MOSS Image Evening Report 2026-11-20 | Top Pick: DC Dark Fantasy Style 1

This issue reviewed 66 new MOSS versions: 13 rated A, 9 rated B, and 44 rated C. There were no S ratings, and no duplicates requiring exclusion were found. The top pick is the MOSS version of DC Dark Fantasy: a dark fantasy style with hyper-realistic CGI texture. The trigger word includes a trailing period, and example images come with workflows for direct reproduction. Other highlights include in-game art styles from *Stellar Blade* and *Genshin Impact*, Dizzle's Jupiter and Twine materials, dusk lighting, pencil sketches, and Fujifilm styles. Regular character LoRAs such as the Kemono Friends series, Soul Tide series, and AI Babe Pack are rated C. The JSON-annotated and standard versions of Canghu and Japanese Wolf have different hashes, so both are retained.

Zhenzhen's AI Radar · MOSS Image · 2026-11-20 Evening Report

This issue features 66 new versions: Strongly Recommended 0 · Recommended 13 · Worth Watching 9 · Included Only 44

Editor's Note This issue reviewed 66 new MOSS versions: 13 rated A, 9 rated B, and 44 rated C. There were no S ratings, and no duplicates requiring exclusion were found. The top pick is the MOSS version of DC Dark Fantasy: a dark fantasy style with hyper-realistic CGI texture. The trigger word includes a trailing period, and example images come with workflows for direct reproduction. Other highlights include in-game art styles from Stellar Blade and Genshin Impact, Dizzle's Jupiter and Twine materials, dusk lighting, pencil sketches, and Fujifilm styles. Regular character LoRAs such as the Kemono Friends series, Soul Tide series, and AI Babe Pack are rated C. The JSON-annotated and standard versions of Canghu and Japanese Wolf have different hashes, so both are retained.

Recommended (13)

[A] DC Dark Fantasy Style 1 · v1.0 [MossBase]

LoRA · Port · Author dkpc69 · Published 11-20 13:07 · Downloads 50

A dark fantasy style LoRA with hyper-realistic CGI texture (grotesque, surreal adventure scenes). This is the MOSS version of an established multi-base model page, trained on MOSS Base. The author states it also works well on the SFT base model.

Images

Example 768x1344

Reasons for Recommendation

  • Port: The same page already has Krea 2, Flux.2 Klein 9B, and Z-Image Turbo versions, with the entire model page having 1363 downloads. The MOSS version was trained on Base for 5080 steps, 40 epochs, rank 32, using GPT4.1 for tagging (Fizgig trainer). The trigger word follows the convention of the Krea 2 version.
  • Serves dark fantasy illustrations, monster and adventure scenes, and surreal concept art. 19 example images are visible; the author states each image includes ComfyUI workflow metadata, allowing reproduction by dragging it into ComfyUI.
  • Free, with a permissive license (generation, rental, and derivatives allowed, no attribution required). The author only requests credit when merging or reposting.

How to Use

  • Trigger word: dcd@rkf@ntsy. (The trailing period is part of the trigger word; the Klein / Z-Image versions use dcdrkfntsy. without the @).
  • Base models: MOSS Base or SFT. If the desired effect is not achieved, the author suggests reducing the strength from 1.0 to .75 (this advice was originally written for the Z-Image / Klein versions).
  • Steps, CFG, and sampler are not listed separately in the model card; they can be read from the workflow metadata of the example images.

Risks and Limitations

  • The MOSS version lacks separate effect descriptions, and the author has not detailed the differences from other base model versions. Additionally, 1 sensitive-level example images are hidden.
  • The model card includes a donation link.

Trigger Word: dcd@rkf@ntsy.

[A] AG: Foras - Mixed Anime Styles · MOSS-Image

LoRA · Port · Author Azriaelis · Published 11-20 09:10 · Downloads 46

Foras, a style LoRA from the "Ars Goetia" series: it mixes multiple anime / thick-paint digital illustration styles into a single weight, with the prompt determining the bias. MOSS image generation parameters are fully provided.

Images

Example 1024x1536

Reasons for Recommendation

  • Port: The same page already has Anima and Krea 2 Turbo versions, with the entire model page having 546 downloads. The MOSS version lacks version notes, but the model card provides specific MOSS parameters (Base model with a 8-step acceleration LoRA).
  • The author explicitly states it is "highly dependent on prompts." Since one LoRA contains multiple art styles, control is achieved by studying the prompt structures in the example images. 15 example images are visible.
  • The trigger word significantly enhances the style; the MOSS version trigger word removes the @ prefix found in the Anima / Krea 2 versions.

How to Use

  • Trigger word: F0R4SH3X
  • MOSS Base model + 8-step acceleration LoRA, euler + simple sampler, LoRA strength 0.8-1, 8-16 steps, CFG 1-3.
  • Copy the prompts from the example images first, then modify them; do not just write the trigger word.

Risks and Limitations

  • The author states the training images mix sources such as Midjourney, NijiJourney, and Tumblr, including other AI-generated images, so the style sources are not singular. The author admits the quality of various LoRAs in the same series is inconsistent.
  • The license does not allow derivatives; commercial use is limited to generation and on-site rental. Additionally, 1 sensitive-level example images are hidden.

Trigger Word: F0R4SH3X

[A] Genshin Impact Artstyle MOSS · v1

LoRA · New Model · Author Only4u_Art · Published 11-20 19:10 · Downloads 27

A LoRA for the in-game visual style of Genshin Impact: hand-drawn sky textures, Liyue/Mondstadt architecture, soft game lighting, and gold-trimmed tassel clothing. Characters lean towards the cel-shaded rendering seen in the game. It can also serve as a "worldview" LoRA layered beneath character LoRAs.

Images

Example 1792x2688

Reasons for Recommendation

  • New model: The author states it was trained on 115 in-game screenshots (landscapes, towns, interiors, quest scenes, NPCs, character close-ups). The released version is the final epoch of 20 epochs, offering the strongest scene fidelity, while character likeness varies by seed.
  • Detailed instructions: Recommended prompt prefixes, full example prompts, negative prompts, and three writing tips are provided. For example, use "one girl" instead of 1girl to avoid flattening the image into a flat-color anime style.
  • Serves Genshin-style scene images and fan art. The author suggests using character or style LoRAs at full strength and this LoRA at 0.6–0.8 to provide scenes, lighting, and clothing. 12 example images are visible.

How to Use

  • Start the prompt with "OnlyGIstyle, 3D game render, 3d, scenery, genshin ingame artstyle, 3d character, solo," followed by a full sentence description (the trigger word is not registered in the version and must be written manually).
  • Strength: 1.0 when used alone; 0.6–0.8 when stacked with other style or character LoRAs.
  • Negative prompt: illustration, painting, bloom, lens flare, sparkles, blurry, lowres, text, watermark. Use plain descriptions for lighting, such as "warm sunset light," and minimize sparkle particles.
  • Steps, CFG, and sampler are not provided by the author.

Risks and Limitations

  • Copyrighted game art style, fan work. The author admits character fidelity is unstable.
  • Additionally, 1 sensitive-level example images are hidden.

[A] Stellar Blade In-Game Style MOSS · v1

LoRA · New Model · Author Only4u_Art · Published 11-20 15:25 · Downloads 42

A LoRA for the in-game rendering style of Stellar Blade: pushes MOSS towards semi-realism, with softer, more realistic faces, glossy skin and materials, and detailed hair, paired with warm foggy light or cool blue cinematic light. It is not a character LoRA and can be used with any character.

The same author's Tachy character LoRA uses this style for its cover; the two can be viewed together.

Images

Example 1183x1774

Reasons for Recommendation

  • New model: The author clearly states what it cannot do (turn images into actual game screenshots) and what it can do (rendering texture, lighting, materials).
  • Strength is specified in three tiers: 1.0 for maximum realism, 0.7–0.8 for a 50/50 blend with the original art style, and 0.4–0.6 for only slight influence on lighting and materials. It also warns of a tendency towards depth-of-field blur and provides positive/negative prompts to remove the blur.
  • The author states it was trained on 95 in-game images (characters, creatures, scenes), releasing epoch 14 because later epochs followed prompts poorly and had more blur. 8 example images are visible.

How to Use

  • Trigger word: OnlySBstyle (not registered in the version, must be written manually), followed by character and scene descriptions. Write content, not style, in the prompt; words like 3D, realistic, and anime will conflict with the trigger word.
  • For clear backgrounds: Add "sharp detailed background, everything in focus" to the positive prompt and "bokeh, depth of field, blurry background" to the negative prompt.
  • Recommended negative prompt: blurry, lowres, jpeg artifacts, text, logo, watermark, signature, subtitles, game ui, hud. Steps, CFG, and sampler are not provided by the author.

Risks and Limitations

  • Copyrighted game art style, fan work. The author states that atmospheric lighting like dusk, firelight, and dim interiors works best; performance in bright scenes is not specified.
  • Additionally, 2 sensitive-level example images are hidden.

[A] Dizzle's Jupiter · v1

LoRA · New Model · Author Maridizzle · Published 11-20 15:54 · Downloads 25

A conceptual LoRA that "Jupiter-izes" everything: subjects become amber, cream, and burnt-orange marble stripes with storm swirls, and the ground flows with lava-like light. This is the latest addition to Dizzle's space series.

Images

Example 1024x1024

Reasons for Recommendation

  • New model: The author states that stones, skin, fabric, and skies will all be painted with gas giant banding patterns. Examples include a dusk noodle stall under a giant planet, dancers jumping on lava ground, and marble-patterned faces.
  • Serves sci-fi posters, space-themed illustrations, or transforming existing characters and objects into planetary textures. Usage requires only one trigger word.
  • Free, with generation, rental, and derivatives allowed, and no attribution required.

How to Use

  • Add "dz_jupiter" to the prompt, then describe the subject. Strength ranges from 0.9 to 1.0.
  • Steps, CFG, and sampler are not provided by the author.

Risks and Limitations

  • The color palette is locked to oranges, creams, and ambers, making it unsuitable for images requiring original colors.
  • 4 example images are visible; additionally, 3 sensitive-level images are hidden.

Trigger Word: dz_jupiter

[A] Dizzle's Twine · v1

LoRA · New Model · Author Maridizzle · Published 11-20 09:57 · Downloads 31

A material LoRA that "weaves" subjects into twine: characters, clothes, hair, stones, and wings are composed of knotted, woven, and frayed threads, with loose strands drifting at the edges.

Images

Example 1024x1024

Reasons for Recommendation

  • New model: The author provides examples such as a Gothic cathedral and gargoyles woven from twine, a woman in a red dress whose skirt unravels into threads, and a ghost wearing a woven kimono, demonstrating its effect on architecture, characters, and clothing.
  • Serves handmade / fiber art style posters, holiday decoration images, and materialized character images. This is a Dizzle series MOSS special effects LoRA, with usage consistent with the rest of the series.
  • 6 example images are visible. Free, with generation, rental, and derivatives allowed, and no attribution required.

How to Use

  • Add "dz_twine" to the prompt, then describe the subject. Strength ranges from 0.9 to 1.0.
  • Steps, CFG, and sampler are not provided by the author.

Risks and Limitations

  • The description is brief; it is not specified whether partial application (e.g., only clothing) is controllable. Additionally, 1 sensitive-level example images are hidden.

Trigger Word: dz_twine

[A] Friendly Beholders of Mischief for Moss · v1.0

LoRA · New Model · Author Razira · Published 11-20 13:27 · Downloads 27

Generate a concept LoRA for the Dungeons & Dragons Beholder (a floating giant-eye monster), allowing it to wear clothes, grow arms and legs, shoot eye rays, and make expressions.

Image

Example 1728x1728

Reasons to Recommend

  • New model: Specific monster designs that the MOSS base model struggles with. The author explains how body parts are triggered by prompts—dressing it in sleeved clothes makes it grow arms, writing "standing" makes it grow legs, and for the traditional floating look, write "floating over".
  • Serves tabletop gaming / fantasy illustration and humorous character art; the author states the training images span multiple art styles but lack real photos, so the output tends toward illustration.
  • 14 example images are visible.

How to Use

  • Trigger phrase: "An image of a beholder with", followed by your desired Beholder description (clothing, actions, eye rays, mouth expressions).
  • Be careful with wording when describing interactions with other objects, otherwise it may grow unwanted extra parts.
  • The author did not provide strength, steps, or CFG values.

Risks and Limitations

  • Strict license: Only on-site rental is allowed; derivatives are not permitted, and attribution is required; the Beholder image involves Dungeons & Dragons copyright.
  • Additionally, 3 sensitive-level examples have been hidden.

Trigger Words: An image of a beholder with

【A】[MOSS Image] Analog film style · FUJI ProPlus II - SFT

LoRA · New Model · Author: hinablue · Published 11-20 12:02 · Downloads 96

FUJI ProPlus II film style LoRA: Trained on photos taken with this film roll, it applies the film's color rendering to MOSS SFT realistic images, while also affecting some poses and clothing.

Image

Example 1120x1488

Reasons to Recommend

  • New model: Serves film-like portraits, street photography, and lifestyle shots; the author specifies in four languages (Chinese, English, Japanese, Korean) that the impact is on color, as well as some body movements and clothing.
  • The filename indicates it is for MOSS Image SFT; the version name indicates it is the first film roll in the "Analog film style" series.
  • 16 example images are visible; published about 9 hours ago with 96 downloads (indicating attention level only).

How to Use

  • Base model: MOSS Image SFT; no trigger word, effective upon loading.
  • The author did not provide strength, steps, or CFG values.

Risks and Limitations

  • The description is only one sentence, with no comparison images without the LoRA; it alters poses and clothing, which may be unsuitable if you only want color grading.
  • The license does not allow derivatives; commercial use only includes image generation and on-site rental; additionally, 4 sensitive-level examples have been hidden.

【A】Sketch Line — Clean Pencil and Ink Lines on White Paper · MOSS

LoRA · New Model · Author: 858654128784 · Published 11-20 11:44 · Downloads 67

Pencil / ink line sketch LoRA on white paper: Clean lines with hatching to represent light and shadow, no coloring, like a page from an artist's sketchbook; capable of drawing architecture, interiors, landscapes, still life, animals, and figures.

Image

Example 896x1152

Reasons to Recommend

  • New model: Serves line art sketches, architectural sketches, figure sketches, and illustration underlays; the output is achromatic line drawing.
  • Usage is clearly documented: Start the sentence with "sketchline" followed by what to draw; you can add "pencil line drawing on white paper, hatching"; the strength range is also provided.
  • The most permissive license (image generation, rental, sale, derivatives allowed, no attribution required).

How to Use

  • Start the prompt with "sketchline", followed by the scene content; you can add "pencil line drawing on white paper, hatching" (the trigger word is not registered in the version, so it must be written manually).
  • LoRA strength 0.8–1.0; the author did not provide steps or CFG values.

Risks and Limitations

  • Most examples are sensitive-level: 5 are visible, and 15 are hidden, leaving few reference images for non-figure subjects.

【A】BBox Studio · v1.5

Workflow · Version Update · Author: plk · Published 11-20 10:28 · Downloads 37

v1.5 of the standalone desktop mini-app BBox Studio: Drag boxes on the canvas to edit MOSS Image region layout prompts and export JSON, independent of ComfyUI; this version mainly involves interface polishing.

It does the same thing as the previous ComfyUI drag-box layout node, the difference being that this runs independently of ComfyUI.

Image

Example 1322x892

Reasons to Recommend

  • Version update: Improved the appearance and behavior of the "Lock Position / Size" and "Lock Aspect Ratio" switches; the main menu, font, and color settings interface have also been polished; Meta list items can be edited or deleted directly via right-click.
  • The size of the description text box is now remembered, and reopening the program basically restores the last state (canvas resolution, grid type, snap toggle, window position, interface color scheme, and font).
  • The tool itself: Written in pure Python, the author claims it only requires Python 3 with no other dependencies, just run the script; the code is public on the author's GitHub, and the entire model page has 348 downloads.

How to Use

  • Download the zip, unzip it, and run the script with Python 3; draw boxes on the canvas, write descriptions for each box, and paste the exported JSON into MOSS's layout prompt.
  • Image generation parameters are irrelevant to this tool (it only generates layout JSON).

Risks and Limitations

  • This is a program, not weights; code distributed as a zip package, you can check against the GitHub repository before running.
  • This version has no new core features, so the upgrade benefit for users already on v1.4 is limited; there are only 2 screenshots.

【A】Sri Lankan Kandyan (bridal saree) · v1.0

LoRA · New Model · Author: instagothic · Published 11-20 12:41 · Downloads 30

Sri Lankan Kandyan-style white bridal saree LoRA: Generates not just the clothing, but also the slicked-back hairstyle and the entire set of jewelry together.

Image

Example 1248x1664

Reasons to Recommend

  • New model: Serves weddings, ethnic costumes, and cultural-themed portraits; one LoRA bundles clothing, hairstyle, and jewelry together, so you don't need to describe them separately.
  • Strength range and CFG are provided, and two trigger words are registered.
  • Permissive license (image generation, rental, sale of fusions, derivatives allowed), free.

How to Use

  • Trigger words: "kan@yan" plus "sri lankan bridal saree".
  • Strength 0.85-0.95, CFG 1.5 (written as CGF on the card); the author did not provide steps or sampler.

Risks and Limitations

  • The model card is only four lines, only 2 example images are visible, and 2 sensitive-level images are hidden.
  • Covers only one style: the white Kandyan bridal saree.

Trigger Words: kan@yan, sri lankan bridal saree

【A】Better Dusk & Twilight— Golden Hour & Blue Hour · MOSS

LoRA · New Model · Author: 858654128784 · Published 11-20 20:38 · Downloads 7

Dusk lighting LoRA, with two lighting options: Golden Hour (low-angle sun outlining hair and body, amber and peach clouds, long shadows) and Blue Hour (deep blue sky, a strip of pink at the horizon, warm lights).

Apply the Golden Hour phrase and a self-written Blue Hour phrase to the same scene respectively to see if both lights can be switched stably.

Image

Example 1152x1536

Reasons to Recommend

  • Serves outdoor portraits and mood shots like beaches, wheat fields, rooftops, balconies, and docks; the author states the MOSS version and the Krea2 version use the same training data.
  • The author provided a complete prompt structure: First write the trigger word, then the number of people and each person's appearance and clothing, then actions and scene, and finally append the light effect description used during training.
  • Strength is provided; permissive license (image generation, rental, sale, derivatives allowed, no attribution required), free.

How to Use

  • Start the prompt with "betterduskhour" (the trigger word is not registered in the version, so it must be written manually), followed by the number of people, each character, actions, and scene.
  • For Golden Hour, append the author's full sentence at the end: "Golden hour backlight catches the hair and edges, the low sun sits just above the horizon, amber and peach clouds and long shadows show the last minutes before sunset."
  • LoRA strength 1.0; the author did not provide steps, CFG, or sampler.

Risks and Limitations

  • The author only provided the ending sentence for Golden Hour; the sentence to append for Blue Hour is not written.
  • The prompt template assumes there are people in the scene; it is not explained if it can be used for pure landscapes.
  • Published less than 1 hour ago; 3 example images are visible, and 15 sensitive-level images are hidden.

【A】Illuminated Manuscript Fables|Medieval Illuminated Manuscript Fables · MOSS

LoRA · Ported · Author: 858654128784 · Published 11-20 15:06 · Downloads 37

Medieval illuminated manuscript illustration style LoRA: Gold leaf, floral borders, flat characters, depicting fables and legend scenes primarily featuring adult male protagonists; ported from the Krea 2 version to MOSS.

Image

Example 832x1216

Reasons to Recommend

  • Ported: The Krea 2 version is already on the same page, with 287 downloads for the entire model page; the MOSS version has no version notes, and the differences from the Krea 2 version are not written by the author.
  • Style features are described specifically (gold leaf, decorative borders, flat characters), serving book illustrations, tabletop game card faces, and retro posters with bordered narrative scenes.
  • 5 example images are visible, and the trigger word is provided.

How to Use

  • Trigger word: "illuminatedmalefable" (not registered in the version, must be written manually).
  • The author did not provide strength, steps, or CFG values.

Risks and Limitations

  • The subject matter is limited to "fables and legends of adult males"; it is not explained if it can be applied to female characters or non-narrative scenes.
  • Commercial license only includes on-site rental; additionally, 2 sensitive-level examples have been hidden.

Worth Watching (9)

【B】V67 Color Palette · V1 MOSS

LoRA · Ported · Author: angelomaiota · Published 11-20 16:23 · Downloads 22

Color palette LoRA: Unifies the image into a set of contrasting colors including deep indigo, electric cyan, acidic lime green, warm coral, and cream white, leaning towards retro-future / cyberpunk neon vibes; the MOSS version of a multi-base-model page.

Image

Example 1024x1024

【B】Takamichi Style [MOSS] · v1.0

LoRA · New Model · Author: Franklin_Sports · Published 11-20 11:24 · Downloads 122

MOSS-style LoRA of illustrator Takamichi's art style, taking the "average art style" of his works from the 2000s to the mid-2010s.

Image

Example 1024x1536

【B】Flat Toon · MOSS

LoRA · Ported · Author: xFennec777 · Published 11-20 16:52 · Downloads 44

Flat cartoon style LoRA sourced from comics, TV, and movies; the Krea 2 version on the same page leans towards Disney / Pixar; the MOSS version has no description at all.

Image

Example 1248x1680

【B】Moss/Anima-style-Oily Wet Skin (nai5th) · Moss-base-V1.0

LoRA · Ported · Author: hardworkingsaltedfish · Published 11-20 17:35 · Downloads 110

Art style LoRA with oily reflective skin texture (the tenth in the author's "nai5" series), trained separately for MOSS Base and Anima versions using the same training set; the author states the style leans towards reflective shiny skin and curvy body types.

Image

Diagram 1086x1448

【B】Tachy (Stellar Blade) MOSS · v1

LoRA · New Model · Author Only4u_Art · Published 11-20 17:13 · Downloads 27

Character LoRA for Tachy from Stellar Blade: The face, long black curly hair, and signature combat outfit can be invoked separately; the combat outfit is tagged by name, so switching to other clothing won't lock it in.

Image

Example 1792x2688

【B】Deal With It: Sunglasses (Meme) [IL] · MOSS

LoRA · Port · Author CitronLegacy · Published 11-20 17:03 · Downloads 36

LoRA based on the “Deal With It” sunglasses meme concept: Puts pixelated sunglasses on characters with a smug expression; can include meme elements like throwing money or giving the middle finger; the Illustrious version is ported to MOSS.

Image

Example 1024x1024

【B】Child of Light (Ubisoft Video Game) Style [Anima] · moss sft v1.0

LoRA · Version Update · Author Connor233 · Published 11-20 19:46 · Downloads 9

The MOSS SFT trained version of the LoRA for the art style of Ubisoft's game Child of Light, listed alongside the MOSS Base trained version released yesterday.

Image

Example 2048x1536

【B】Japanese Wolf (Kemono Friends) · MOSS SFT JSON

LoRA · Version Update · Author quintronanthony265 · Published 11-20 17:55 · Downloads 19

MOSS SFT character LoRA for the Japanese Wolf from Kemono Friends, trained with JSON format annotations; a standard annotation version was also released on the same day.

Image

Example 1440x1024

【B】Pale Fox (Kemono Friends) · MOSS SFT JSON

LoRA · Version Update · Author quintronanthony265 · Published 11-20 18:22 · Downloads 15

MOSS SFT character LoRA for Pale Fox from Kemono Friends; this version's training annotations have been changed to JSON format; a standard annotation version was also released on the same page on the same day.

Image

Example 1440x1024

Included Only (44)


Data: Civitai public API (base model MOSS Image), inclusion window 11-20 08:31 → 11-20 21:01 (Beijing Time). Recommendations have been reviewed by editors; download counts only reflect attention.

Civitai

Explore the latest AI Radar updates