Zhenzhen's AI Radar · MOSS Image · 2026-11-21 Evening Report
This issue features 54 new versions: Strongly Recommended 0 · Recommended 9 · Worth Watching 16 · Included Only 29
Editor's Note This issue reviewed 57 new MOSS versions, with 9 rated A, 18 rated B, and 30 rated C, with no exclusions. Top pick: Genshin Impact in-game art style v2.0: removed old trigger words, uses strength as a slider, produces full game style when used alone at 1.0, and handles only scene and lighting when stacked with character LoRAs. Other highlights include 1561 Nuremberg Sky Battle alternate history scenes, Zenless Zone Zero toon rendering, Flat Color v4 flat color style, Ken Sugimori Pokémon original art style, and 2010 vintage overexposed film snapshots. The HaDeS metadata workflow only adds parameter logging, downgraded to B; the AI Babe Pack 8 parts and most copyrighted character LoRAs are listed as C.
Recommended (9)
[A] Genshin Impact Artstyle MOSS · v2.0
LoRA · Version Update · Author Only4u_Art · Published 11-21 19:02 · Downloads 24
v2.0 of the Genshin Impact in-game visual style LoRA: Teyvat landscapes, towns, interiors, and in-game 3D character rendering style; this version no longer uses trigger words, relying on strength to control style intensity.
Run the same prompt at strengths 0.6 / 1.0 / 1.5 to see if the "slider" truly only adjusts style intensity. (Editor downgraded to A: steps, CFG, and sampler not provided, training epoch description is inconsistent)
Images

Reasons for Recommendation
- Version update: The author states that the old trigger word OnlyGIstyle pulled characters toward flat 2D; removing it makes the 3D game feel more obvious, so it takes effect upon loading. Switched to using strength as a slider: 0.6 for subtle, 1.0 for full game style (recommended), 1.2–1.5 to override other styles in the prompt
- The author claims the strongest aspect is the scenes: hand-drawn feel skies, dense vegetation, Liyue and Mondstadt architecture, soft game lighting, and clothing details like gold-trimmed tassels; characters approach in-game cel shading but vary with the seed
- Can be used as a "worldview LoRA" for stacking: Use character or style LoRAs at full strength, set this to 0.6–0.8 to handle only scenes, lighting, and clothing; 7 example images are visible, free
How to Use
- Do not write trigger words; start the prompt with 3D game render, 3d, scenery, genshin ingame artstyle, 3d character, solo, then describe the subject in full sentences (write one girl instead of just 1girl)
- Use strength 1.0 when used alone; use 0.6–0.8 when mixing with other style/character LoRAs; negative prompt can include illustration, painting, bloom, lens flare, sparkles, blurry, lowres, text, watermark
- For lighting, use plain descriptions like warm sunset light, soft overcast daylight to better match game visuals; steps, CFG, and sampler were not provided by the author
Risks and Limitations
- Training epoch descriptions are contradictory: version notes say 30 of 30, while the model card training section says 20 of 20; the registered trigger word field still contains the old word plus "for testing," do not copy it into prompts
- Game copyright art style (author notes it is fan-made and unrelated to miHoYo); additionally, 3 sensitive-level example images are hidden
Trigger Words: OnlyGIstyle DO NOT USE UNLESS FOR TESTING
[A] DC Nuremberg 1561 · v1.0 [MossBase]
LoRA · New Model · Author dkpc69 · Published 11-21 13:56 · Downloads 50
An alternate history sci-fi LoRA themed around the 1561 Nuremberg "Sky Battle" sighting event: 16th-century cities and soldiers, paired with giant spaceships, metal spheres, flying monoliths, and time portals, capable of generating both close-up and wide-shot battlefield scenes.
Starting with example images that include workflows is the easiest approach: drag them directly into ComfyUI to see the author's strength and steps.
Images

Reasons for Recommendation
- New model: Trained on MOSS Base (99 images, 7000 steps, rank 32), the author claims it also works on SFT; reference material is a Nuremberg event trailer by an X user
- Serves alternate history, ancient vs. alien invasion, and cinematic poster-style war scenes; examples cover character close-ups and large scenes
- 14 example images are visible, the author claims each includes prompts and ComfyUI workflows, which can be dragged in and modified directly
How to Use
- Start the prompt with DCNRMBRGINCD, (the author registered it with a comma), then describe 16th-century cities, flying objects in the sky, and characters
- Strength, steps, and CFG were not provided by the author; take them from the workflows embedded in the example images
Risks and Limitations
- Training material was taken from someone else's trailer, authorization status is not stated by the author; attribution to the author's page is required for fusion or reposting to other platforms (license requires attribution)
- Additionally, 6 sensitive-level example images are hidden; the text description lacks parameters
Trigger Words: DCNRMBRGINCD,
[A] V67 Cloud Dancer · V1 MOSS
LoRA · Port · Author angelomaiota · Published 11-21 15:40 · Downloads 23
A "white on white" realistic style LoRA: themed around the Pantone Color of the Year Cloud Dancer, making interiors, people, white animals, and snowscapes feature white as the protagonist, distinguishing layers through texture and soft light; a MOSS port of the Krea 2 / Z-Image Turbo version.
(Editor downgraded to A: parameters are all written for other base models, the MOSS version has only 3 example images)
Images

Reasons for Recommendation
- Port: The same page already has Krea 2 and Z-Image Turbo versions, with 408 downloads for the entire model page; the MOSS version filename is moss-image-base-epoch-02, trained on Base, differences from other versions are not written by the author
- Serves minimalist interiors, high-key studio portraits, white-haired/white-clothed characters, white pets, snowfields, and other "low saturation, all white" images; the author provided a method of placing only one dark accent (like a black sheep) within a white subject
- Prompt writing is detailed: start with cloud dancer followed by 2–3 synonymous terms, specify materials (silk, cashmere, fur, marble) and lighting; 3 example images are visible, free, with the most permissive license
How to Use
- Start the prompt with (cloud dancer:1.1), then pick 2–3 terms from white on white, albino style, platinum white, snowy minimalism, pale ethereal, etc., followed by subject, material, and soft light
- The author's Quick Start is LoRA strength 0.7, 1024x1024 or 1344x896; but these are written for Z-Image Turbo / Flux, steps, CFG, and sampler for the MOSS version were not provided by the author
- Exclude saturated colors (red, blue, green, etc.) in the negative prompt, otherwise the white tone will be broken
Risks and Limitations
- The entire model card is written for Z-Image Turbo / Flux (CFG 4.5, 30 steps, etc. are not MOSS parameters), the MOSS version has no separate instructions
- The filename indicates epoch 2, it is not stated whether it is an early checkpoint; large parts of the model card are concept promotion and Buzz cover bounties
Trigger Words: (cloud dancer:1.1), white on white, albino style, platinum white, snowy minimalism, pale ethereal, arctic whisper, frost bloom, pearl tone, milk drift, chalk veil, ghost silk
[A] Flat Color - Style · v4.0 [moss]
LoRA · Port · Author motimalu · Published 11-21 18:15 · Downloads 42
The MOSS version of the veteran "flat color, no lineart" style LoRA: removes outlines, uses only flat color blocks, and barely expresses depth, similar to flat illustration.
The tagging includes region box formats, you can try using it with MOSS's layout prompts.
Images

Reasons for Recommendation
- Port: The same page already has over ten versions including Anima, Z-Image, Flux.2 Klein, Qwen, Illustrious, Wan, etc., with 57122 downloads for the entire model page; the author says this is his fixed dataset when testing LoRA training on new models
- MOSS version training details are written: diffusion-pipe branch, 1024^2 resolution, MOSS timestep sampling implemented by Kohaku Blueleaf, tagging mixes tags, natural language, and region box formats
- Serves flat illustration, posters, and compositions with negative space; recommended tags are provided, 4 example images are visible
How to Use
- Trigger words: flat color, no lineart; the author recommends adding blending, negative space, and "[some color] background"
- Strength, steps, CFG, and sampler were not provided by the author
Risks and Limitations
- The version name is v4.0, but the filename is moss_flat_color_v5, the author did not explain; the MOSS version has no separate effect description
- Additionally, 3 sensitive-level example images are hidden; the license does not allow selling
Trigger Words: flat color, no lineart
[A] Ken Sugimori - Pokemon - Style MOSS LORA · v1.0
LoRA · New Model · Author Konan · Published 11-21 10:32 · Downloads 48
Ken Sugimori (original official illustrator for the first generation of Pokémon) style LoRA: the retro 2D illustration style of 90 era Pokémon game original art, trained directly on MOSS.
Images

Reasons for Recommendation
- New model: Author Konan's style template, provided a full starting prompt sentence (sugimori_style + classic Ken Sugimori art style + 1990 era Pokémon game art + 2D retro illustration)
- Serves Pokémon fan art and character design sheets in retro game art style; strength range is specified as 0.8-1.3
- 3 example images are visible; free, license allows output and rental
How to Use
- Start the prompt with sugimori_style, authentic classic Ken Sugimori art style, 1990s Pokémon game art style, 2D retro illustration of, followed by the content to be drawn
- Strength 0.8-1.3; steps, CFG, and sampler were not provided by the author
Risks and Limitations
- Imitates a real artist's style, involves Nintendo/Pokémon copyright, authorization is not stated by the author
- Additionally, 5 sensitive-level example images are hidden; the latter half of the model card is donation and commission ads, style characteristics are not described in text
Trigger Words: sugimori_style, authentic classic Ken Sugimori art style, 1990s Pokémon game art style, 2D retro illustration of
[A] Zenless Zone Zero In-Game Style - MOSS · v1
LoRA · New Model · Author Only4u_Art · Published 11-21 17:05 · Downloads 38
Zenless Zone Zero in-game visual style LoRA: Pulls MOSS's default semi-realistic look toward the game's 3D cel-shaded cartoon rendering—cel shading, clean outlines, with scenes featuring the game's streets and interiors.
Running the same prompt across 0.6 / 1.0 / 1.5 intensity levels allows you to directly observe the transition from semi-realistic to full cartoon rendering.
Image

Reasons for Recommendation
- New model: The author states that cartoon rendering is far from MOSS's default semi-realistic style; the first few rounds were too weak, so the final version uses round 30 of the 30 rounds; training materials consist of approximately 100 in-game screenshots and renders (characters, NPCs, scenes), without using promotional images or works by other artists
- Intensity is used as a slider, with three levels clearly defined: 0.6 for a slight touch, 1.0 for full game style (recommended), and 1.2–1.5 for when the prompt is pulled away by other styles
- Suitable for game fan art, 2D 3D character portraits, and neon street scenes; the author states it can be stacked with character LoRAs; 6 examples are visible, it is free, and the license allows for image generation and selling fused works
How to Use
- No trigger word is required; it takes effect upon loading (the author states that adding a trigger word makes it flatter and more like 2D); start intensity from 1.0
- Start the prompt with 3D game render, 3d, cel shading, then specify the character and scene in detail; if the prompt is too short, it will slide back to MOSS's semi-realistic style
- Steps, CFG, and sampler are not provided by the author
Risks and Limitations
- Training materials are game screenshots, involving game copyright; the author themselves states it will not always perfectly match the game
- Additionally, 7 sensitive-level examples have been hidden; released approximately 4 hours ago
[A] Soft Film 2010 — Overexposed Pastel Snapshot, Orange Light Leak · MOSS
LoRA · Port · Author 858654128784 · Released 11-21 15:13 · Downloads 56
2010s film snapshot style LoRA: Overexposed, washed-out pastel colors, lifted cyan-gray shadows, creamy highlights, fine grain, with a faint orange light leak along the left edge of the image.
Image

Reasons for Recommendation
- Port: The same page already has a Krea 2 version, with 289 downloads for the entire model page; the MOSS version has no separate description, following the prompt structure of the full page
- Suitable for film-like realistic images of daily life scenes: The author gives examples such as rooftops with laundry, sea walls, railway tracks, and old apartments, stating that multi-person frames are also stable
- The prompt structure is provided specifically, and the full-sentence effect description used during training is given, which can be copied directly; 5 examples are visible, it is free, and the license is permissive
How to Use
- Start with softfilm2010, followed by the number of people and each person's settings (ethnicity, age, hairstyle, clothing), then describe the action and scene
- End with the training effect sentence: An overexposed 2010s film snapshot with washed pastel colors, lifted cyan gray shadows, creamy highlights, milky overexposure, fine grain and a faint orange light leak along the left edge.
- The trigger word is not registered in the version and must be written manually; intensity, steps, and CFG are not provided by the author
Risks and Limitations
- The light leak is fixed on the left edge, and the tone is fixed to pastel overexposure, making it unsuitable for images requiring normal exposure
- Additionally, 8 sensitive-level examples have been hidden; the differences between the MOSS version and the Krea 2 version are not written by the author
[A] GTA6 Cinematic Style · v1.0
LoRA · New Model · Author un0wn · Released 11-21 14:24 · Downloads 36
GTA6 trailer / promotional image style LoRA: Glossy, highly detailed cinematic 3D rendering, shallow depth of field, strong rim lighting, high-saturation color grading, covering everything from pink-purple neon night scenes to Florida noon sunlight and golden hour.
Image

Reasons for Recommendation
- New model: Trained on MOSS Base, the training set consists of 28 widescreen images (1920×1080 and 3840×2160), paired with long natural language descriptions; topics cover character close-ups, car chases and gunfights, nightclubs, neon gas stations, swamp crocodiles, etc.
- The prompt writing method is clearly explained: Write full sentences rather than tags after the trigger word, specify lighting and colors, and copying the closing sentence from the training description at the end can strengthen the style; widescreen 16:9 best fits the training data
- Suitable for crime / urban theme posters, game fan promotional images, and cinematic character close-ups; 6 examples are visible
How to Use
- Start the prompt with gt6, followed by a full scene description, specifying light colors (e.g., hot pink and cyan lighting, golden hour)
- Add The style is a glossy, highly detailed cinematic 3D render with shallow depth of field, conveying a tense mood. at the end; use 16:9 for aspect ratio
- The trigger word is not registered in the version and must be written manually; intensity, steps, and CFG are not provided by the author
Risks and Limitations
- Training materials are Rockstar trailers and promotional images, involving game copyright; the license is strict: derivatives are not allowed, attribution is required, and commercial use only includes on-site rental
- Additionally, 3 sensitive-level examples have been hidden; training topics often involve guns and crime scenes
[A] Advertising Pylons of Shopping Mall · MOSS
LoRA · Port · Author 858654128784 · Released 11-21 09:53 · Downloads 43
Concept LoRA for standalone advertising pylons at shopping mall entrances: Old-style small-town stacked lightboxes, modern lightboxes, vertical LED towers, and slanted metal signs, covering daytime, dusk, and night scenes; a MOSS port of the Krea 2 version.
Can be combined with MOSS's text-writing ability: Specify store names in quotes for each sign in the prompt to see if the text on the pylon follows along. (Editor upgraded to A: Usage and image structure are clearly written, there are non-sensitive examples, only parameters are missing)
Image

Reasons for Recommendation
- Port: The same page already has a Krea 2 version, with 246 downloads for the entire model page; the MOSS version has no separate description, following the same model card
- Suitable for urban street scenes, commercial scene concept art, and "pylon advertising" elements in posters and sign design; 3 examples are visible, it is free, allows for image generation, selling, and derivatives, with no attribution required
How to Use
- The author has not registered a trigger word; write the prompt directly following the structure of "pylon type + time of day + central portrait ad + upper/lower sign names + pedestrians and bicycles at the bottom"
- Intensity, steps, CFG, and sampler are not provided by the author
Risks and Limitations
- The MOSS version's version notes are empty, and the differences from the Krea 2 version are not written by the author
- Additionally, 5 sensitive-level examples have been hidden; there is no explanation on whether sign text can be written correctly according to the prompt
Worth Watching (16)
[B] HaDeS MOSS Suite Workflows · Metadata v0.9.3
Workflow · Version Update · Author _HaDeS_ · Released 11-21 19:24 · Downloads 18
The newly added "Metadata" workflow in the HaDeS MOSS Suite workflow page: Follows the Turbo 8 steps + x2 upscale process, but allows each save node to write its own segment's parameters, so the 1024 original image, x2 upscaled image, and low-redraw refined image each remember how they were generated.
Image

[B] SoulTide 灵魂潮汐 Erinnern 爱瑞妮 DiT · MOSSTagFlowV1.0
LoRA · Port · Author zwa73 · Released 11-21 18:21 · Downloads 16
Character LoRA for Erinnern from the mobile game SoulTide (a beast-girl with squirrel ears and tail, pink eyes, and a single braid), a MOSS port of the Anima version, with separate trigger words for two outfits.
Image

[B] NexBlend 13 Asian Realistic LoRA - MOSS Image · v1.0
LoRA · New Model · Author NexBlend · Released 11-21 12:01 · Downloads 71
East Asian realistic LoRA with a Korean aesthetic bias: The author states it enhances Korean facial features, as well as Korean street scenes, signs, hanoks, and traditional clothing differentiated by status from the Joseon Dynasty.
Image

[B] Zeon (zzeeonn) - Style MOSS LORA · v1.0
LoRA · New Model · Author Konan · Released 11-21 10:13 · Downloads 77
Anime art style LoRA by artist Zeon (zzeeonn), an art style template by author Konan, trained directly on MOSS.
Image

[B] Mikoshii's Ultra Lightweight MOSS Workflow - txt2img, img2img, Model Merging, Auto-prompting and Image Upscaling · v1.1
Workflow · Version Update · Author Mikoshii · Released 11-21 17:31 · Downloads 29
Mikoshii's all-in-one MOSS ComfyUI workflow v1.1: Switches control text-to-image, image-to-image, real-time model merging, auto-prompting from images, and upscaling; this version replaces the upscaler with the faster LC Relay VSR.
Image

[B] poderdohokuto (conto) - MOSS Image v1 · v2.0
LoRA · New Model · Author inpaint · Released 11-21 10:38 · Downloads 63
Art style LoRA v2.0 by artist poderdohokuto (conto), the author states v1 was severely overexposed, so it was retrained after expanding the dataset; the file is only 7 MB.
Image

[B] moss-北岸网图 · v1.0
LoRA · New Model · Author 949141972741 · Released 11-21 15:34 · Downloads 60
LoRA for improving realistic portrait appearances: Trained on MOSS Base, the author states generated images have higher attractiveness and realism for characters, but do not lock onto a specific face; tags lean towards atmospheric photos of Asian women.
Image

[B] Hollywood Technicolor Glamour Three-Strip Color · MOSS
LoRA · New Model · Author 858654128784 · Released 11-21 17:59 · Downloads 40
Hollywood Three-Strip Technicolor old movie style LoRA: High-saturation colors, warm studio lighting with rim lighting, hand-painted sets, fine grain, with black bars on the sides.
Image

[B] Yixi Style1 | MOSS-Image · v1.0
LoRA · New Model · Author Leafyixi · Released 11-21 10:57 · Downloads 21
"yixi style" anime art style LoRA (tags include anime coloring, soft shadows, thin line art); the author explicitly states it was not trained but is a weight merge of several public LoRAs.
Image

【B】SoulTide Cecilia DiT · MOSSTagFlowV1.1
LoRA · Port · Author zwa73 · Published 11-21 18:27 · Downloads 14
Character LoRA for Cecilia from the mobile game Soul Tide (dark skin, black-and-white twin braids, horned monster girl), a MOSS port of the Anima version, with a separate trigger word for each outfit.
Image

【B】SoulTide Superbia DiT · MOSSTagFlowV1.1
LoRA · Port · Author zwa73 · Published 11-21 16:52 · Downloads 21
Character LoRA for Superbia from the mobile game Soul Tide (white drill-shaped long hair, star-shaped pupils, small crown), a MOSS port of the Anima version, with trigger words for casual and Valentine's outfits.
Image

【B】SoulTide Alisa DiT · MOSSTagFlowV1.1
LoRA · Port · Author zwa73 · Published 11-21 16:41 · Downloads 23
Character LoRA for Alisa from the mobile game Soul Tide (gray-white twin braids, wearing glasses), a MOSS port of the Anima version, with trigger words for maid outfit, casual outfit, and wedding dress.
Image

【B】oval face slider · MOSS
LoRA · Port · Author ghostpaint · Published 11-21 14:53 · Downloads 118
Oval face slider LoRA: uses positive and negative strength to adjust facial features toward or away from an oval shape; a MOSS port of the Anima and Illustrious versions.
Image

【B】Kigurumi Onesie: Absol Pokémon [Krea2/MOSS] · MOSS_v1.0
LoRA · Port · Author AuthentiQ · Published 11-21 12:37 · Downloads 17
Kigurumi onesie clothing LoRA featuring the Pokémon Absol design, dressing realistic characters in this hooded onesie; a MOSS port of the Krea 2 version.
Image

【B】Kigurumi Onesie: Bonsly Pokémon [Krea2/MOSS] · MOSS_v1.0
LoRA · Port · Author AuthentiQ · Published 11-21 10:53 · Downloads 20
Kigurumi onesie clothing LoRA featuring the Pokémon Bonsly design, dressing realistic characters in this hooded onesie; a MOSS port of the Krea 2 version.
Image

【B】RedAnts · RedAnts
LoRA · New Model · Author PanPDX · Published 11-21 15:32 · Downloads 27
Red ant concept LoRA: makes swarms of red ants crawl over people, the ground, and object surfaces.
Image

Included Only (29)
- Evie (Stellar Blade: Blood Rain) - MOSS · v1 — LoRA: Character LoRA for Evie, the protagonist of the new Stellar Blade title Stellar Blade: Blood Rain (short dark brown hair covering one eye, freckles, mechanical arm, white tight combat suit), trained on MOSS.

- [MOSS] Virtual Influencer - xuenhui · xuenhui v1.0 — LoRA: Realistic character LoRA for the original virtual influencer xuenhui (22-year-old college student, long straight black hair, cool temperament); the model card is mostly a character backstory.

- my comfyui workflow · Moss_Image_workflow — Workflow: MOSS version of a personal ComfyUI workflow collection; the version note is just one line: added Krea 2 secondary sampling, LLM reverse inference for layout control, and SeedVR2 upscaling.

- Vladimir Matyukhin Style MOSS · v1.0 — LoRA: Art style LoRA for artist Vladimir Matyukhin; the model card contains only two sentences: the artist's name and the recommended strength.

- Maho Hiyajo · MOSS-Image E9 — LoRA: Character LoRA for Maho Hiyajo from Steins;Gate 0, a MOSS port of the Illustrious / Pony version, with very comprehensive MOSS image generation parameters.

- Kurisu Makise · MOSS-Image E9 — LoRA: Character LoRA for Kurisu Makise from Steins;Gate, a MOSS port of the Illustrious / Pony version, using the same MOSS training and image generation parameters as the author's Maho Hiyajo LoRA.

- Maomao - Kusuriya no Hitorigoto - The Apothecary Diaries - MOSS LORA · v1.0 — LoRA: Character LoRA for Maomao from The Apothecary Diaries (dark green hair, straight bangs, two thin braids, freckles), with prompts for two outfits (casual wear and tea ceremony dress), trained on MOSS.

- Wakaba_Mutsumi_MOSSimage_lora · v1.0 — LoRA: Character LoRA for Mutsumi Wakaba from BanG Dream! Ave Mujica, trained on MOSS-SFT, with full item-by-item descriptive prompts provided for four outfits: stage costume, summer uniform, autumn uniform, and casual wear.

- Adriana · MOSS Image — LoRA: Character LoRA for the AI-generated fictional woman Adriana, a MOSS port of the Krea 2 version; the model card still contains the Krea 2 description.

- Moss Image AI Babe Pack ( Part 02 ) by Sarcastic TOFU · MI_Babe_Britania_v1 — LoRA: Character LoRA for the AI-generated realistic woman Britania (English persona, ivory skin, thick curly chestnut brown hair, sapphire blue eyes), a new character in the second part of the AI Babe Pack.

- Misty: Team Rocket | カスミ ロケット団だん (Pokemon) [MOSS & IL] · MOSS — LoRA: Character LoRA for Misty from Pokémon wearing a Team Rocket uniform (white short jacket, white skirt, long gloves, black knee-high socks), a MOSS port of the Illustrious version.

- Moss Image AI Babe Pack ( Part 03 ) by Sarcastic TOFU · MI_Babe_Evanthia_v1 — LoRA: Character LoRA for the AI-generated realistic woman Evanthia (Greek persona, olive skin, chestnut hair, blue-gray eyes), a new character in the third part of the AI Babe Pack.

- Asia Argento (High School DxD) - MOSS · v1 — LoRA: Character LoRA for Asia from High School DxD, trained on MOSS, learning her face, hairstyle, summer/winter school uniforms, and two sets of nun habits (black and white).

- Moss Image AI Babe Pack ( Part 03 ) by Sarcastic TOFU · MI_Babe_Erkhjin_v1 — LoRA: Character LoRA for the AI-generated realistic woman Erkhjin (Mongolian persona, warm bronze skin, dark long braided hair, golden brown eyes), a new character in the third part of the AI Babe Pack.

- Moss Image AI Babe Pack ( Part 02 ) by Sarcastic TOFU · MI_Babe_Haneul_v1 — LoRA: Character LoRA for the AI-generated realistic woman Haneul (Korean persona, fair skin, smooth black hair, dark brown eyes), a new character in the second part of the AI Babe Pack.

- Moss Image AI Babe Pack ( Part 01 ) by Sarcastic TOFU · MI_Babe_Sueno_v1 — LoRA: Character LoRA for the AI-generated realistic woman Sueño (Latina persona from Brazil or Venezuela, light brown skin, smooth black hair, gray eyes), a new character in the first part of the AI Babe Pack.

- Moss Image AI Babe Pack ( Part 04 ) by Sarcastic TOFU · MI_Babe_Alma_v1 — LoRA: Character LoRA for the AI-generated realistic woman Alma (Mexican persona, warm bronze skin, long black wavy hair, golden brown eyes), a new character in the fourth part of the AI Babe Pack.

- Moss Image AI Babe Pack ( Part 04 ) by Sarcastic TOFU · MI_Babe_Anima_v1 — LoRA: Character LoRA for the AI-generated realistic woman Anima Segreta (Italian persona, olive-tinted fair skin, dark chestnut hair, emerald green eyes), a new character in the fourth part of the AI Babe Pack.

- Moss Image AI Babe Pack ( Part 01 ) by Sarcastic TOFU · MI_Babe_Engel_v1 — LoRA: Character LoRA for the AI-generated realistic woman Engel (German-Nordic persona, fair skin, blonde hair, blue eyes), a new character in the first part of the AI Babe Pack.

- A style lora by Ryia | 2 · v1.3 — LoRA: Anime style LoRA trained by the author Ryia using images generated with NovelAI v5, MOSS third version; the author claims it was retrained with different parameters for greater stability.

- 𝕮𝖆𝖋𝖊 Style (Cafe Style) [MOSS] · v1.0 — LoRA: Anime style LoRA, data taken from NovelAI-generated images in posts by Tieba user 𝕮𝖆𝖋𝖊, no trigger word required; the author provides no textual description of the style characteristics.

- mojo_moss · v1.0 — LoRA: Style LoRA imitating the art style of artist mojo; the model card contains only the single sentence "Art style of artist mojo".

- ningenmame_moss にんげんまめ · v1.0 — LoRA: Anime style LoRA imitating the art style of artist にんげんまめ (ningenmame); the model card contains only the single sentence "LoRA in the style of artist にんげんまめ".

- megauber-EG · MOSS Image — LoRA: Female styling style LoRA for "cute / emo / gamer / streamer" aesthetics; the model card contains only this line of tag-style description.

- Oyomechan(Batta) · MOSS SFT — LoRA: Character LoRA for Oyomechan (a fox-girl married woman character with blonde hair and fox ears) from artist Batta's works, a MOSS port of the Illustrious version, trained on MOSS SFT.

- yomuko · v1.0 (moss) — LoRA: Character LoRA for the cat-eared girl yomuko (black hair, black eyes, cat ears) from the manga yani_neko, a MOSS port of the Anima / Illustrious version.

- Ookami Ranka (Seton Gakuen) · MOSS SFT — LoRA: Character LoRA for the wolf-girl Ookami Ranka (pink hair, wolf ears) from Murenase! Seton Gakuen (Seton Gakuen), a MOSS port of the Illustrious version, trained on MOSS SFT.

- imoko (yani neko) · v1.0 (moss) — LoRA: Character LoRA for the silver-haired red-eyed cat-girl imoko (girl) from the manga yani_neko, a MOSS port of the Anima / Illustrious version.

- Neon Siren Moss Turbo · 1.0 — Base Model / Checkpoint: Paid MOSS Turbo merged base model: cyberpunk plus high-gloss seductive style, iridescent skin, glowing makeup, electric blue-purple and orange neon, strong rim lighting and holographic flares, with clothing leaning towards latex, metal, and translucent materials.

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