SOLORDZ MOSS Base versionOur first MOSS beta version of the modelOur settings:Based on base model, work with 24 steps, cfg=2-3er_sde/betaBut you can experiment with the parameters.If you like our work and want support us please support us directly, since I am in a country which cannot get money from Civitai <a target="_blank" rel="ugc" href="https://www.buymeacoffee.com/tripleheadg">https://www.buymeacoffee.com/tripleheadg</a><a target="_blank" rel="ugc" href="https://civarchive.com/user/Triple_Headed_Monkey">Also check out our other models Triple_Headed_Monkey Creator Profile | Civitai</a><h3 id="solordz_realism_moss_v1">SolordZ_Realism_Moss_v1</h3>This version of the model focuses on photographic realism, with an emphasis on believable faces, natural skin, coherent anatomy, and convincing materials and environments. It is suited to a wide range of photographic approaches, including casual snapshots, portraits, documentary imagery, fashion photography, and atmospheric scenes.The goal is realism with photographic character: natural variations in skin texture, imperfect expressions, realistic fabric behaviour, and lighting that belongs to the scene. Depending on the prompt, images can range from ordinary phone photographs and direct-flash snapshots to carefully composed portraits.Alongside the model, we developed and tested a structured six-part prompting system for describing images clearly and consistently.This approach separates photographic treatment, period, theme, fashion, visual identifiers, and scene description. It provides a practical way to organize complex requests while keeping the main description readable.We would also like to express our sincere appreciation to KandooAI, the creator of Juggernaut MOSS v1. This release incorporates weights from their model as part of the merging process. We greatly appreciate their work and contribution to the MOSS community.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/90cb1711-6298-414e-bcd8-fad7667211dc/original=true/90cb1711-6298-414e-bcd8-fad7667211dc.jpeg" /><h2 id="recommended-prompting-system">Recommended Prompting System</h2>1. ModeDefines the photographic approach and the image’s overall visual character.Include relevant framing, camera viewpoint, lighting behaviour, depth of field, focus, image texture, and processing. Examples include an available-light documentary portrait, a direct-flash compact-camera snapshot, or a softly faded colour-film photograph.Use photographic characteristics that support the intended result. Not every image needs shallow depth of field, film grain, dramatic lighting, or extreme sharpness.2. EraEstablishes the photographic period or aesthetic.Examples include late-1970s colour photography, 1990s consumer film snapshots, early-2000s compact digital photography, and contemporary smartphone images.When describing a period-inspired image, distinguish the intended aesthetic from an actual capture date.3. ThemeDefines the main subject, activity, atmosphere, and context.This section establishes the creative intention briefly: an interrupted conversation, a quiet domestic moment, an outdoor working portrait, or a spontaneous nighttime encounter.4. FashionDefines the overall clothing category and relevant fashion influences.Examples include practical workwear, contemporary casualwear, formal eveningwear, or vintage sportswear. Detailed garments, accessories, fabric textures, and their placement belong in the main Prompt.5. Additional TagsA concise, comma-separated collection of important visual anchors.These may include subject counts, approximate age groups, actions, expressions, objects, locations, weather, and distinctive photographic effects. Keep them consistent with the detailed description.6. PromptThe main descriptive section, written in connected natural language.Describe the composition, appearance, facial expression, gaze, posture, clothing, object interactions, environment, lighting, and colour relationships.For portraits, useful details include facial proportions, eyelid position, mouth shape, hair arrangement, and the skin’s base complexion. Describe skin texture proportionately: pores, fine lines, pigmentation, and small imperfections should vary naturally across the face rather than appear equally pronounced everywhere.For complex scenes, specify spatial relationships and limb ownership: which hand holds an object, where each arm rests, how bodies overlap, and where each subject is looking.Separate material colour from illumination. Warm light on neutral skin, for example, is different from a naturally warm complexion. Describe reflections, moisture, fabric sheen, and surface wear where they contribute to the scene.<h2 id="why-use-this-structure">Why Use This Structure?</h2>The six sections connect the overall photographic treatment with the physical details of the scene.In our testing, this format has been useful for organizing detailed prompts and making revisions more precise. It also makes it easier to identify conflicting instructions—for example, requesting matte skin while repeatedly emphasizing glossy highlights.The structure is a guide, not a guarantee of perfect anatomy or composition. Clear, compatible descriptions remain more useful than a large number of quality keywords.For everyday realism, describe the qualities of an ordinary photograph: uneven framing, local differences in sharpness, modest exposure imperfections, recognizable surroundings, and natural expressions. For polished photography, specify the lighting and finish deliberately.<h2 id="recommended-sampling-settings">Recommended Sampling Settings</h2>The following sampler and scheduler combinations have worked well in our testing:Sampler/Scheduler <code>deis_3m_odesgm_uniform or dpmpp_2m_sderadiance_shift</code>Treat these as starting points and adjust them to suit your workflow and preferred photographic result.SolordZ_Moss Art &amp; Anime ModelThis model has been trained on a diverse collection of artistic styles, anime aesthetics, and recognizable characters, covering a wide range of visual approaches, from stylized illustrations and Japanese animation to more realistic imagery.Alongside the model, we developed and tested a structured six-part prompting system designed to achieve clearer, more detailed, and visually consistent generations.Rather than relying on a single collection of keywords, this approach separates visual style, artistic context, subject information, and scene description into distinct sections. This helps the model interpret complex prompts more effectively, improving composition, character consistency, anatomical accuracy, and the reproduction of fine visual details.Recommended Prompting SystemOur prompting system uses six sections:1. ModeDefines the visual medium and overall artistic direction.For anime, this may include the animation studio, artist, franchise, adaptation, illustration technique, linework, cel shading, color palette, and visual effects.For photography, it describes the photographic approach, camera characteristics, image texture, lighting behavior, and processing.2. EraEstablishes the visual period and its associated aesthetic characteristics.Examples include 1990s Japanese animation, contemporary anime production, vintage film photography, or modern digital photography.3. ThemeDefines the subject, atmosphere, genre, and narrative context of the image.This section communicates the overall creative intention without overloading the detailed scene description.4. FashionDescribes clothing styles, accessories, materials, and relevant fashion influences.It helps establish the intended wardrobe while leaving precise garment placement and physical details to the main prompt.5. Additional TagsA concise, comma-separated collection of important visual identifiers.This may include character names, franchises, subject counts, poses, expressions, objects, environmental elements, and distinctive visual features.6. PromptThe main descriptive section and the most detailed part of the system.It describes the complete scene in natural language, including subject appearance, facial expressions, body positioning, limb relationships, clothing, object interactions, environment, composition, lighting, colors, and textures.Particular attention is given to spatial relationships and anatomical clarity: which hand holds an object, where each limb is positioned, how subjects interact, and how foreground and background elements overlap.The objective is to eliminate unnecessary ambiguity and give the model a precise, coherent visual description.Why use this structure?The six sections work together to establish both the overall visual identity and the smallest scene details.In our testing, this approach has produced more controlled, detailed, and visually coherent results than less structured prompting.The format is suitable for anime, illustration, character-focused artwork, and photorealistic generation.Recommended Sampling SettingsThese sampler and scheduler combinations have worked well in our testing:Generation type Sampler/SchedulerAnime / Illustration <code>deis_3m_ode / sgm_uniform</code>Realism / Photography <code>dpmpp_2m_sde / radiance_shift</code>For the realism configuration, install the following ComfyUI custom-node package:<a target="_blank" rel="ugc" href="https://github.com/silveroxides/ComfyUI-UtilsCollection">ComfyUI-UtilsCollection</a>Since this is the basic model, you can use CFG and negative prompt.For realism, we recommend CFG=2 and 24-30 stepsfor anime and art CFG=3.5 30-50 stepsGeneration Quality and WorkflowAll showcased images are generated directly with the model using a basic ComfyUI workflow.No external upscalers, post-processing enhancers, detail restoration tools, or additional image improvement stages are used.The examples represent the model's native generation capabilities rather than results enhanced through a separate processing pipeline.For the best results, we recommend using the structured six-part prompting system described above.<h2 id="solordz_zero_restart-(krea2)">SolordZ_Zero_Restart (Krea2)</h2>Our an experimental version based on the RAW model.It works in 8 steps and CFG=1, but the prompt needs to be written in more detail than for the Turbo version.There are two versions available at once Int8_Convrot and FP8ZIT/ZIB/Krea 2 project collaboration with TripleHeadedMonkeyKrea2 project<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Our first TE for Krea2 - Improved prompt understanding and details <a target="_blank" rel="ugc" href="https://civarchive.com/models/2662833/tesolordzzimagekrea2">TE_SoLordZ_ZImage_Krea2 - Krea2_SolordZ_TE_v1.0 | Krea 2 Text Encoder | Civitai</a>ZIB projectReturning after a pause.I left Civitai for a short time due to personal circumstances. However, during my absence, together with Triple Headed Monkey, we did not stop working and conducted several major projects.Today I decided to publish one of the very first, but at the same time strong versions of our joint development — the improved ZIT base. Key changes compared to what was before: increased detail, reduced background blurring (less blurring) and improved realism.Thanks for waiting. We are back and ready to continue working.We use our training data and our soft for training.!!! For best result use this TE <a target="_blank" rel="ugc" href="https://civarchive.com/models/2650573/thmuncz-imagete?modelVersionId=2976257">THM_UNC_Z-image_TE - v1.0 | ZImage Text Encoder | Civitai</a>We released our another uncensored TE <a target="_blank" rel="ugc" href="https://civarchive.com/models/2662833/zimagetesolordz?modelVersionId=2990083">Z_Image_TE_SoLordZ - Z_Image_TE_SoLordZ_v1 | ZImage Text Encoder | Civitai</a>If you like our work and want support us please support us directly, since I am in a country which cannot get money from Civitai <a target="_blank" rel="ugc" href="https://www.buymeacoffee.com/tripleheadg">https://www.buymeacoffee.com/tripleheadg</a>our WF's <a target="_blank" rel="ugc" href="https://civarchive.com/models/2663318/solordz-wf-upscale?modelVersionId=2990649">SoLordZ WF (upscale) - v1.0 | ZImage Workflows | Civitai</a>and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2662358?modelVersionId=2989530">https://civarchive.com/models/2662358?modelVersionId=2989530</a>for realism use euler/normal, for art sometimes better result give euler_a/normalAll images in the gallery just raw generation w/o any tricks or upscaleZIB SoLordZ v1ZIB SoLordZ is a new chapter in the evolution of Z—Image and our collaboration with Triple Headed Monkey. The basic Z-Image model (up to 50 steps, CFG 5-9) is flexible, but requires fine tuning. Our goal was ambitious: to keep control of the "slow" model, but to teach it to generate in 8 steps with CFG=1.We are releasing ZIB SoLordZ, one of our first public releases in a series of improvements to the Z Image Base.ZIB SoLordZ is designed to work in "turbo mode": 8 steps, CFG=1. Due to this, it ignores the negative prompts. Everything that you want to exclude should be written directly in a positive prompt. This is especially important for anatomy, where control is crucial.SoLordZ Krea2 Turbo v1Together with TripleHeadedMonkey (<a target="_blank" rel="ugc" href="https://civarchive.com/user/Triple_Headed_Monkey">https://civarchive.com/user/Triple_Headed_Monkey</a>) we present our first merged version based on the stunning Krea2 Turbo base.We are actively working on future updates, so stay tuned! Let us know your feedback and don't forget to show us what you create. ❤️We use er_sde/beta sampler for the first pass and euler/simple for the 2nd pass.You can use our WF (in each image).

SoLordZ ZIT/ZIB/Krea2 8steps

FP8 version SoLordZ_ZIB_v3, the FP8 file size is almost the same as that of the BF16 version because most of the layers are saved in order not to lose quality. If necessary, I can publish a smaller version, but it has lost a lot of quality after quantization.

Files with SHA256: b210f8f681c02b3ed382153a5db585ac76ce4ea02fbb404a4ef5affbf784b024