ€Œใ€= SHA256

  • BloodNightOrangeMix.ckpt
    [ffa7b160]ใ€Œf8aff727ba3da0358815b1766ed232fd1ef9682ad165067cac76e576d19689e0ใ€
  • BloodOrangeMix_half.ckpt
    [ffa7b160]ใ€Œb2168aaa59fa91229b8add21f140ac9271773fe88a387276f3f0c7d70f726a83ใ€
  • BloodOrangeMix.ckpt
    [ffa7b160] ใ€Œ25cece3fe303ea8e3ad40c3dca788406dbd921bcf3aa8e3d1c7c5ac81f208a4fใ€
  • BloodOrangeMix.safetensors
    ใ€Œ79a1edf6af43c75ee1e00a884a09213a28ee743b2e913de978cb1f6faa1b320dใ€

โ–ผ How to choice models

  • _base : SFW๐Ÿ˜‰
  • _Night : SFW ๏ฝž Soft NSFW๐Ÿฅฐ
  • _half : SFW ๏ฝž NSFW๐Ÿ‘„
  • unlabeled : SFW ๏ฝž HARDCORE ๏ฝž๐Ÿคฏ ex)AbyssOrangeMix, BloodOrangeMix...etc

โ–ผHow to use

  • VAE: orangemix.vae.pt

BloodOrangeMix (BOM)

โ–ผUse Models

  1. AnythingV3.0 Model Database pruned [2700c435]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]
  4. Gape60 [25396b85]

โ–ผInstructions

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)

โ–ผ? Anything+Nai+Gape0.5 +Gape0.5 version BloodOrangeMix. NSFW expression will be softer and have less impact on the Anything style painting style.

โ–ผUse Models

  1. AnythingV3.0 Model Database pruned [2700c435]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]
  4. Gape60 [25396b85]

โ–ผInstructions

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)

โ–ผ?

It is a merged model that just did AnythingV3+ (NAI-NAISFW) 0.3.

โ–ผUse Models

  1. AnythingV3.0 Model Database pruned [2700c435]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]

โ–ผInstructions

Step Interpolation Method Primary Model Secondary Model Tertiary Model Merge Name
1 Add Difference @ 0.3 AnythingV3.0 NovelAI animefull NovelAI sfw BloodNightOrangeMix

ElderOrangeMix

โ€ปI found this model to be very prone to body collapse. Not recommended.

โ–ผ?
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.

โ–ผHash [3a46a1e0]

โ–ผUse Models

  1. anything and everything mix ver.1.5 [5265dcf6]
  2. NovelAI animefull-final-pruned [925997e9]
  3. NovelAI sfw [1d4a34af]
  4. Gape60 [25396b85]

โ–ผInstructions:**

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

  1. 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 (๐ŸˆMEME ZONE๐Ÿฆ)

Trash zone.

โ–ผAOM3A1?
R.I.P.

โ–ผNo, AOM2 (only hentai models)

โ–ผNooo^()&*%#NG0u!!!!!!!!็ธบใ‚…โ™€็นง?็ธบๅŒป?็ธบ๏ฝค็น๏ฝผ็ธบ๏ฝจ็ธบๅŒป?็ธบๅถ๏ฝŠ็น๏ฝผ็ธบ๏ฝฏ้ฉ•ไธป๏ฝญ๏ฝฆ้„™ๅต?็นง๏ฝด็นๆบ˜ใ€’็ธบ? (ใ€ŒAOM3A2 and A3 are overlearning and Trash. delete!ใ€)

โ€getting_excitedโ€

โ–ผNoo, Too many models. Tell me which one to choose.

โ†’ ๅ…จ้ƒจๅŒใ˜ใ˜ใ‚ƒใชใ„ใงใ™ใ‹

โ–ผNooo, not work. This guy is Scammer
STEP1: BUY HUGE PC

โ–ผNoooo, can't generate image like samples.This models is hype.

โŒ

๐ŸŸข

โ–ผNooooo, 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.

Downloads last month
6,793
Hosted inference API
Text-to-Image
This model can be loaded on the Inference API on-demand.

Dataset used to train WarriorMama777/OrangeMixs

Spaces using WarriorMama777/OrangeMixs 165