MOSS Image Evening Report 2026-11-08 | Top Pick: Fire Palms: Umbrium

This issue reviews 35 new MOSS versions, with 2 rated S and 8 rated A. The top pick is the Palm Flame LoRA: the flame itself acts as a light source, illuminating the palm and face, with guidance provided on hand gestures, colors, and lighting techniques. Another S-rated item is a high-quality landscape LoRA, complete with sampling parameters and prompt templates. The A-tier includes a PVC figure texture base model, two cinematic styles (V67 Fairy Tale World and Cosplay Photography), Edo-period princess attire, and more; the MisterMR series fictional characters and several paid art style LoRA examples are almost entirely hidden by rating filters, resulting in a C rating.

2026-11-08T13:17:32Z

Zhenzhen's AI Radar · MOSS Image · 2026-11-08 Evening Report

This issue features 35 new versions: Strongly Recommended 2 · Recommended 8 · Worth Watching 9 · Included Only 16

Editor's Note This issue reviews 35 new MOSS versions, with 2 rated S and 8 rated A. The top pick is the Palm Flame LoRA: the flame itself acts as a light source, illuminating the palm and face, with guidance provided on hand gestures, colors, and lighting techniques. Another S-rated item is a high-quality landscape LoRA, complete with sampling parameters and prompt templates. The A-tier includes a PVC figure texture base model, two cinematic styles (V67 Fairy Tale World and Cosplay Photography), Edo-period princess attire, and more; the MisterMR series fictional characters and several paid art style LoRA examples are almost entirely hidden by rating filters, resulting in a C rating.

Strongly Recommended (2)

[S] Fire Palms: Umbrium · Moss v1.0

LoRA · Ported · Author umbrium2835 · Published 11-08 19:23 · Downloads 20

Palm Flame LoRA: Places a living flame above the character's open palm. The flame itself acts as a light source, illuminating the palm, face, and clothing while reflecting onto the surrounding environment.

Images

Example 1248x1824

Reasons for Recommendation

Ported: The same page already features Krea 2, Anima, and Illustrious versions.
Actionable: Trigger word palmflame, strength 1.0; the author clearly specifies hand gesture prompts (e.g., outstretched hand, palm up, one flame per hand), flame colors (default orange, but blue / white / green / purple fire can be specified), and optimal dark lighting scenarios.
Suitable for one or two characters, various poses (standing, sitting, kneeling, lying down), and different camera angles; 4 visible examples provided.

How to Use

Include palmflame in the prompt, then specify hand gestures (e.g., palm up) and flame color; works best in dark lighting scenarios; strength 1.0.

Risks and Limitations

The training words displayed on the page show palmstyle, but the trigger word specified in the main text is palmflame; follow the main text.
This entry has not been tested.

Trigger Word: palmstyle

[S] High Landscape | Moss Image · MOSS High Landscape v1.0

LoRA · New Model · Author svnx7x_ · Published 11-08 10:09 · Downloads 47

High-Quality Landscape LoRA: Primarily focused on landscapes, but can also add detail to any scene.

Images

Example 800x1600

Reasons for Recommendation

New model; serves landscape and environmental concept art.
Complete parameters: Euler sampler, 40–50 steps, CFG 3–5, strength 0.8–1, with positive and negative prompt templates provided; also usable with MOSS-Image SFT.
5 visible examples provided.

How to Use

Start the prompt with highlandscape, best quality, highres,; use negative prompts like worst quality, bad anatomy, watermark, jpeg artifact; Euler sampler with 40–50 steps, CFG 3–5, strength 0.8–1.

Risks and Limitations

No comparison images provided without the LoRA.
This entry has not been tested.

Trigger Word: highlandscape

Recommended (8)

[A] PVC Figurerizer · v1.0 MOSS-Image

Base Model / Checkpoint · Ported · Author aueki4g467 · Published 11-08 13:27 · Downloads 65

MOSS version of the PVC Figurerizer base model: Deliberately aims for a "real PVC figure" texture (even if it looks slightly cheap), rather than generic semi-realism.

Downgraded from automatic S to A: The concept is clear and parameters are complete, but the author admits stability is insufficient.

Images

Example 1024x1536

Reasons for Recommendation

Ported: The original page had an SD 1.5 version; the model page has accumulated 4999 downloads.
Sampling baseline provided: 12–15 steps, CFG approx. 2; offers both INT8 and BF16 file formats.
7 visible examples provided.

How to Use

Load PVCFigurerizer_MOSS (use INT8 for lower VRAM), with 12–15 steps and CFG approx. 2.

Risks and Limitations

The author states that anatomical stability is low and prompt adherence is "not yet sufficient".
3 examples are hidden by rating filters; this entry has not been tested.

[A] V67 World of Fairy Tales · V1 MOSS

LoRA · Ported · Author angelomaiota · Published 11-08 16:32 · Downloads 27

MOSS version of the V67 "Fairy Tale World" cinematic style LoRA: Features castles, magical forests, and court ball scenes reminiscent of Grimm's Fairy Tales and Aesop's Fables.

Downgraded from automatic S to A: The direction is clear, but strength and sampling parameters are missing.

Images

Example 1024x1024

Reasons for Recommendation

Ported: The same page already features Krea 2 and ZImageTurbo versions.
The author provides a long list of primary trigger words (including weight syntax); 4 visible examples provided.

How to Use

Start the prompt with (World of Fairy Tales:1.1), fablecore, fairy tale world, etc., using the author's provided trigger strings; strength not specified by the author.

Risks and Limitations

The trigger string is very long, indicating that much of it is promotional material for an event.
This entry has not been tested.

Trigger Word: (World of Fairy Tales:1.1), fablecore, fairy tale world, Brothers Grimm fairy tale, Grimmian storybook, Aesop fable, Aesopian narrative, Collodi storybook, European fairy-tale tradition, classic fable, storybook fantasy, enchanted kingdom, royal court, fairy-tale castle, magical forest, mushroom ring, misty lake, grand palace ballroom, candlelit castle hall, ornate medieval fantasy costumes, elegant royal attire, well-dressed adult fairy-tale characters, regal queen, adult king, adult princess, adult prince, sorceress, wizard, knight, fairy godmother, huntsman, bard, enchantress, duchess, royal advisor, gentle giant, mermaid noble, elf noble, dwarf artisan, witch, fairy queen, cinematic fantasy lighting, detailed storybook illustration, romantic fairy-tale atmosphere

[A] Kurumi style Korean Semi-Realistic CG Thick Paint · v1.0

LoRA · New Model · Author yor_briar · Published 11-08 11:13 · Downloads 96

A Korean semi-realistic CG thick paint style LoRA trained on the author's own artwork; effective even without a trigger word.

Downgraded from automatic S to A: It is the artist's personal style with clear licensing, but parameters are sparse.

Images

Example 1248x1680

Reasons for Recommendation

New model; the author notes that many training images feature Boa Hancock, so this character will look more accurate.
Examples use the kromoss_8_step_dmd2_v2 acceleration LoRA; approximately 96 downloads in half a day since release (indicating attention level only).

How to Use

You can write ohstyle, or omit it; use the 8 step acceleration LoRA for generation; strength not specified by the author.

Risks and Limitations

The author claims to be a training novice; 8 examples are hidden by rating filters.
This entry has not been tested.

Trigger Word: ohstyle,

[A] V67 Cosplay photography · V1 MOSS

LoRA · Ported · Author angelomaiota · Published 11-08 15:04 · Downloads 37

MOSS version of the V67 Japanese Cosplay photography style LoRA: Features cosplay portraits in Akihabara, Harajuku, Comiket venues, and studios.

Images

Example 1024x1024

Reasons for Recommendation

Ported: The same page already features Flux.2 Klein, Krea 2, and ZImageTurbo versions; the model page has accumulated 835 downloads.
Provides trigger strings with weight syntax; 4 visible examples provided.

How to Use

Start the prompt with (Cosplay photography:1.1), japan, cosplay, photo, studio, etc., using the author's provided words; strength not specified by the author.

Risks and Limitations

Strength and sampling parameters are not provided.
This entry has not been tested.

Trigger Word: (Cosplay photography:1.1), japan, cosplay, photo, studio, akiba, tokyo, harajuku, shibuya, comiket, nippon, costume, portrait, fashion, street, convention, fantasy, sci-fi, cyber, modern, retro

[A] FecchiBabes · MOSS - v1.0

LoRA · Ported · Author ChocoGirlTummy · Published 11-08 19:00 · Downloads 16

MOSS version of a LoRA based on the art style of Italian Disney comic artist Massimo Fecchi (used by the author to draw beautiful female characters).

Images

Example 1280x720

Reasons for Recommendation

Ported: The same page already features Krea 2, Illustrious, and ZImageTurbo versions; the author claims that using the same dataset yielded surprisingly good results.
6 visible examples provided; trigger word is clear.

How to Use

Include fecchistyle in the prompt; strength not specified by the author.

Risks and Limitations

Mimics a comic artist's style; 10 examples are hidden by rating filters.
This entry has not been tested.

Trigger Word: fecchistyle

[A] Japanese Princess / Edo Period Princess · Moss-Image

LoRA · Ported · Author Takec5031 · Published 11-08 19:36 · Downloads 26

MOSS version of a LoRA recreating Japanese Edo-period princess attire.

Images

Example 896x1184

Reasons for Recommendation

Ported: The same page already features ZImageTurbo and Illustrious versions.
13 visible examples provided; trigger word is clear.

How to Use

Include Princess of Edo period, edohime, in the prompt; strength not specified by the author.

Risks and Limitations

The description is only one sentence; parameters are not provided.
This entry has not been tested.

Trigger Word: Princess of Edo period, edohime,

[A] RDBT | MOSS · v0.2

LoRA · Version Update · Author reakaakasky · Published 11-08 17:18 · Downloads 53

RDBT v0.2: A digital art / painting LoRA trained on the author's personal comprehensive dataset, focusing on lighting and texture, covering everything from wallpaper-quality digital paintings to 5-minute sketches.

Images

Example 896x1152

Reasons for Recommendation

Version update: The author states this version was trained on MOSS Base (not SFT), and the cover image was also generated using Base.
The model page has accumulated 646 downloads; 6 visible examples provided.

How to Use

Load using MOSS Base; trigger word and strength not specified by the author.

Risks and Limitations

The author does not specify the differences from v0.1.
This entry has not been tested.

[A] Kajiura style [Pony/Illustrious/anima/MOSS] · MOSS

LoCon · Ported · Author rotechacon667 · Published 11-08 10:13 · Downloads 29

MOSS version of the Kajiura style LoCon; the same page also features Pony, Illustrious, and Anima versions.

Images

Example 1024x1536

Reasons for Recommendation

Ported: The model page has accumulated 1089 downloads; 5 visible examples provided.
Trigger word is clear.

How to Use

Prompt specifies Kajiura; strength not provided by author

Risks and Limitations

The author notes that the number of training images was reduced; the description mainly covers the production process
This entry was not run

Trigger words: Kajiura

Worth Noting (9)

[B] MOSS Style · v2 · 300 Steps Fine-tuned

LoRA · Version Update · Author Mistbell · Published 11-08 17:38 · Downloads 30

“Mixed Anime Style” LoRA v2: Continued training from the 200 step checkpoint of the same training run up to 300 steps.

Images

Example 1152x832

[B] A style lora by Ryia | 3 · v1.0

LoRA · New Model · Author Ryia1417 · Published 11-08 17:12 · Downloads 30

A style LoRA trained on images generated by the author using NovelAI v5 (Ryia’s 3th model).

Images

Example 1920x1072

[B] Cutesexyrobutts Style | MOSS Image · v1.0

LoRA · New Model · Author HissingCat · Published 11-08 10:24 · Downloads 76

A quick test version v1 of the MOSS LoRA for artist cutesexyrobutts’ style; the author says v2 will be uploaded later.

Images

Example 896x1184

[B] MOSS Style · v1 · 200 Steps

LoRA · New Model · Author Mistbell · Published 11-08 17:37 · Downloads 13

The v1 (official 200 step version) of the “Mixed Anime Style” LoRA, specifically for MOSS-Image-SFT.

Images

Example 1088x768

[B] Mayumi Saegusa · v1.0

LoRA · New Model · Author Hayakake · Published 11-08 18:24 · Downloads 22

A MOSS character LoRA for Mayumi Saegusa from The Irregular at Magic High School, providing trigger phrases for three outfits: school uniform, white dress, and tennis attire.

Images

Example 1024x1536

[B] NexBlend 04 Asian Realistic LoRA - MOSS Image · v1.0

LoRA · New Model · Author NexBlend · Published 11-08 20:40 · Downloads 9

A realistic portrait LoRA featuring Korean-style East Asian faces (NexBlend 04); the author claims it maximizes the realistic potential of the MOSS base model.

Images

Example 1024x1536

[B] Maid Classic / maid classic · v1.0 moss

LoRA · Ported · Author bolero537 · Published 11-08 10:39 · Downloads 98

The MOSS version of the classic maid outfit LoRA; other base model versions are available on the same page.

Images

Example 768x1280

[B] Koto and Kumi (Posthumous Work) MOSS-image LORA · 002

LoRA · New Model · Author tabinezumi · Published 11-08 13:01 · Downloads 37

A dual-character LoRA for two characters from the same work (Asakawa Koto and Takashima Kumi); the author provided complete Japanese descriptions for their hairstyles and clothing.

Images

Example 1024x1024

[B] AFK Journey Anima Style · v2.0 - STF

LoRA · Ported · Author pedrodenovox549 · Published 11-08 19:17 · Downloads 19

The MOSS SFT version v2.0 of the art style LoRA for the mobile game AFK Journey, leaning towards 3D cartoon rendering.

Images

Example 1088x1440

Included Only (16)

Edith Belwyn · MossImage — LoRA: The MOSS version of the character LoRA for Edith Belwyn (a fictional figure from a paper art family in Bruges), part of the MisterMR series.
Style na_tarapisu153 style 【anima】 · moss — LoRA: The MOSS version of the artist na_tarapisu153’s style LoRA, no trigger words required.
Tentacle Clothes / tentacles clothes(pony) · v1.0 moss — LoCon: The MOSS version of the “Tentacle Clothes” concept LoCon, recommended strength 1.0–3.0.
Veronika_Krea2 · Veronika_AKT_MOSS — LoRA: The MOSS version of the LoRA for Veronika, a fully AI-generated fictional female character (black hair, freckles, blue eyes).
Chulianzi (Cute, Breakable) / kawaii wa kowaseru · v1.0 — LoRA: The MOSS character LoRA for the Chinese internet character “Chulianzi”: black dress, pink socks, with injury makeup such as red marks and bruises on the forehead.
Sora · MOSS Image — LoRA: The MOSS version of the custom LoRA for Sora, a slim female character aged 21.
comiclo-style/c-lo style · v1.0 — LoRA: The “c-lo” style LoRA.
boleromix(moss-image) · v1.03 — Base Model / Checkpoint: The MOSS fusion base model boleromix v1.03, which the author describes as being fused “quite casually.”
Iman Belwyn · MossImage — LoRA: The MOSS version of the character LoRA for Iman Belwyn, part of the MisterMR Bruges Belwyn family series.
Style MeioN style 【anima】 · moss — LoRA: The MOSS version of the artist MeioN’s style LoRA, no trigger words required; the author says the eyes are mainly drawn similarly.
Cutesexyrobutts Style | MOSS Image · v2.0 — LoRA: The paid cutesexyrobutts style LoRA v2: the author made aesthetic adjustments, claiming better style fusion, hands, and backgrounds.
Leyla Phoenix · MossImage — LoRA: The MOSS version of the character LoRA for Leyla Phoenix, part of the MisterMR series.
Nevaeh Belwyn · MossImage — LoRA: The MOSS version of the character LoRA for Nevaeh Belwyn, part of the MisterMR series.
@citrusted · v1.0 — LoRA: The MOSS LoRA for pixiv artist @citrusTED’s style.
Herb Ritts Style · v1 — LoRA: The paid Herb Ritts style LoRA: sculptural bodies, strong directional light, and deep shadows from 90s black-and-white male fashion / physique photography.
aboutorange Moss style · v1.0 — LoRA: The paid aboutorange style LoRA.

Data: Civitai Public API (Base Model MOSS Image), inclusion window 11-08 09:33 → 11-08 21:01 (Beijing Time). Recommendations reviewed by editors; download counts reflect attention only.