Zhenzhen's AI Radar · MOSS Image · 2026-11-14 Evening Report
This issue features 40 new versions: Strongly Recommended 1 · Recommended 6 · Worth Watching 17 · Listed Only 16
Editor's Note This issue reviewed 41 new MOSS versions, with 1 rated S, 6 rated A, 17 rated B, and 17 rated C; no items were excluded. Top pick: MOSS_3DReal_Portrait, a 3D realistic portrait LoRA that enhances skin translucency, facial dimensionality, and pore texture. The author provides intensity tiers and a 8-step recipe, along with same-seed comparison images. Fiat concept cars, top-down tactical map assets, Slavic primeval forest environments, double-exposure landscapes, the Arknights: Endfield Operator Collection v5.0, and Romiks Western comic style are rated A. Automatic S-rated items such as Glutinous Rice Ink Wash, Zuole, and MementoMori v1.1 were downgraded due to insufficient descriptions, few examples, or the author's admission that they are inferior to previous versions. Standard character LoRAs and style LoRAs providing only weights are mostly rated C.
Strongly Recommended (1)
[S] MOSS_3DReal_Portrait · v1.0
LoRA · New Model · Author: baowy · Published 11-14 14:58 · Downloads: 73
MOSS_3DReal_Portrait: A 3D realistic portrait enhancement LoRA that strengthens subsurface scattering, skin texture, and facial volume, making portraits look like 3D renders while keeping facial proportions unchanged.
Images

Reasons for Recommendation
- New capability: The author claims four effects appear simultaneously—skin translucency, dimensionality of cheekbones, jawline, and nose bridge, pore-level texture, and more concentrated highlight falloff; it does not alter facial shape, drift towards anime style, or feminize male subjects.
- Serves realistic portraits, half-body, and full-body shots; the author claims via same-seed comparisons that it has almost no effect on scenes and interiors, so it will not pollute backgrounds.
- Complete parameters, no trigger words: Provides an intensity tier table and a 8-step recipe; the 3 visible examples are comparison images using the same seed with 0 / 1.0 / 1.5 / 1.8.
How to Use
- Pair with MOSS_8-steps-dmd2-lora (intensity 1.0), set this LoRA's intensity to 1.0–1.5 (author recommends 1.5), use 8 steps, CFG 1.0, euler / simple sampler, and resolution 1024×1360 (multiples of 16 are acceptable); no trigger word required.
- Intensity 0.8–1.0 adds only slight depth and texture; intensity 1.8 provides the strongest volume and may slightly alter facial features.
Risks and Limitations
- Maximum intensity is 2.0, at which point skin begins to look waxy; negative values cannot be used, as negative intensity degrades image quality.
- Effective only on people; effects are weak on objects and macro shots, and largely ineffective on landscapes; there are only 3 visible examples.
Recommended (6)
[A] Romiks – Rafael Albuquerque × Kittyhand | Western Comic Style · v1.0
LoRA · New Model · Author: Shashami · Published 11-14 18:12 · Downloads: 25
Romiks Western Comic Style LoRA: Originally intended to mix the styles of artists Rafael Albuquerque and Kittyhand, it resulted in a unique style—thick outlines, dark blocky shadows, high contrast, traditional comic texture combined with cleaner digital coloring.
Downgraded from Automatic S to A: The author describes this as an experimental first work, and trigger word usage is inconsistent.
Images

Reasons for Recommendation
- New model; trained on MOSS-Image Base using AI Toolkit. The author claims it is suitable for fantasy, sci-fi, modern, action, and other genres, leaning towards dark tones and strong shadows, but can also produce bright scenes.
- The author provides four types of example prompts and three optional lighting descriptions (dark purple/magenta, blue cyber, warm orange/amber); using strong directional light better reveals shadows and ink lines.
- 5 visible examples (15 hidden); free to use.
How to Use
- Start prompts with the trigger word (the model card lists romiksMs, while the site registration uses romiksMsA), intensity 1.0; on Base, use 30–50 steps, or pair with the 8-step Turbo LoRA using its quick settings.
- For natural fair skin, add color descriptions like subtle pink undertones, soft rosy cheeks; in negative prompts, include speech bubbles, dialogue balloons, text.
Risks and Limitations
- The author states this is their first MOSS training and may be over-trained; fine hatching in facial shadows is unstable, and an improved version will be released later.
- Writing 'pale skin' often results in grayish-white, almost vampire-like complexions; speech bubbles and comic text sometimes appear unexpectedly.
- Trigger words are inconsistent between the model card and site registration; CFG was not provided by the author; the license does not allow derivatives; the page contains mixed content.
Trigger Word: romiksMsA
[A] Fiat Multiplina Concept [Moss] · v1
LoRA · New Model · Author: denrakeiw · Published 11-14 12:44 · Downloads: 18
Fiat Multiplina Concept Car LoRA: A flat-fronted micro van with a yellow and white two-tone body, a gray canvas top rolled up, and retro disc wheels, including details like front and rear lights, logo text, and striped interior seats.
Downgraded from Automatic S to A: Examples and prompts are complete, but step count and CFG are missing, and usage is limited to this specific car model.
Images

Reasons for Recommendation
- New model; serves automotive product images, concept car scenes, and retro micro-car settings. The author claims training images cover front three-quarters, side profile, rear three-quarters, and close-ups.
- The author provides 5 complete example prompts (studio shots, side views, rear views, front close-ups, interior views through windows); the text FIAT and MULTIPLINA on the body are written in English double quotes.
- 10 visible examples (7 hidden); the license allows commercial use of generated images and sales.
How to Use
- Write 'Fiat Multiplina Concept' in the prompt, starting with 'Create a photograph of ...' as per the author's examples, specifying camera angle, lighting background, two-tone paint, and required text; intensity 0.85–1.0.
- Step count, CFG, and sampler were not provided by the author.
Risks and Limitations
- The model card title says Flux LoRA, and annotations follow Flux's long-sentence style; there is no separate explanation for the MOSS version.
- Trained only on this one car; most examples are studio shots, and performance in complex scenes like street views is not specified.
Trigger Word: Fiat Multiplina Concept
[A] Dizzle's Surreal Landscape Photography · v1
LoRA · New Model · Author: Maridizzle · Published 11-14 14:30 · Downloads: 29
Dizzle's Surreal Landscape Photography: Double-exposure style surreal composites—blending subjects into landscapes like misty mountains and sea surfaces, with soft, grayish tones.
Downgraded from Automatic S to A: Same tier as other single-effect LoRAs by the same author that have complete parameters.
Images

Reasons for Recommendation
- New model; serves poetic posters and album cover-style double-exposure images. The author suggests naming the two elements to be blended, connected by 'blended with', e.g., 'a woman blended with a foggy mountain range'.
- Complete parameters: Intensity 0.9–1.0, 30–50 steps, CFG 3, Euler flowmatch; use full sentences for descriptions.
- 6 visible examples (3 hidden).
How to Use
- Start prompts with 'dz_land' (the description says 'dz_land, surreal landscape photo', while registration uses 'dz_land, faded surreal landscape'), then use the structure 'A blended with B' to describe the subject and landscape; intensity 0.9–1.0, 30–50 steps, CFG 3, Euler flowmatch.
Risks and Limitations
- The author states it is best suited for scenes with mist, low saturation, and soft transitions; hard edges and high-contrast images are not specified.
- The page contains mixed content.
Trigger Word: dz_land, faded surreal landscape,
[A] Arknights: Endfield | Operators Collection · v5.0 [moss]
LoRA · Ported · Author: motimalu · Published 11-14 14:52 · Downloads: 69
Arknights: Endfield Operator Collection LoRA v5.0 MOSS version: A single file covering game operators from versions 1.0 / 1.1, unreleased operators, and antagonist characters.
Character collections are generally rated C; this one is kept at A due to the large number of covered characters and complete tag formatting.
Images

Reasons for Recommendation
- Ported: Previous versions were trained on Anima; this version is trained on MOSS Image using a diffusion-pipe branch and Kohaku Blueleaf's implementation of MOSS timestep sampling, with tags mixed with natural language annotations.
- Serves Arknights: Endfield fan art; the model card lists tag formats for each character. The author claims it works best to name the character first, then describe basic appearance in natural language.
- 13 visible examples (7 hidden); cumulative page downloads are 4136, with 392 likes, mostly from the Anima version.
How to Use
- Write character tags in prompts (e.g., perlica (arknights), yvonne (arknights)), then describe appearance like hairstyle and clothing in full sentences; intensity, step count, and CFG were not provided by the author.
Risks and Limitations
- Some examples have the author's Masterpiece aesthetic LoRA overlaid; using this LoRA alone will result in a different style.
- The author restricts usage to non-commercial fan creations; designs of unreleased operators are unconfirmed; the page contains mixed content.
[A] Top Down BattleMap Assets for Moss · v1.0
LoRA · New Model · Author: Razira · Published 11-14 09:58 · Downloads: 36
Top-Down BattleMap Assets LoRA: Forces objects into a strict top-down perspective, with an illustration style common in tabletop RPG (TTRPG) tactical maps, used for creating map stickers that can be cut out.
Images

Reasons for Recommendation
- New model; serves battle map assets for TTRPGs like D&D (statues, furniture, props, etc.). The author claims most objects can be stably rendered in top-down view.
- Clear usage: Start with 'A top-down view of a [subject]', use 'A top-down view of a statue' for statues, and add 'The entire [subject] is in frame' to ensure full inclusion; specify background color for easy post-processing removal.
- 13 visible examples (7 hidden); the author claims the MOSS version makes it easier to fully frame objects compared to their Krea 2 version.
How to Use
- Start prompts with 'A top-down view of a [subject]', add 'The entire [subject] is in frame, Illustrated concept art style', and specify 'white background' or 'black background'.
- Intensity, step count, and CFG were not provided by the author.
Risks and Limitations
- The author notes that some objects and descriptions can disrupt the top-down perspective, requiring prompt corrections; they also admit that MOSS itself follows prompts well, so this LoRA is not essential, and image quality is slightly lower than Krea 2
- The license only permits on-site rental and does not allow derivatives; the page contains mixed content
Trigger words: A top-down view of a [subject], A top-down view of a statue, The entire [subject] is in frame,, Illustrated concept art style,
【A】Dremuch (Primeval Forest & Swamp Environment) · MOSS Image v.1.0
LoRA · Port · Author GreenRazor · Published 11-14 13:56 · Download 39
Dremuch: A LoRA for northern/Slavic primeval forest and swamp environments—moss-covered ancient spruce forests, fallen log piles, peat bogs, forest floor macro shots, and moonlit misty forests.
Images

Reasons for Recommendation
- New model: The author positions it as a companion environment expansion for their Tales of Old Rus (Slavic dark fantasy characters); it is labeled as a port because an SDXL version exists on the same page, but the SDXL version was uploaded slightly later on the same day
- The author provides five sets of ecological description prompts that can be used directly: ancient spruce moss forests, fallen logs (burelom), peat bogs, forest floor macro, and moonlit shrubs
- 4 visible examples, no hidden content; the license allows commercial use and sale of generated images
How to Use
- Write dremuch in the prompt, followed by one of the author's ecological description sets (e.g., misty peat bog, dark still peat water, sphagnum moss ...); the author does not specify strength, steps, or CFG
Risks and Limitations
- The trigger word dremuch is only mentioned in the description and is not registered; the model card lacks MOSS-specific parameters
- Intended for standalone use in pure landscapes; combining it with Tales of Old Rus characters requires a MOSS version of that model, which the author does not specify
Worth Watching (17)
【B】MementoMori style · MOSS_memento style_v1.1
LoRA · Version Update · Author yyumt · Published 11-14 10:06 · Download 61
MOSS v1.1 of the watercolor CG style LoRA for the mobile game Memento Mori: Retrained, the author claims it is closer to the original style than the previous version when paired with the 8 step acceleration LoRA.
Images

【B】Zuo Le (Arknights) · v1.0
LoRA · New Model · Author DSLM · Published 11-14 11:25 · Download 18
Character LoRA for Zuo Le from Arknights: Blue hair, blue eyes, pointed ears, snake tail, covering the base skins for Elite 0/Elite 2 and three skins (Youthful Journey, Silent Breakthrough, Musical Synesthesia 2025).
Images

【B】Nuomici Ink Wash Style · v1.0
LoRA · New Model · Author fjfcj · Published 11-14 15:31 · Download 37
"Nuomici Ink Wash" style LoRA: The name indicates an ink wash style, and it is one of three style LoRAs released by author fjfcj during the same period.
Images

【B】Painterly Concept Art Style for MOSS · v1.0
LoRA · New Model · Author Razira · Published 11-14 17:19 · Download 47
Painterly Concept Art Style: A semi-realistic digital concept art style with traditional oil painting brushstrokes.
Images

【B】Genshin Impact Miliastra Wonderland - Dreamkeeper Plush | Thousand Star Realm - Dreamkeeper Plush · v1.0 MOSS
LoRA · Port · Author LS3171244 · Published 11-14 19:55 · Download 8
MOSS version of the Dreamkeeper Plush outfit LoRA from Genshin Impact's "Thousand Star Realm": Striped off-shoulder puff-sleeve top, black pleated skirt, striped knee-high socks, black animal-ear beanie, and black boots, optionally with a plush toy.
Images

【B】Lemon Jingjing Jingjing / nmjjjj style for moss · v1.0
LoRA · New Model · Author azukiuy · Published 11-14 19:08 · Download 21
MOSS LoRA imitating the art style of illustrator "Lemon Jingjing Jingjing" (nmjjjj).
Images

【B】DeepSeek Whale Girl Dafeiyu · MOSS-Image LoRA · Proportion Version · SFT Base Model (adult)
LoRA · Version Update · Author WYR233 · Published 11-14 11:32 · Download 21
Proportion version SFT base model of DeepSeek Girl "Dafeiyu": An adult female proportion version of the same character, trigger word dafeiyu, consistent with the chibi art style.
Images

【B】Kamitani George - Dragon's Crown - Odin Sphere - Style MOSS LORA · v1.0
LoRA · New Model · Author Konan · Published 11-14 09:59 · Download 83
Art style LoRA for Kamitani George (Vanillaware's Dragon's Crown and Odin Sphere): Exaggerated heroic proportions, thick ink lines, oil painting texture, and strong chiaroscuro 2D dark fantasy illustrations.
Images

【B】MOSS All in One Inpaint, Outpaint, & Rotate Workflow | Partial Redraw Outpainting · v1.0 Beta
Workflow · New Model · Author lonecatone23 · Published 11-14 19:51 · Download 17
MOSS integrated partial redraw / outpainting / rotation workflow module (Beta): Can be inserted into existing ComfyUI workflows for partial redraws, outpainting, rotation, and object replacement.
Images

【B】DeepSeek Whale Girl Dafeiyu · MOSS-Image LoRA · Chibi Version · SFT Base Model (chibi)
LoRA · New Model · Author WYR233 · Published 11-14 11:32 · Download 18
Chibi version SFT base model of DeepSeek Girl "Dafeiyu" (Whale Girl): Big head and small body chibi proportions, dark blue gradient long curly hair, whale fin ears, whale tail, and navy blue maid outfit.
Images

【B】Vintage CRT Television · MOSS
LoRA · New Model · Author 858654128784 · Published 11-14 13:54 · Download 80
Vintage cathode ray tube (CRT) television object LoRA: Black plastic, beige, or wood-grain shell TVs on wooden tables or cabinets, featuring knobs, speaker grilles, scan lines, and curved glass reflections, with freely specifiable screen content.
Images

【B】Dizzle's Surreal Red Photography · v1
LoRA · New Model · Author Maridizzle · Published 11-14 11:53 · Download 27
Dizzle's Surreal Red Photography: Immerses the image in saturated crimson red, with a hard-edged geometric shape like a cube or prism floating in the background, resembling fashion editorials and installation art.
Images

【B】Silkshade · Moss-v2.0
LoRA · Version Update · Author soda_Cookie · Published 11-14 10:00 · Download 42
MOSS v2.0 of the comic illustration style LoRA Silkshade, which the author claims fixes the hand deformations found in v1.
Images

【B】Retro Quiet — Cream Walls and Deep Blue Skies, a Calm Vacation Look | MOSS · MOSS
LoRA · New Model · Author 858654128784 · Published 11-14 09:56 · Download 53
Retro Quiet: Quiet retro vacation scenes—low cream-colored houses, stone tables and rattan chairs, grassy slopes and coastlines, paired with rich deep blue skies and large areas of negative space, resembling vacation photos from old magazines.
Images

【B】ULTIMATE LIGHT REALISM — MOSS IMAGE · v1.0
LoRA · New Model · Author Vince_AI · Published 11-14 15:00 · Download 2 · Paid
ULTIMATE LIGHT REALISM: A LoRA that enhances MOSS lighting and photographic realism, leaning towards backlighting, volumetric light, reflection/refraction, and soft natural shadows.
Images

【B】Juemingzi Style · moss v1.0
LoRA · Port · Author fjfcj · Published 11-14 13:42 · Download 28
MOSS version of illustrator Juemingzi's art style LoRA: Illustrations leaning towards anime character portraits and postcard styles, which the author claims are particularly good for Hatsune Miku.
Images

【B】MossImage Layout Editor · v0.2
Other · New Model · Author hybskgks28275 · Published 11-14 14:40 · Download 24
MossImage Layout Editor v0.2: A standalone Windows mini-program that uses mouse-drawn bounding boxes for layout, exporting JSON prompts for MOSS-Image without needing to open ComfyUI.
Images

Included Only (16)
- viclim Moliu Style · v1.0 — LoRA: A MOSS LoRA that mimics the art style of illustrator viclim (Moliu).

- Cute Style · v1.0 — LoRA: A "Cute" style LoRA (filename "Cute Sleeping"), one of three style LoRAs released by author fjfcj during the same period.

- DeepSeek Whale Girl Big Fat Fish · MOSS-Image LoRA · Proportionate Version · Base Model (adult) — LoRA: The Base model version of the proportionate "Big Fat Fish" DeepSeek Girl: Uses the same trigger words and parameters as the proportionate SFT version, intended only for those generating images directly with the MOSS-Image Base.

- Faith NordhavenLove · MossImage — LoRA: A MOSS version LoRA for the original fictional female character Faith NordhavenLove (part of author NordhavenLove's family series), set as a lighthouse keeper's assistant on a Swedish cape, often wearing a navy blue wool coat and a knit cap.

- DeepSeek Whale Girl Big Fat Fish · MOSS-Image LoRA · Chibi Version · Base Model (chibi) — LoRA: The Base model version of the chibi "Big Fat Fish" DeepSeek Girl: Uses the same trigger words and parameters as the chibi SFT version, intended only for those generating images directly with the MOSS-Image Base (without SFT).

- uekura eku Uekura Style · moss v1.0 — LoRA: The MOSS version of the Uekura Eku art style LoRA: Flat 2D character illustrations, with the author recommending a blank background.

- Comfyui simple MOSS workflow with guide · v1.0fix — Workflow: PotatCat's simple MOSS text-to-image ComfyUI workflow v1.0fix: Fixes connection errors, where previously NegPip only affected the negative prompt.

- ogino jun style · moss base v1.0 — Base Model / Checkpoint: The MOSS Base version of the Ogino Jun (ogino jun) art style LoRA, with training images mostly being manga pages.

- [Moss] Ass Slider · v1.0 — LoRA: A slider LoRA for adjusting the size of a character's hips, which can be positive or negative.

- Xia Die Castorice Honkai: Star Rail MOSS Image · v1.0 — LoRA: A MOSS character LoRA for Xia Die (Castorice) from Honkai: Star Rail.

- nai5 style-High Contrast Lines · v1.0 — LoRA: An art style LoRA named "nai5 style High Contrast Lines", with the trigger word nai5, providing files for two training steps.

- Florence NordhavenLove · MossImage — LoRA: A character LoRA for the fictional person Florence NordhavenLove (part of author Nordhaven's family series), set as a watercolor painter by the Swedish coast.

- Android 18 MOSS · v1.0 — LoRA: The first version of the MOSS character LoRA for Android 18 from Dragon Ball.

- Eldreda Sunbringer エルダ・サンブリンガー, 2 Outfits — Zenless Zone Zero ゼンレスゾーンゼロ [Anima+Illustrious+MOSS] · v1.0 MOSS Image — LoRA: A MOSS version character LoRA for Eldreda Sunbringer (Aierdelida) from "Zenless Zone Zero": Long gray hair in a half-up braid, small black horns, heterochromia (blue and yellow eyes), with one file containing both the default outfit and the Moonlit outfit.

- zhangfetumiao · v1.0 — LoRA: A character LoRA for a skin of Zhang Fei from Honor of Kings (written as Tu Miao by the author).

- Yanyun Sixteen Voices - Yi Dao / Where Winds Meet - Yi Dao · v1.0 — LoRA: A MOSS character LoRA for Yi Dao from "Yanyun Sixteen Voices": A sturdy martial arts man wearing a blue-green robe with scale-patterned shoulder armor, a high bun, a purple-brown leather headband, thick black beard, and a light scar on his cheek.

Data: Civitai Public API (Base Model MOSS Image), collection window 11-14 08:48 → 11-14 21:01 (Beijing Time). Recommendation reasons have been reviewed by editors; download counts only represent attention levels.