The Deep Dark is an anime-style checkpoint that generates images that vary between an animated 2.5D look and a semi-realistic style. The model has undergone many iterations, and most of the information below applies to Version 6 models. Differences in other model versions are described in their version information.

Version 6.1 is a hybrid model created by fine-tuning a Pony XL model, surgically replacing its text encoder, and injecting it with unet blocks from a fine-tuned SDXL Turbo checkpoint. The result is a checkpoint with the desired stylistic appearance that responds well to natural language prompts, Danbooru tags, and anything in between. The Deep Dark has most of the conceptual knowledge and understanding of both SDXL and Pony models; however, it does not need Pony’s score prompts or other quality keywords required by other checkpoints – it always tries for the best quality. It produces excellent images in only 10 steps (at a CFG of 2 to 4).

Usage

To use the most recent version of this model, use the following settings:

• Prompting: Anything goes. Natural language, tags, broken English, haikus - Use whatever. Tell the model what you want. SFW and NSFW.

• Sampler & Scheduler: Again, the checkpoint isn’t picky so that you can use your favorite combination; I default to DPM++ 2S a or DPM++ SDE (Automatic, Karras, or Exponential). It is guaranteed to work with those.

• Steps & CFG: Anything works. Thanks to its turbo injection, you can get good images with 6 to 10 steps with a CFG of 2 to 4, and you can continue to see tiny improvements up to 30-40 steps with a CFG of 8 to 10. I think between 20-25 steps with a CFG of 5 to 7 is the sweet spot for a good balance, but the quality at 10 steps is excellent too.

• Styles: This model is intentionally fixed to generate only two styles based on the Deep Dark Anime look. By default, it produces more anime-style images; however, adding “realistic” to the prompt will push your images toward a semi-realistic appearance.

• LoRA compatibility: One of the intentions behind developing the Deep Dark models was to make using LoRAs unnecessary. It knows quite a lot on its own. However, its hybrid nature means it will work with various SDXL, Pony, Animagine, and Illustrious-trained LoRAs. It does work well with style LoRAs, as the style is inherent to the base model, and it may put up more of a fight if you attempt to change the style with a LoRA.

• Safety: If attempting anything NSFW, adding "children, loli, child, kids, teens" to your negative prompt and “aged up” to your positive prompt is never a bad idea.

Note: As it's no longer a traditional merged checkpoint, I’ve dropped “mix” from the checkpoint name and have rebranded this model simply as The Deep Dark. My start in fine-tuning and making Version 6.0 and beyond wouldn't have been possible without the collaboration with DucHaiten, so if you like my models, check out his stuff and consider supporting him.

Model Credits

The following models have been in at least one Deep Dark Hentai Mix version. Credit goes to the model creators (and, of course, the creators of the models used in their merges/training). In addition, several trained models or trained LoRAs may have also been used that I haven't released for one reason or another.

• DucHaiten-Infinity-Pony https://civarchive.com/models/412724/duchaiten-infinity-pony

• Trifecta XL https://civarchive.com/models/286936/trifecta-xl

• Hentai ProtoGenesis https://civarchive.com/models/244551?modelVersionId=275930

• ShikiAnimeXL https://civarchive.com/models/119163/shikianimexl

• Hentai Mix XL - Road to Freedom https://civarchive.com/models/199554?modelVersionId=296678

• Hentai Mix XL - World is flat! - https://civarchive.com/models/289879/hentai-mix-xl-world-is-flat

• Aesthetic Mix XL https://civarchive.com/models/262187/aesthetic-mix-xl

• SDXL Nuclear - General Purpose v2 https://civarchive.com/models/138331?modelVersionId=162051

• NSFW POV All In One SDXL LoRA https://civarchive.com/models/144203?modelVersionId=160240

• Deep Dark Hentai Mix https://civarchive.com/models/221751?modelVersionId=256648 

• Pyro's NSFW SDXL https://civarchive.com/models/133808/pyros-nsfw-sdxl 

• duchaitenReal3dNSFW_v10 - https://civarchive.com/models/247266/duchaiten-real3d-nsfw-xl?modelVersionId=286445

• omnigenxlNSFWSFW_v10 https://civarchive.com/models/203014/omnigenxl-nsfw-and-sfw

• betterThanWords_v30 https://civarchive.com/models/149359?modelVersionId=233092

• FappXL LyCORIS https://civarchive.com/models/248856/fappxl-multi-nsfw-concept 

• And a few other models that have vanished now and are no longer posted on model cards.

Legal Stuff

AI model output may sometimes be unpredictable and is not guaranteed to match the input prompt nor to be of any given nature or quality. Do not attempt to generate content that is illegal in your legal jurisdiction or within your country or district of residence. By using this model, you agree that the creator of this model shall in no event be responsible or liable to you or to any third party, whether under contract, warranty, tort (including negligence), strict liability, indemnity or other theory, for any direct, indirect, special, incidental, consequential, exemplary, liquidated or punitive damages or any other damages, including but not limited to loss of profit, revenue or business, cost of substitute procurement, arising in whole or in part from your use of (or inability to use) the model.

Model Restrictions:

To be clear, this model is a derivative of the SDXL 1.0 model and the following restrictions are part of the 2023 Stability AI CreativeML Open RAIL++-M License dated July 26, 2023, with applies to this model and all models that use any component or derivatives of this model. You agree not to use the Model or Derivatives of the Model:

• In any way that violates any applicable national, federal, state, local or international law or regulation; 

• For the purpose of exploiting, harming or attempting to exploit or harm minors in any way; 

• To generate or disseminate verifiably false information and/or content with the purpose of harming others; 

• To generate or disseminate personal identifiable information that can be used to harm an individual; 

• To defame, disparage or otherwise harass others; 

• For fully automated decision making that adversely impacts an individual’s legal rights or otherwise creates or modifies a binding, enforceable obligation; 

• For any use intended to or which has the effect of discriminating against or harming individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics; 

• To exploit any of the vulnerabilities of a specific group of persons based on their age, social, physical or mental characteristics, in order to materially distort the behavior of a person pertaining to that group in a manner that causes or is likely to cause that person or another person physical or psychological harm; For any use intended to or which has the effect of discriminating against individuals or groups based on legally protected characteristics or categories; 

• To provide medical advice and medical results interpretation; 

• To generate or disseminate information for the purpose to be used for administration of justice, law enforcement, immigration or asylum processes, such as predicting an individual will commit fraud/crime commitment (e.g. by text profiling, drawing causal relationships between assertions made in documents, indiscriminate and arbitrarily-targeted use).

Clarifications on the permissible use of this model:

• Versions of this model are merges. Therefore, they inherit the permissions and restrictions of all models used in their creation.

• You may not make merges, extractions, or derivatives of this model and use a less restrictive license than the 2023 Stability AI CreativeML Open RAIL++-M License.

• This model may be run and hosted for download on Civitai.com. At this time, no other organization or individual is authorized to host or distribute this model.

• You may not sell this model or any of this model’s merges or derivatives.

• You must credit this model’s creator if you use this model to create a new merged, mixed, trained, or extracted model.

• You may sell images you generate with this model.

• You may share merges using this model as long as any merge has the same or more restrictive permissions.