Zhenzhen's AI Radar · MOSS Image · 2026-11-04 Morning Report
This issue features 41 new versions: Strongly Recommended 1 · Recommended 8 · Worth Watching 11 · Included Only 21
Editor's Note This issue reviews 42 new MOSS versions. Top pick: Polaroid instant film style LoRA: renders any scene as a Polaroid photo with a wide white bottom border, faded warm tones, and flash reflections. Trigger words, strength, dimensions, steps, CFG, sampler, and fixes for missing frames are all provided; training image sources are public. Tier A includes BBOX Prompter v2 for region-specific prompts, anime-fusion base models MOSS_LUC v3 and cocoamix BF16 version, Warhammer Ultramarines armor, Chenbo and Orcavice art styles, gladiator sandals, and amputee character LoRAs. Of the 9 auto-S entries, only 1 were kept; the rest were downgraded due to few examples, parameters provided only for other base models, or being merely precision versions. Entries with overly short descriptions, title-only entries, and standard character LoRAs are rated C; the paid corrugated cardboard style is rated B.
Strongly Recommended (1)
[S] SEEDRAAI | Polaroid Instant Film | MOSS Image · v1.0
LoRA · New Model · Author cfgscale · Released 11-03 23:13 · Downloads 98
Polaroid instant film style LoRA: renders any scene as a Polaroid photo with a wide white bottom border, faded warm tones, soft shadows, and flash reflections.
Top pick for this group: The use case is clear in one sentence, parameters and troubleshooting tips are complete, and training data sources are public. The author also has a Qwen Image 2.1 version.
Images
Reasons for Recommendation
- New model: The author claims it was trained for 3000 steps on MOSS Image SFT using 54 real Polaroid photos. Training images are CC BY / CC BY-SA licensed photos from Wikimedia Commons, with photographers listed.
- Serves the retro film feel in text-to-image: applicable to portraits, nightclub flash photos, travel, food, cars, and interiors. The author claims it focuses on texture and is theme-agnostic.
- Actionable: Trigger words, prompt opening templates, strength, resolution, steps, CFG, and sampler are all specified, along with a fix for when the frame disappears. The author claims example image metadata includes prompts.
- Free, license allows output usage; less than half a day after release, it has 98 downloads (indicating attention only).
How to Use
- Load p0laroid_moss_v1.safetensors, start the prompt with "p0laroid, a polaroid instant photo, <your scene>"
- Strength 1.0 (reduce to 0.8 if faces distort); 896×1088, 25 steps, CFG 2.5, er_sde + simple
- Square images work, but the author claims the frame appears less often; if the frame disappears, add "white polaroid border with a wider bottom edge" to the end of the prompt
- Avoid the words "frame" and "cropped" in scene descriptions, as the author claims they are interpreted as "viewfinder" and cause the white border to disappear
Risks and Limitations
- Only 4 visible examples (plus 3 sensitive-level examples hidden); cross-theme performance relies mainly on the author's claims
- Training set contains only 54 images; the author warns that faces may distort at strength 1.0; this entry was not tested
Trigger Word: p0laroid
Recommended (8)
[A] MOSS_LUC [ moss-image-checkpoint ] · sft-int8-v3
Base Model / Checkpoint · Version Update · Author ukr8b3g201 · Released 11-04 00:38 · Downloads 62
v3 (SFT INT8 quantized version) of MOSS_LUC, an anime-oriented MOSS-Image fusion base model. The author claims it stabilizes scheduling, adjusts contrast, and makes the art style "cuter."
Downgraded from auto-S to A: Parameters are copy-paste ready, but the update notes are only three sentences long and visible examples are few.
Images
Reasons for Recommendation
- Version update: The version update notes section is empty; the V.3 section of the model card lists only three points—stabilized general scheduler, adjusted contrast, and more KAWAII than before.
- Serves anime and illustration styles in text-to-image; the author claims it is a full model fused from their Lucy series and quantized to INT8 via convrot, not suitable for realism.
- Complete parameters: Sampler, scheduler, steps, CFG, Shift value, accompanying DMD acceleration LoRA and its strength, and positive/negative prompts are all specified.
- Free, permissive license (allows selling and selling fusions); this version has 62 downloads, model cumulative 378 downloads (indicating attention only).
How to Use
- Switch base model to mossLUCMossImage_sftInt8V3.safetensors, no trigger word
- DPM++ SDE or DPM++ 2M SDE; Beta scheduler or MOSS Scheduler (Shift, value 5); 8–10 steps (starting from 8), CFG 2–5 (starting from 3, lower for realism)
- Pair with lodestones' MOSS-Image-SFT-8-steps-dmd acceleration LoRA (lora_latest.safetensors), strength 0.7–0.8
- The author recommends installing their scheduler extension: sd-webui-forge-neo-moss-scheduler for Forge Neo, ComfyUI-MOSS-Shift-Scheduler for ComfyUI
- Positive prompts can use "masterpiece, best quality, amazing quality"; negative prompts "worst quality, low quality, lowres, bad anatomy, bad hands, jpeg artifacts," etc.
Risks and Limitations
- The author did not write update notes (version notes section is empty); V.3 changes are only three sentences in the model card, making it hard to see the difference from v2
- 6 visible examples, 14 sensitive-level examples hidden (more than visible); realism is not a strength
- The recommended settings of 8–10 steps rely on the DMD acceleration LoRA, and the author's custom scheduler extension is also recommended; the INT8 full model is about 5.3 GB; this entry was not tested
[A] cocoamixMOSS (moss image) · BF16 v1.0
Base Model / Checkpoint · Version Update · Author ccooccoo · Released 11-04 05:07 · Downloads 32
cocoamix series MOSS Image fusion base model (BF16 version). The default art style does not draw pupils for characters, capable of producing both anime and realistic outputs.
Downgraded from auto-S to A: Configuration and prompt writing are clear, but the license prohibits commercial use, no sampler is provided, and this version is merely a precision update, not a new capability.
Images
Reasons for Recommendation
- Version notes: This is not a capability update but the BF16 precision version of the same v1.0, released simultaneously with the 8INT 10 step version; the update notes section is empty.
- The author claims it is a fusion of MOSS-Image and their own trained model; the default cocoamix style does not draw pupils unless "dark_circle shaped pupils" is specified; tags include realistic, anime, mecha, and chibi.
- Three output configurations are provided: BF16 + 8 steps DMD2 LoRA (marked as best by the author), INT8 10 steps, and BF16 without LoRA 50 steps CFG4, along with prompt writing for vivid colors and realistic photos.
- Free; 9 example images; 32 downloads in about 4 hours after release (indicating attention only).
How to Use
- Switch base model to cocoamixmossMossImage_bf16V10.safetensors, no trigger word
- Recommended: Attach 8-steps-dmd2 LoRA, 8 steps, CFG 1; without LoRA, use 50 steps, CFG 4 (author claims it takes longer)
- For pupils, write "eyes has dark_circle shaped pupils"; for vivid colors, write "vivid dark color"; for realistic photos, write "photorealistic style has realistic background, realistic hair, realistic face ..."
- Sampler, scheduler, and resolution are not provided by the author
Risks and Limitations
- Strict license: Commercial use is prohibited, generated images are for non-commercial use only; selling generated images or publishing fusion models requires prior author permission; reposting to unauthorized sites incurs fees.
- The author did not write update notes; this version and the 8INT 10 step version differ only in precision, choose one.
- The BF16 full model is about 9.5 GB; the author warns about sensitive content; 1 sensitive-level examples are hidden; this entry was not tested
[A] The Space Marines Warhammer 40K | Flux + Pony + illustrious + Anima + Krea2 · Ultramarines MOSS
LoRA · Port · Author VelvetS · Released 11-04 05:09 · Downloads 32
MOSS port of the Warhammer 40K Space Marines power armor LoRA. This version features Ultramarines livery, using a trigger word to equip characters with full armor.
Downgraded from auto-S to A: Use case is clear and examples are plentiful, but MOSS version parameters can only be inferred from older versions.
Images
Reasons for Recommendation
- Port: The series already has versions for Flux, Pony, Illustrious, Anima, Krea 2, etc., released separately by chapter; MOSS currently only has the Ultramarines version.
- Serves Warhammer-themed character images and sci-fi armor concept art; the author claims the purpose is to equip your character with Space Marines power armor.
- Has a dedicated trigger word UltraM40k; 14 visible examples.
- Free, permissive license; the entire model has cumulative 35906 downloads, mostly from older base model versions (indicating attention only).
How to Use
- Load MOSSUltraM40k.safetensors, include "UltraM40k" in the prompt
- The author did not provide strength, steps, or CFG for the MOSS version; the strength of 0.6-1 (preference 0.9) in the model card refers to the Flux / Pony versions and can serve as a starting point.
Risks and Limitations
- The model card contains only instructions for older base models (Flux / Pony), with no MOSS-specific parameters or instructions.
- Warhammer 40K is a commercial copyright IP; assess commercial output risks yourself; 6 sensitive-level examples are hidden; this entry was not tested
Trigger Word: UltraM40k
[A] Chenbo (Bo Chen) - Artist Style · MOSS-Image
LoRA · Port · Author Azriaelis · Released 11-03 23:31 · Downloads 25
A LoRA imitating the style of former Riot artist Chenbo (Bo Chen), rendering characters as semi-realistic digital illustrations in the style of League of Legends splash art.
Downgraded from automatic S to A: MOSS parameters and examples are complete, but it is a style imitation of a living artist, and the required turbo LoRA is not specified.
Image
Reasons for Recommendation
- Port: Previously available for Anima (including high-resolution version) and Krea2Turbo; this time, parameters were specifically written for MOSS-Image: Base model + 8 step turbo LoRA
- Serves text-to-image game concept art and character promotional images, featuring a style of detailed semi-realistic digital painting
- Trigger word: chenbo; the author states that adding the trigger word significantly strengthens the style; 14 visible examples
- Free; this version has 25 downloads, with cumulative model downloads at 431 (indicating attention only)
How to Use
- MOSS-Image Base + 8 step turbo LoRA, then load CHENBO-MOSS-V1 (1).safetensors, and include "chenbo" in the prompt
- euler + simple, LoRA strength 0.8-1.2, 8-16 steps, CFG 1-3
- The author did not specify which turbo LoRA to use
Risks and Limitations
- Imitates the personal style of a living professional artist; commercial use requires self-assessment; the license does not allow publishing derivative models
- The description is very short, with most content being parameters for the Anima version; 6 sensitive-level examples are hidden; this entry has not been tested
Trigger Word: chenbo
[A] Orcavice Style - MOSS LoRa · v1.0
LoRA · New Model · Author Minokai · Published 11-03 21:12 · Downloads 60
A vibrant anime illustration style LoRA that uses the trigger word MKAI_0rcav1ce to steer generated images toward this artistic style.
Downgraded from automatic S to A: Parameters can be copied directly, but there are few visible examples, and it was not created on the official base model.
Image
Reasons for Recommendation
- New Model: The author claims it is a style LoRA for MOSS-Image, aiming for colorful, illustrative anime visuals
- Serves text-to-image anime character portraits; the author has publicly shared complete prompts, strengths, and seeds for the examples, allowing for reproduction
- Complete parameters: Starting strength 1.00, 50 steps, CFG 3, Euler + sgm_uniform, 832 × 1216, skip-layer guidance skips only a single layer 22 and is enabled throughout
- Free with a permissive license; 60 downloads within half a day of release (indicating attention only)
How to Use
- Load MKAI_Orcavice_Style_v1_MOSS.safetensors, start the prompt with "MKAI_0rcav1ce", strength 1.00
- 50 steps, CFG 3, Euler + sgm_uniform, 832 × 1216; skip-layer guidance single layer 22, do not skip double layers, active from 0–100%
- The base model used in the author's examples is MOEMOSS-Base, not the official MOSS base model
Risks and Limitations
- Only 3 visible examples; among the 4 examples described by the author, only 1 use the trigger word and strength 1.00, while the rest are early 0.80 samples without the trigger word
- All examples were generated on the third-party fine-tuned base model MOEMOSS-Base; effects on the official base model are unknown; examples frequently feature well-known game and anime characters; 1 sensitive-level examples are hidden; this entry has not been tested
Trigger Word: MKAI_0rcav1ce
[A] Moss Image BBOX Prompter · v2.0
Workflow · Version Update · Author ukr8b3g201 · Published 11-04 01:24 · Downloads 64
MOSS-Image ComfyUI region box prompt editor v2: Place boxes on the canvas, write separate prompts for each box, and newly added automatic generation of descriptions within boxes from reference images.
Unlike the concurrent Compact BBox workflow, this is a standalone node usable in the classic interface; Compact requires enabling Nodes 2.0.
Image
Reasons for Recommendation
- Version Update: v2.0 adds reference image analysis—select a reference image and a box, run it to automatically generate descriptions of people or objects within that box; turning off analysis preserves the result for manual editing; target can be Auto or a specified box (including sub-images); V1 workflow remains compatible
- Serves text-to-image tasks requiring composition control: Place people, objects, and text into different boxes, suitable for multi-character scenes, posters, and layouts with dialogue text
- The author's GitHub documentation states that analysis directly uses the Qwen3.5-9B INT8 text encoder already loaded by MOSS, requiring no additional models or cloud APIs; reference images are converted only into text descriptions, not used directly as image conditions
- MIT license, free; includes Standard and DMD example workflows; this version has 64 downloads, with cumulative model downloads at 780 (indicating attention only)
How to Use
- Installation: In ComfyUI's custom_nodes directory, git clone https://github.com/ukr8b3g-cmyk/MossImage-BBOX-Prompter.git，重启 into ComfyUI and refresh the browser (press Ctrl+F5 for residual old interface elements)
- Use the V2 workflow, select a reference image in Load Image; set Image analysis enable to true and Run normally, results are written to the target box; switch back to false to manually edit and reuse
- Additional models required: MOSS-Image SFT (ComfyUI INT8) or the author's MOSS_LUC, Qwen3.5-9B INT8 text encoder, FLUX.2 Small Decoder VAE; DMD workflow additionally requires MOSS-Image SFT 8 step DMD acceleration LoRA
- Supports ComfyUI classic node interface, no need to enable Nodes 2.0
Risks and Limitations
- Re-running with analysis ON will overwrite manually edited prompts; each run analyzes only one reference image for one box, text boxes are not analyzed
- The author does not guarantee restoration of character identity, text, or exact reproduction of the reference image; the analysis workflow was verified by the author only on a single RTX 5060 Ti
- Depends on custom nodes and multiple model files, requiring self-installation and verification; 1 sensitive-level examples are hidden; this entry has not been tested
[A] Flat Gladiator Sandals MOSS · v1.0
LoRA · New Model · Author gkhan_ · Published 11-04 03:16 · Downloads 35
Flat Roman gladiator sandal clothing LoRA: Generates knee-high, flat-soled, open-toed gladiator sandals with straps wrapping around the legs; strap thickness, complexity, and color can be changed via prompts.
Maintained at A, top recommendation in this group: Single-purpose but immediately understandable, with the clearest trigger sentence and optional tags; visible examples outnumber hidden ones.
Image
Reasons for Recommendation
- New Model: Clothing item LoRA, serving text-to-image tasks by specifying a particular style of sandals for characters (fashion, classical, summer outfits) without relying on long descriptive strings
- The author provides a minimal trigger sentence that can be copied directly, along with a set of optional tags: thick / thin straps, complex / simple, buckles, criss-cross wrapped cords, tied at the knee, braided wrap; colors can be written directly
- The author recommends strength 1; 4 visible examples, 1 sensitive-level examples are hidden
- Free; 35 downloads within approximately 6 hours of release (indicating attention only)
How to Use
- Load Clothing_Sandals_Gladiator_Moss_v1.safetensors, strength 1
- Use the author's minimal trigger sentence in the prompt: "gldsandals, a woman wearing knee-high gladiator sandals, flat sole, open toe, thick straps, more intricate, horizontal cage straps, metal studs, back zipper", then replace with optional tags such as thin straps / less intricate / buckles / criss-cross wrapped cords / tied at the knee / braided wrap as needed; write colors like black, brown, tan
- The author did not provide steps, CFG, sampler, or applicable base models (Base / SFT)
Risks and Limitations
- Narrow application scope, covering only one type of footwear; the author states that optional tags "vary in effect by image"
- The trigger sentence starts with "a woman"; the author did not specify applicability to male characters or non-character scenes
- License does not allow derivatives and requires attribution; this entry has not been tested
Trigger Word: gldsandals, a woman wearing knee-high gladiator sandals, flat sole, open toe, thick straps, more intricate, horizontal cage straps, metal studs, back zipper
[A] Moss SAK Amputee · v1.0
LoRA · New Model · Author droidy001142 · Published 11-04 02:29 · Downloads 24
Limb difference character LoRA: Generates female characters with unilateral above-knee amputation using forearm crutches; prompts can specify left or right leg, and whether the residual limb is visible or covered by a skirt / pants.
Maintained at A: Uncommon direction with practical demand, parameters and description sentences are specific; the author admits stability is average, suitable for those willing to generate multiple times.
Image
Reasons for Recommendation
- New Model: Addresses body feature representation where the base model is weak, serving portraits, illustrations, and character design sheets that need to accurately depict amputees and crutch users
- No trigger word; the author lists description sentences used during training, which can be directly incorporated into prompts: left / right above-knee single leg amputation, residual limb visible / not visible / covered by skirt or pants, walking / standing with forearm crutches
- The author provides strength 0.8-1.0, CFG 1.0-1.2, and notes that example images simultaneously stack the 8 step DMD2 acceleration LoRA and moss_image_sft LoRA
- Free; 3 visible examples; 24 downloads within approximately 6 hours of release (indicating attention only)
How to Use
- Load Moss_amputee.safetensors, strength 0.8-1.0, CFG 1.0-1.2; author's examples pair with 8-step-dmd2-lora and moss_image_sft LoRA
- Include the author's description sentences in the prompt, such as "left above knee single leg amputee", "right above knee single leg amputee", "her residual limb is visible", "her residual limb is covered by her skirt", "walking with Forearm crutches", "stood with forearm crutches"
- The author states that sometimes disabling prompt enhancement yields better results; steps and sampler were not provided by the author
Risks and Limitations
- The author describes it as experimental, "usable but imperfect," requiring repeated trials; notes that MOSS shows little variation with different seeds, so changing prompts is necessary for different results
- Trained only on female characters with unilateral above-knee amputation
- 7 examples are classified as sensitive and hidden, outnumbering visible examples; license allows only in-site rental, no derivatives; this entry has not been tested
Worth Watching (11)
[B] Roseglass Conservatory · v1.0
LoRA · New Model · Author CelestialCanvas · Published 11-04 04:00 · Downloads 28
Original fantasy scene LoRA: Uses a trigger word to generate a glass conservatory with rose-colored crystal arches, platinum filigree decorations, filled with pink roses, and reflective water surfaces.
Image
[B] Jadetilaurant Style [Anima/Illustrious/Krea2/MOSS] · MOSS v1.0
LoRA · Port · Author Imagine__That · Published 11-03 21:42 · Downloads 79
A LoRA imitating the style of artist jadetilaurant (art for "Reverse:1999"), featuring a thick-paint style with sketch-like brushstrokes.
Image
【B】MOSS-RiceBranOilMix · v01.0
Base Model / Checkpoint · Version Update · Author toshi_autumn · Published 11-04 00:46 · Downloads 37
The official release v01.0 of the anime illustration fusion base model MOSS-RiceBranOilMix, which adds coloring and an oil painting feel, rendering eyes and eyelashes more delicately.
Image
【B】Compact BBox T2I Moss Image · v1.0
Workflow · New Model · Author NeuralVandal · Published 11-04 03:50 · Downloads 34
A streamlined MOSS Image bounding box (BBox) text-to-image workflow that divides the canvas into multiple regions, each with its own prompt.
Image
【B】Vo1ez Style · MOSS-Image v1.0
LoRA · Port · Author emerjk · Published 11-04 07:35 · Downloads 10
A MOSS version of the LoRA mimicking artist Vo1ez's style. The author states the images feature high contrast and sharp lines, with Chiaki Nanami from Danganronpa appearing so frequently in the training data that she can be generated by name directly.
Image
【B】cocoamixMOSS (moss image) · 8INT 10steps v1.0
Base Model / Checkpoint · New Model · Author ccooccoo · Published 11-04 04:06 · Downloads 30
The 8-bit integer (INT8) version of the cocoamix-style MOSS fusion base model: it integrates a 8-step DMD2 acceleration LoRA, and the author claims images can be generated in 10 steps with CFG 1.
Image
【B】V67 Melancholic Scenarios · V1 MOSS
LoRA · Port · Author angelomaiota · Published 11-04 00:17 · Downloads 41
A MOSS version of the melancholic rainy atmosphere LoRA: featuring wet streets and cobblestones, fog, low-saturation blue-gray tones, solitary figures, and abandoned old objects.
Image
【B】Construction Site — Scaffolding, Rebar, Hard Hats, Site Cabins | MOSS · v1.0
LoRA · New Model · Author 858654128784 · Published 11-03 21:19 · Downloads 80
Construction site scene LoRA: featuring scaffolding, exposed rebar, concrete, tower cranes, hard hats, and reflective vests, alongside workers' site cabins.
Image
【B】Realistic Faces · v1
LoRA · New Model · Author fyrtyv · Published 11-04 04:04 · Downloads 103
Realistic facial detail LoRA: the author claims it adds pores, wrinkles, fine vellus hair, and more detailed eyes to faces.
Image
【B】[Moss] Male/Female Administrator Dual LoRA Endministrators | Arknights: Endfield · v1.0
LoRA · New Model · Author rayzerosr · Published 11-04 01:35 · Downloads 19
A dual-character LoRA for the male and female Administrators from Arknights: Endfield: one LoRA contains two characters with two trigger words, used to generate images of both together, trained on MOSS-Image SFT.
Image
【B】Corrugated Cardboard [Krea2] · MOSS v1.0
LoRA · Port · Author TeraWatt · Published 11-04 04:06 · Downloads 0 · Paid
A MOSS version of the corrugated cardboard craft style LoRA: the author claims it can turn any prompt into handmade crafts made of cardboard, featuring cardboard flutes, irregular edges, and a rough texture.
Image
Included Only (21)
- [OC] Athene Character LoRA for Krea 2 · 1.0 (MOSS) — LoRA: Character LoRA for the fictional adult female character Athene, ported from Krea 2 to MOSS, trigger word athnblk.
- MOSS_Realism_sft_bf16 · bf16 — Base Model / Checkpoint: Realistic-oriented full MOSS model MOSS_Realism_sft_bf16, accompanied by a 8-step acceleration LoRA and workflow; the author claims it focuses on realistic skin and cinematic lighting.
- Vlada - Inked Braids (Original Character) · v1.0 MOSS — LoRA: LoRA for the AI-synthesized fictional female character Vlada: twin braids, bleached front hair, chest decorative tattoo, leaning towards realistic street photography portraits.
- Rose Fae · v1.0 — LoRA: Character LoRA for the original fantasy character Rose Fae: rose-pink side-braided thick braid, amber eyes, long pointed ears, translucent pink petal wings, platinum floral accessories.
- Ast-light | Yukina (東雲ゆきな) — Character LoRA · MOSSv1.0 — LoRA: MOSS version of the character LoRA for Yukina (東雲ゆきな) from the original idol project Ast-light: silver hair with a light purple gradient, half-up braided hair, purple eyes, and a beauty mark under the left eye.
- Diya Von Vergin · MossImage — LoRA: MOSS version of the character LoRA for the fictional female character Diya Von Vergin; the same author released versions for multiple base models around the same time.
- MossArt · v3.0 — LoRA: v3 of the MossArt style LoRA trained on a mix of multiple art styles; the author only states that the training images cover hundreds of styles including illustration, anime, and webtoons, and recommends lowering the strength.
- [MOSS] Future Julie Power and Katie Power (Power Pack, Earth-68326) · v1 — LoRA: Dual-character LoRA for the "future adult versions" of sisters Julie Power and Katie Power from Marvel's Power Pack, created based on a commission.
- Elise : Perfect Curves · MOSS — LoRA: MOSS version of the character LoRA for the fictional female character "Elise"; the model card contains only a single promotional sentence.
- Kigurumi Onesie: Jester [Krea2/MOSS] · MOSS_SFT_v1.0 — LoRA: MOSS SFT version of the Jester outfit LoRA from the AuthentiQ kigurumi onesie series: according to the title and tags, it makes characters wear a jester-themed kigurumi onesie.
- MOSS 35mm · v1.0 — LoRA: 35mm simulated film texture LoRA; the author claims it is triggered by "analogue film" or similar terms.
- Mr Piglets Oldies and brats Moss image · Lux48 — LoRA: Lux48 from the Mr_Piglet female character LoRA collection: this model page releases many characters as separate "versions," Lux48 is one of them, and the trigger word is simply Lux48.
- Mavis Dracula - Scrag Boy · Moss — LoRA: MOSS version of the character LoRA for Mavis Dracula from Hotel Transylvania, featuring a set of 3D / CGI art style trigger words; the author does not explain "Scrag Boy" in the title.
- Seaweed Dress (MOSS) [Anime & Realism] · v1.0 — LoRA: Seaweed dress outfit LoRA: makes characters wear a dress composed of seaweed and shells; the title claims it works for both anime and realism.
- L-shaped sideburns · v0.2 — LoRA: "L-shaped sideburns" hairstyle detail LoRA; the author marks it as a v0.2 test version still in progress.
- Selfie Stick Selfie · moss-image_v1.0 — LoRA: MOSS version of the selfie stick selfie composition LoRA: generates illustrations of characters taking selfies with a selfie stick.
- Creeps art style · v1.0 MOSS-Image — LoCon: Style model imitating the art style of illustrator Creeps (artbycreeps.com); tags are furry and style.
- Mai Yoneyama Style · v1.0 — LoRA: LoRA imitating the art style of illustrator Mai Yoneyama; the author claims the training set was self-selected and self-tagged, with the trigger word @yoneyama.
- Mr Piglets Oldies and brats Moss image · Lux49 — LoRA: Lux49 from the Mr_Piglet female character LoRA collection: another character from the same collection page, with the trigger word simply being Lux49.
- Prompt Helper [html] - Kromoss - Analog · v1.0 — Other: HTML prompt helper tool configured for the Kromoss base model (a few KB compressed package); the description lists four film / photography style items.
- AuthentiQ Onesie: Capybara [Krea2/MOSS] · MOSS_SFT_v1.0 — LoRA: Capybara version from the AuthentiQ onesie pajama series: makes characters wear a capybara-themed kigurumi onesie pajama, MOSS SFT version.
Data: Civitai Public API (Base Model MOSS Image), inclusion window 11-03 20:31 → 11-04 09:01 (Beijing Time). Recommendation reasons have been reviewed by editors; download counts only reflect attention.