MOSS Image Evening Report 2026-11-13 | Top Pick: [MOSS-5B] Modern TV Anime Screencap / Sakuga Broadcast (in sakuga_screencap style)

This issue reviews 43 new MOSS versions, with 1 rated S, 5 rated A, 14 rated B, and 22 rated C; 1 duplicate uploads were excluded. The top pick is the Sakuga Broadcast style LoRA: it transforms MOSS's default semi-thick paint style into crisp line art and high-contrast cel shading typical of TV anime screencaps. Trigger phrases, strength settings, 36 steps / CFG 3.0, and ComfyUI workflows are all provided. Clairveil (Clear Color Pale Wash) style, the unidirectional Asian idol face beauty slider, BYD Shark vehicle model, Y2K Neon, and billiard hall scenes with surveillance camera perspectives are rated A; soda_Cookie's 8 style LoRAs released consecutively using the same template only include trigger words and example sentences, so they are rated B or C based on the number of visible examples. Character/experimental entries automatically rated S, such as Sitri and Ada Wong, were downgraded.

Zhenzhen's AI Radar · MOSS Image · 2026-11-13 Evening Report

This issue features 42 new versions: Strongly Recommended 1 · Recommended 5 · Worth Watching 14 · Included Only 22

Editor's Note This issue reviews 43 new MOSS versions, with 1 rated S, 5 rated A, 14 rated B, and 22 rated C; 1 duplicate uploads were excluded. The top pick is the Sakuga Broadcast style LoRA: it transforms MOSS's default semi-thick paint style into crisp line art and high-contrast cel shading typical of TV anime screencaps. Trigger phrases, strength settings, 36 steps / CFG 3.0, and ComfyUI workflows are all provided. Clairveil (Clear Color Pale Wash) style, the unidirectional Asian idol face beauty slider, BYD Shark vehicle model, Y2K Neon, and billiard hall scenes with surveillance camera perspectives are rated A; soda_Cookie's 8 style LoRAs released consecutively using the same template only include trigger words and example sentences, so they are rated B or C based on the number of visible examples. Character/experimental entries automatically rated S, such as Sitri and Ada Wong, were downgraded. The Sakuga author uploaded weights and workflows with identical SHA256 hashes to another model page, which was excluded as a duplicate upload.

Strongly Recommended (1)

[S] [MOSS-5B] Modern TV Anime Screencap / Sakuga Broadcast (in sakuga_screencap style) · v1.0 - MOSS-Image 5B (Sak

LoRA · New Model · Author MaisonLatentia · Published 11-13 10:05 · Downloads 43

Modern TV Anime Screencap (Sakuga Broadcast) style LoRA: Transforms MOSS's default semi-thick paint output into crisp line art, high-contrast cel shading, glowing effect lines, and clean background art typical of TV anime keyframes.

Retains automatic S rating: The only new style LoRA this issue with complete parameters, workflows, training information, examples, and licensing. You can generate images directly by following the provided workflow.

Images

Example 896x1152

Reasons for Recommendation

  • New model; serves text-to-image needs for anime-style character art, action scenes, and cyberpunk street scenes that look like "a frame from a TV anime screencap." The author provides two complete copy-paste prompts: one for fantasy action (1344x768) and one for cyberpunk street characters (896x1152).
  • Complete parameters: Trigger phrase, strength settings for standalone use and when stacking with character LoRAs, Euler / Simple / 36 steps / CFG 3.0 / Shift 5.0 / SLG Layer 22, accompanied by a ComfyUI workflow JSON. Example images also embed generation metadata.
  • Training information disclosed: MOSS-Image Base 5B (bf16), 60 multi-resolution images, 2040 steps / 34 epochs, Rank 16 / Alpha 16; 12 visible example images. Free, with a license allowing commercial use and derivatives.

How to Use

  • Start the prompt with "in sakuga_screencap style," followed by a complete natural language scene description; standalone strength is 0.80 (0.65–1.00).
  • When stacking with character LoRAs, set style strength to 0.65–0.70 and character strength to 0.85–0.95; use Euler + Simple, 36 steps, CFG 3.0, Shift 5.0, SLG Layer 22. Simply drag the included sakuga_screencap_moss5b_workflow.json into ComfyUI.
  • Use the author's provided negative prompt string: photorealistic, live action, 3d render, cgi, realistic skin pores … letterbox, black bars, sakuga_screencap text.

Risks and Limitations

  • The author notes that close-ups in dense scenes may occasionally result in awkward hands or limbs at strength 0.80. You may need to increase style strength to 1.00, or lower it to 0.65 and combine with a character LoRA.
  • Shift and SLG settings depend on nodes within the included workflow; 6 examples are hidden; identical weights were uploaded to another model page by the author (3048636), which has been excluded as a duplicate.

Trigger Words: in sakuga_screencap style

Recommended (5)

[A] Byd Shark [Moss] · Moss

LoRA · New Model · Author denrakeiw · Published 11-13 16:47 · Downloads 17

BYD Shark (2027 model) vehicle LoRA for the double-cab open-bed plug-in hybrid pickup: Locks down body contours, light clusters, grille, wheels, cockpit, and bed details to produce realistic photos in a car review or advertising style.

Narrow application scope, but the usage guide and perspective checklist are among the most detailed this issue. Useful as a reference for writing vehicle or product LoRAs.

Images

Example 2736x1536

Reasons for Recommendation

  • New model; serves automotive product images, car review illustrations, and off-road/wading scene photos. The author states the training set consists of 72 official news photos and studio shots, including only the pure double-cab open-bed body style to ensure consistent contours across all angles.
  • Coverage is clearly defined: Front three-quarter, rear three-quarter, side, front, and rear views; outdoor / off-road / wading / studio environments; plus center console, seats, cargo bed, external power outlet, and engine bay.
  • Provides multiple complete example prompts (rural front three-quarter, blue-gray studio, wading, cockpit); 10 visible example images. Free, with a license allowing commercial use and derivatives.

How to Use

  • Start the prompt with "Create a photograph of the BYD Shark," followed by view angle, paint color (metallic orange or metallic slate-blue), lighting, and environment; write body text in quotes, such as BYD, DMO AWD, SHARK.
  • Strength 0.8 – 1.0; steps and CFG were not provided by the author.

Risks and Limitations

  • The model card title says Flux LoRA, and annotations are written for Flux; it is unclear if the MOSS version uses the same annotation set. No trigger words are registered on the page.
  • Serves only this specific vehicle model; 10 examples are hidden, the page contains mixed content, and involves automotive brand trademarks.

[A] Clairveil · 澄彩 | MOSS-Image Style LoRA · v1.0

LoRA · New Model · Author MengY_Madrigal · Published 11-13 17:49 · Downloads 28

Clairveil · Clear Color style LoRA: Fine lines, transparent and light color palettes, and soft character rendering, suitable for character illustrations and scenic compositions with figures.

Images

Example 1681x1300

Reasons for Recommendation

  • New model; the author states it was trained on MOSS-Image-heal using AI Toolkit and tested on original MOSS-Image-SFT and Base INT8 ConvRot. The released checkpoint is 9750 steps, rank 32.
  • Works with both natural language prompts and MOSS layout JSON; the author has placed full comparisons, prompts, and ComfyUI workflows on Hugging Face, along with a RunningHub comparison workflow.
  • 5 visible example images (generated from SFT at strength 0.75); free, with a license allowing commercial use and derivatives.

How to Use

  • Add the trigger word "clairveil" to the positive prompt, then describe the character, action, and scene. Note that the weight filename is spelled "claireveil.safetensors," but the trigger word does not have an 'e' in the middle.
  • Strength 0.75; try 1.0 for a stronger effect. Steps, CFG, and sampler were not provided by the author on the page; refer to the workflow on Hugging Face.

Risks and Limitations

  • The Civitai page itself does not list steps and CFG; key parameters are off-site.
  • 2 examples are hidden; released only 3 hours ago, with very little download feedback.

Trigger Words: clairveil

[A] MOSS_Pretty_AsiaFace_Slider · v1.0

LoRA · New Model · Author bwy392138239 · Published 11-13 13:42 · Downloads 119

Unidirectional "Asian Idol Face" beauty slider LoRA: Positive values simultaneously whiten and cool skin tone, narrow the face shape, soften facial features, and smooth out pores and freckles, while maintaining realistic photography for camera and lighting.

Suitable for creating a four-panel comparison with the same seed at 0 / 0.5 / 1.0 / 1.5.

Images

Example 1430x2012

Reasons for Recommendation

  • New model; serves adjustment of retouching levels for realistic portraits—progressively pushing from natural skin to a magazine-retouched look with the same prompt, without changing the text.
  • Strength tiers are clearly defined: 0.5 retains freckles and pores, 1.0 is the recommended value, 1.5 nearly eliminates pores and freckles, and 2.0 results in an extremely narrow face with details starting to flatten. The author explicitly states that negative values do not make it more natural but slide towards an anime style.
  • Complete settings: Paired with MOSS_8-steps-dmd2-lora 1.0, 8 steps, CFG 1.0, euler + simple, 1024x1360; no trigger word required; free, with a license allowing commercial use and derivatives; 119 downloads and 3 comments since release.

How to Use

  • Load and adjust strength directly; commonly used range is 0.5–1.5; if it looks too retouched, lower to 0.3. Do not use negative values.
  • Author's combination: MOSS_8-steps-dmd2-lora @ 1.0 + this slider, 8 steps, CFG 1.0, euler / simple, resolution multiples of 16.

Risks and Limitations

  • Only 1 example images (comparison with same seed at 0 / 0.5 / 1.0 / 1.5).
  • The effect pushes faces towards a single aesthetic of fair, narrow features, erasing personal characteristics like freckles. The author states that negative values starting from −0.5 cause skin to turn orange and oversaturated, and below −1.5 it becomes a flat anime face.

[A] Y2K Neon | MOSS · MOSS

LoRA · New Model · Author 858654128784 · Published 11-13 14:34 · Downloads 59

Y2K Neon concept LoRA: Millennium-era dreamy neon color palette—purple-pink skies, glowing city skylines and water reflections, often featuring a huge moon and starry sky, paired with silver reflective jackets and high-gloss materials.

Images

Example 896x1152

Reasons for Recommendation

  • New model; the author states it works for architecture, interiors, landscapes, still life, and characters, making it one of the broader-application atmosphere LoRAs this issue.
  • Usage is clear: Start with "y2kneon," followed by short comma-separated phrases (subject, details, background), with complete example sentences provided.
  • Free, with a license allowing commercial use and derivatives.

How to Use

  • Write following the structure of y2kneon, glass-fronted house on a rocky grassy slope, huge moon and starry sky, wide view: trigger word + subject + details + background
  • If you don't want the moon, specify the background clearly; strength, steps, and CFG were not provided by the author, and the page does not list the trigger word either

Risks and Limitations

  • Only 3 visible examples are shown, 5 are hidden, and the page contains mixed content
  • The author states that a large moon may appear even if not specified, so it needs to be actively suppressed in the prompt

【A】Billiard Hall | MOSS · MOSS

LoRA · New Model · Author 858654128784 · Published 11-13 15:44 · Downloads 60

Concept LoRA for old-style billiard hall scenes: green felt tables, brown wooden frames, low-hanging dark green metal lamps, wood paneling, and cue racks; capable of generating eye-level photos as well as high-angle surveillance footage with white timestamps.

The same author also released three concept LoRAs using the same template this period: gym, dormitory poster, and Y2K neon. This entry's dual-perspective design is the most distinctive.

Images

Example 896x1152

Reasons for Recommendation

  • New model; serves billiard hall scene photos, cinematic framing, and pseudo-surveillance screenshots; the author states that half of the training set consists of eye-level photos and half of high-angle CCTV footage, so both angles can be generated
  • Starting sentences are provided: use "An eye-level photograph shows a billiards hall." for eye-level views, and "A high corner surveillance view looks across a billiards hall ..." for surveillance perspectives, adding a timestamp sentence for surveillance images
  • Free; the license allows commercial use and derivatives

How to Use

  • Start with "billiardhallconcept." (trigger word followed by a period), then write the angle sentence, character actions (aiming, setting up balls), lights, and walls; you can also describe an empty venue after closing time
  • For the surveillance version, add "White timestamp and camera text appear along the upper edge."; strength, steps, and CFG were not provided by the author

Risks and Limitations

  • Only 2 visible examples are shown, 4 are hidden, and the page contains mixed content; the trigger word is not listed on the page
  • Pseudo-surveillance images should not be passed off as real surveillance screenshots

Worth Watching (14)

【B】MOSS-FET · v0.7

Base Model / Checkpoint · Version Update · Author aueki4g467 · Published 11-13 20:03 · Downloads 33

MOSS-FET v0.7: An experimental base model where the author performed full fine-tuning on MOSS using Fizgig; this version shifts further towards a realistic style from previous versions' anime/semi-realistic mix.

Images

Example 1024x1536

【B】AG: Sitri - Niji 7 Girls · MOSS-Image

LoRA · Ported · Author Azriaelis · Published 11-13 09:10 · Downloads 74

MOSS version of Sitri from the Ars Goetia series: A semi-realistic Niji-style female character art style LoRA trained on mixed materials generated by Midjourney / NijiJourney.

Images

Example 1024x1536

【B】Chainsaw Man Anime Style | MOSS Image · Movie Reze arc

LoRA · Version Update · Author HissingCat · Published 11-13 17:45 · Downloads 43

New version of the Chainsaw Man anime style LoRA: The version is named after the theatrical release "Reze Arc" and trained on movie visuals; the previous version was based on the TV Season 1 style.

Images

Example 896x1184

【B】Mistline · MOSS-v1.0

LoRA · New Model · Author soda_Cookie · Published 11-13 17:43 · Downloads 35

Mistline (Wu Cai Xian Miao) anime illustration style LoRA; the Chinese name points to fine lines + misty colors, but the author provides no textual description of the art style characteristics.

Images

Example 832x1216

【B】clearshade · MOSS-v1.0

LoRA · New Model · Author soda_Cookie · Published 11-13 17:38 · Downloads 31

Clearshade (Cheng Ying Man Hui) anime illustration style LoRA: Example prompts emphasize the clear shadow shapes left by strong side lighting on bangs, gloves, and skirt folds, corresponding to "Cheng Ying" in the Chinese name; the author provides no other art style descriptions.

Images

Example 1024x1024

【B】Porcelia · Moss-v1.0

LoRA · New Model · Author soda_Cookie · Published 11-13 19:22 · Downloads 22

Porcelia (Qing Ci Ping Cai) anime illustration style LoRA; the Chinese name points to porcelain-like texture and flat coloring, but the author provides no textual description of the art style characteristics.

Images

Example 1024x1024

【B】Cyber Graphic · Moss

LoRA · Ported · Author CN_Signal · Published 11-13 15:22 · Downloads 50

MOSS version of Cyber Graphic: A style LoRA combining cyberpunk + computer art + poster graphics.

Images

Example 1280x1920

【B】Neon Style [MOSS] · Neon Style

LoRA · New Model · Author RalFinger · Published 11-13 16:56 · Downloads 29

MOSS version of RalFinger's neon style LoRA (tags: neon / light); the model card contains no art style descriptions.

Images

Example 832x1248

【B】lumiline · moss-v1.0

LoRA · Ported · Author soda_Cookie · Published 11-13 15:11 · Downloads 43

MOSS version of the Lumiline art style LoRA: An anime illustration style by the same author, soda_Cookie; the author provides no textual description of the art style characteristics.

Images

Example 1024x1536

【B】Gym · MOSS

LoRA · New Model · Author 858654128784 · Published 11-13 16:15 · Downloads 55

Concept LoRA for ordinary gym scenes: black rubber flooring, dumbbell racks, squat racks, treadmills, bench presses, and large windows; includes basements, industrial-style exposed beams, and small private training rooms; characters are tying shoelaces, wiping benches, or walking on treadmills with natural, non-posed actions.

Images

Example 896x1152

【B】Layered Papercut Style [MOSS] · Layered Papercut

LoRA · New Model · Author RalFinger · Published 11-13 20:22 · Downloads 10

MOSS version of RalFinger's layered papercut style LoRA (tags: layered / paper cut); the model card contains no art style descriptions.

Images

Example 832x1248

【B】Dormitory Wall Poster — Pin-Up Posters of Men on Dorm and Locker-Room Walls | MOSS · MOSS

LoRA · New Model · Author 858654128784 · Published 11-13 10:29 · Downloads 68

Concept LoRA for dormitory wall posters: Generates a vertical paper poster stuck or pinned to a wall, featuring male figures such as basketball, football, tennis, boxing athletes, guitarists, or singers; the paper surface has creases and reflections, surrounded by metal lockers, bunk beds, or desks.

Images

Example 896x1152

【B】KIDRIOT STYLE · v1.0

LoRA · New Model · Author Vince_AI · Published 11-13 11:44 · Downloads 1 · Paid

KIDRIOT urban graffiti chibi style LoRA: Street graffiti + designer toys + big-head chibi characters, featuring thick black outlines, exaggerated proportions, vibrant colors, and clean cel shading.

Images

Example 896x1152

【B】MossImage Layout Editor · v0.1

Other · New Model · Author hybskgks28275 · Published 11-13 11:40 · Downloads 25

MossImage Layout Editor: Turns the "Create Bounding Boxes (MossImage)" node in ComfyUI into a standalone Windows desktop application, allowing users to drag boxes for layout and export MOSS JSON layout prompts without opening ComfyUI.

Included Only (22)

  • Realistic Ada Wong - Resident Evil - MOSS LORA · v1.0 — LoRA: A realistic character LoRA for Ada Wong from Resident Evil, with trigger strings split into two phrases for appearance and clothing. Example 1280x1840
  • Prismcoat · Moss-v1.0 — LoRA: Prismcoat (Liu Guang Hou Cai) anime illustration style LoRA; the Chinese name refers to a glossy, thick-paint aesthetic. The author provides no textual description of the style characteristics. Example 832x1216
  • Princess Zelda (The Legend of Zelda: Twilight Princess) [MOSS Image] · v1.0 — LoRA: A character LoRA for Princess Zelda from The Legend of Zelda: Twilight Princess, with a trigger string consisting of a full sequence of appearance tags. Example 1024x1344
  • Helen Parr - The Incredibles - MOSS LORA · v1.0 — LoRA: A character LoRA for Helen Parr, the superhero mom from The Incredibles, capable of generating realistic or Pixar-style 3D renders. Example 1280x1840
  • Petalmist · MOSS-v1.0 — LoRA: Petalmist (Hua Wu Rou Hui) anime illustration style LoRA; the Chinese name refers to a soft style featuring petals and mist. The author provides no textual description of the style characteristics. Example 832x1216
  • Carys · MOSS v1.0 — LoRA: A character LoRA for the fictional female character Carys (MOSS version): approximately 35 years old, with a blonde bob and an athletic build. The author states that all training images were AI-generated. Example 1248x1680
  • Doll Likeness (MOSS) · Dutch Dolls — LoRA: The Dutch Dolls version of Doll Likeness: rather than targeting a specific person, it imbues generated females with Dutch facial features. The author categorizes this as a clothing LoRA, used to swap models for their other clothing LoRAs. Example 768x1088
  • xMarna · MOSS Image — LoRA: xMarna: A character LoRA for a fictional young Taiwanese woman (messy double buns). The author states that all training images were AI-generated. Example 1568x2016
  • IREADING - STYLE - MOSSImage+ANIMA · MOSS v1.0 — LoRA: The MOSS version of an anime-style LoRA imitating the art style of X platform artist ireading, including weights from epochs 8, 9, and 10. Diagram 1086x1448
  • Genevieve NordhavenLove · MossImage — LoRA: A character LoRA for the fictional woman Genevieve NordhavenLove, set as a boatbuilder repairing wooden boats by the Swedish coast. The author declares that this does not correspond to a real person. Example 832x1248
  • Silkshade · Moss-v1.0 — LoRA: Silkshade (Man Hui) style LoRA: The model card contains only the name and trigger words. Example 832x1216
  • xThick · MOSS Image — LoRA: xThick: A female body type LoRA that adjusts the figure to a fit yet curvy hourglass shape. The author states it does not affect the head or face.
  • xKorina · MOSS Image — LoRA: xKorina: A character LoRA for a fictional young blonde woman, including an "adult braces" appearance tag. The author states that all training images were AI-generated.
  • Moss Image AI Babe Pack ( Part 02 ) by Sarcastic TOFU · MI_Babe_Houri_v1 — LoRA: A character LoRA for the fictional Iranian woman Houri (dark brown curly hair, light hazel-green eyes) from Part 2 of the Moss Image AI Babe Pack. Example 1141x1636
  • Velvetglow · Moss-v1.0 — LoRA: Velvetglow (Rou Hui) style LoRA: The model card contains only the name and trigger words. Example 1024x1024
  • Moss-yulNan Shuishui · MOOS-yulNan Shuishui. — LoRA: A character LoRA for the ancient-style character "Nan Shuishui": black hair in an updo with floral hairpins and dangling ornaments, a pink-purple modified qipao (with gold cloud embroidery, frog buttons, and tassels), and pink stiletto heels. The author describes it as having a BJD doll texture. Example 1536x1536
  • Moss Image AI Babe Pack ( Part 01 ) by Sarcastic TOFU · MI_Babe_Maya_v1 — LoRA: A character LoRA for the fictional Indian woman Maya (straight black hair, brown eyes) from Part 1 of the Moss Image AI Babe Pack.
  • PromptHelper [html] - Kromoss - StillLife Studio · v1.0 — Other: A prompt helper for Kromoss (a MOSS fine-tuned base model): a small HTML zip package containing 30 examples of studio photography prompts for food/drinks still life. Example 1440x1440
  • Luo Tianyi | LouTianyi_Anima_v1 · moss_v1.0 — LoRA: A character LoRA for the virtual singer Luo Tianyi (MOSS version): The author states that MOSS originally lacked the concept of the Luo Tianyi character. Example 2304x2048
  • Moss Image AI Babe Pack ( Part 04 ) by Sarcastic TOFU · MI_Babe_Luljeta_v1 — LoRA: A character LoRA for the fictional Albanian woman Luljeta (chestnut hair, reddish-brown eyes) from Part 4 of the Moss Image AI Babe Pack. Example 1024x1024
  • Moss Image AI Babe Pack ( Part 03 ) by Sarcastic TOFU · MI_Babe_Hemla_v1 — LoRA: A character LoRA for the fictional Norwegian woman Hemla (platinum hair, blue-gray eyes) from Part 3 of the Moss Image AI Babe Pack.
  • Long legs slider · Moss-image — LoRA: The MOSS version of the Long Legs Slider LoRA: adjusts the leg length ratio of characters.

Data: Civitai public API (base model MOSS Image), collection window 11-13 08:31 → 11-13 21:01 (Beijing Time). Recommendation reasons have been reviewed by editors; download counts only reflect attention.

Civitai

Explore the latest AI Radar updates