MOSS Image Morning Report 2026-11-10 | Top Pick: RISOGRAPH

This issue reviews 36 new MOSS versions, with the top pick being the RISOGRAPH LoRA in Risograph print style: The MOSS version is trained separately, uses the dedicated trigger word silbriorisograph, and provides full details on strength, steps, CFG, and ink color schemes. Also included are three other styles by the same author (woodblock, color field, and Arcane style), a concept LoRA for adding eyes by body part, a cinematic still style, a Hanasaki Taru watercolor style, and the WeVisOne editing workflow (listed as A) which automatically generates masks from text for local inpainting. Several character/copyrighted character LoRAs with only one or two sentences of description have been downgraded from S/A/B to C. The nvfp4 and int8 versions of MOSSRang2-lite are two quantizations of the same paid base model, with int8 downgraded to C; Nyx's 1.0

Zhenzhen's AI Radar · MOSS Image · 2026-11-10 Morning Report

This issue features 33 new versions: Strongly Recommended 1 · Recommended 9 · Worth Watching 10 · Included Only 13

Editor's Note This issue reviews 36 new MOSS versions, with the top pick being the RISOGRAPH LoRA in Risograph print style: The MOSS version is trained separately, uses the dedicated trigger word silbriorisograph, and provides full details on strength, steps, CFG, and ink color schemes. Also included are three other styles by the same author (woodblock, color field, and Arcane style), a concept LoRA for adding eyes by body part, a cinematic still style, a Hanasaki Taru watercolor style, and the WeVisOne editing workflow (listed as A) which automatically generates masks from text for local inpainting. Several character/copyrighted character LoRAs with only one or two sentences of description have been downgraded from S/A/B to C. The nvfp4 and int8 versions of MOSSRang2-lite are two quantizations of the same paid base model, with int8 downgraded to C; Nyx's 1.0 (Krea 2) is a Krea 2 weight mistakenly labeled as MOSS and has been excluded.

Strongly Recommended (1)

[S] RISOGRAPH · MOSS Image v1.0

LoRA · Port · Author Silbrio · Published 11-10 07:39 · Downloads 18

The MOSS version of the Risograph style LoRA: featuring spot inks, fine halftone dots, ink overprinting, and exposed white paper, it renders people, landscapes, animals, and still life as illustrations with a printed grain texture.

Top pick for this issue: The MOSS version is trained separately with a dedicated trigger word, complete parameters and prompt writing guidelines, a clear style direction, and a low barrier to entry.

Images

Example 1344x2016

Reasons for Recommendation

  • Port: The same page already features Krea 2 and Anima versions, with the model page accumulating 448 downloads; the MOSS version is trained separately by the author on MOSS-Image-Base (60 curated images, 1,800 steps, rank/alpha 32), and the trigger word has been changed to the MOSS-specific silbriorisograph
  • Serves text-to-image generation for illustrations, posters, and editorial images: The author provides prompt writing guidelines—describe the subject and composition, specify two or three ink colors, and assign them to highlights, shadows, or overprint areas
  • Complete parameters: Strength 1.0; MOSS-Image-Base BF16, 25 steps, CFG 2.5, ER-SDE / Simple, Qwen3.5 9B INT8 text encoder + FLUX.2 VAE, prompt enhancement disabled; display sizes 1344 × 2016, 2016 × 1344, or 1664²
  • 10 examples; free, with license and statement files attached, the license allows for image output and on-site rental

How to Use

  • Write silbriorisograph in the prompt, then describe the subject, composition, and two or three ink colors, requesting halftone shadows, visible ink overprinting, and exposed white highlights; LoRA strength 1.0
  • MOSS-Image-Base, 25 steps, CFG 2.5, ER-SDE / Simple; if the image is too busy, the author suggests simplifying the scene, leaving more large areas of solid color and blank paper

Risks and Limitations

  • Do not copy the Krea 2 instructions in the lower half of the model card: The trigger word vtxrisograph and the 8 steps CFG 1 parameters there belong to the Krea 2 version
  • The author states that some comparison samples retain their original sizes; visible examples include 1248² and 1024 × 1536 images, but it is not specified which ones are not newly generated by MOSS; there are no same-seed comparisons with and without the LoRA
  • Parameters are based on the unaccelerated Base 25 steps; the author has not stated the performance with SFT or accelerated LoRAs; the license does not allow off-site rental or sale
  • This entry has not been run

Trigger Word: silbriorisograph

Recommended (9)

[A] MINERAL INK · MOSS Image v1.0

LoRA · Port · Author Silbrio · Published 11-10 08:08 · Downloads 13

The MOSS version of the woodblock/engraving illustration style LoRA: featuring engraved outlines, directional hatching, and warm-toned paper, paired with mineral colors like petroleum blue, ochre, charcoal black, and vermilion, making objects and scenes look like printed illustrations.

Downgraded from automatic S to A: Parameters are complete, but the training set is small, and the MOSS-specific instructions are only one paragraph; the rest must be distinguished from the Krea 2 version instructions.

Images

Example 1024x1536

Reasons for Recommendation

  • Port: The same page already features Anima and Krea 2 versions, with the model page accumulating 261 downloads; the MOSS version is trained separately on MOSS-Image-Base (24 curated images, 1,000 steps, rank/alpha 32), and the trigger word is changed to silbriomineralink
  • Serves text-to-image generation for illustrations of portraits, wildlife, plants, fantasy, and sci-fi scenes; the author's starting color palette is petroleum blue, ochre, charcoal black, and vermilion, and suggests specifying the subject and negative space in the prompt
  • Complete parameters: Strength 1.0; MOSS-Image-Base BF16, 25 steps, CFG 2.5, ER-SDE / Simple, Qwen3.5 9B INT8 + FLUX.2 VAE
  • 10 examples; free, license allows for image output and rental

How to Use

  • Write silbriomineralink in the prompt, describe the subject and negative space, and start with a color palette of petroleum blue, ochre, charcoal black, and vermilion; strength 1.0
  • MOSS-Image-Base, 25 steps, CFG 2.5, ER-SDE / Simple

Risks and Limitations

  • The main body of the model card contains Krea 2 version instructions: The Krea 2 version trigger word is vtxmineral, and the strength range and CFG 1 are also for Krea 2; the MOSS version requires silbriomineralink
  • The author states in the Krea 2 version instructions: At strength 1.0, realistic or anime prompts will be converted into woodblock illustrations, and simple backgrounds will be made more complex; to preserve the background, specify it clearly and lower the strength
  • The MOSS version training set has only 24 images; the author claims this is their first LoRA; the author states that most comparison samples use different seeds and should only be used as style references
  • This entry has not been run

Trigger Word: silbriomineralink

[A] SERRAVÉ · MOSS Image v1.0

LoRA · Port · Author Silbrio · Published 11-10 07:43 · Downloads 11

The MOSS version of the original painting style LoRA: filling large, rounded shapes with velvet-like matte rich colors, revealing adjacent colors only through a few short, sharp incisions at select edges, similar to shaping with pastel and pigment ink using a palette knife.

Downgraded from automatic S to A: Same author and template as RISOGRAPH, with complete parameters, but the training set is small, and the style's applicability is narrower than print-based styles.

Images

Example 1344x2016

Reasons for Recommendation

  • Port: The same page already features Krea 2 and Anima versions, with the model page accumulating 288 downloads; the MOSS version is trained separately on MOSS-Image-Base (25 curated images, 1,000 steps, rank/alpha 32), and the trigger word is changed to silbrioserrave
  • Serves text-to-image generation for painting-style illustrations of subjects like people and landscapes; the author's prompt writing guideline suggests a small color palette, one concentrated area of rich color, and a few quiet edge accents
  • Complete parameters: Strength 1.0; MOSS-Image-Base BF16, 25 steps, CFG 2.5, ER-SDE / Simple
  • 10 examples; free, license allows for image output and rental

How to Use

  • Write silbrioserrave in the prompt, pair with a small color palette, specify one area where the color is most concentrated and a few edge accents; strength 1.0
  • MOSS-Image-Base, 25 steps, CFG 2.5, ER-SDE / Simple

Risks and Limitations

  • The latter half of the model card contains Krea 2 version instructions, where the trigger word vtxserrave belongs to the Krea 2 version and cannot be used for the MOSS version
  • The training set has only 25 images; the style description is largely the author's self-description, with no same-seed comparisons with and without the LoRA
  • The page is marked with a mixed content rating; this entry has not been run

Trigger Word: silbrioserrave

[A] ARKANE · MOSS Image v1.0

LoRA · Port · Author Silbrio · Published 11-10 03:08 · Downloads 30

The MOSS version of the cinematic style LoRA inspired by the painterly feel of the animated series Arcane: featuring sculptural volumes, expressive faces, and large painted color fields, primarily targeting character scenes.

Downgraded from automatic S to A: Parameters are complete, but the examples have been upscaled, and it is a fan imitation of existing animated art.

Images

Example 2731x4096

Reasons for Recommendation

  • Port: The same page already features Krea 2 and Anima versions, with the model page accumulating 227 downloads; the MOSS version is trained separately on MOSS-Image-Base (60 curated images, 1,800 steps, rank/alpha 32), and the trigger word is silbrioarkane
  • Serves narrative illustrations for original characters and worldviews: The author suggests clearly describing the character and gestures, camera angle and lighting, and then providing a clear emotional moment (encounter, decision, pause)
  • Complete parameters: Strength 1.0; MOSS-Image-Base BF16, 25 steps, CFG 2.5, ER-SDE / Simple; 10 examples, free, license allows for image output and rental

How to Use

  • Write silbrioarkane in the prompt, then describe the character and gestures, camera angle, lighting, and emotional moment; strength 1.0
  • MOSS-Image-Base, 25 steps, CFG 2.5, ER-SDE / Simple, output sizes 1344 × 2016, 2016 × 1344, or 1664²

Risks and Limitations

  • The new example images are final products upscaled to a long side of 4096 using IllustrationJaNai DAT2 + Lanczos after generation, not direct MOSS output effects
  • The author states that complex interactions and special props require multiple attempts, and hands, props, and spatial relationships need to be checked
  • Fan style study, imitating copyrighted animated art; the author declares no affiliation with any studio
  • This entry has not been run

Trigger Word: silbrioarkane

[A] Zero_Q_0q - Artist Style · MOSS-Image

LoRA · Port · Author Azriaelis · Published 11-09 22:00 · Downloads 38

The MOSS version of the artist Zero_Q_0q style LoRA: ornate dark fantasy anime illustrations with thick paint textures, cinematic lighting, elaborate costumes, and a romantic fantasy atmosphere, with tags leaning towards bishonen and manhwa.

Downgraded from automatic S to A: MOSS parameters are fully provided, but it is merely an imitation of a single artist's style.

Images

Example 1024x1536

Reasons for Recommendation

  • Port: The same page already has Anima and Krea 2 versions, with a cumulative 432 downloads on the model page; the author provided parameters for three base models, and the MOSS version uses Base with a 8-step acceleration LoRA
  • Complete parameters: Trigger word Zero_Q_0q (the author claims adding it significantly strengthens the style); MOSS-Image Base + 8-step turbo LoRA, euler / simple, strength 0.8-1.2, 8-16 steps, CFG 1-3
  • 13 sample images; this version has 38 downloads and 5 likes (representing attention only)

How to use

  • Write Zero_Q_0q in the prompt; use MOSS-Image Base with a 8-step acceleration LoRA, euler / simple, LoRA strength 0.8-1.2, 8-16 steps, CFG 1-3

Risks and limitations

  • Mimics a specific artist's style, with no authorization stated by the author; the license does not allow derivatives, nor does it allow off-site rental or sale
  • 7 of the 13 sample images are hidden due to rating; there is no description of the training set or training parameters
  • This entry has not been tested

Trigger word: Zero_Q_0q

[A] Absolute CINEMA | MOSS-KREA2-CHROMA | Movie style · 🌱MOSS-Image

LoRA · Port · Author Kaalciv · Published 11-09 23:20 · Downloads 99

The MOSS version of a movie still style LoRA: pushes text-to-image generation toward the texture and composition of realistic cinematic visuals, avoiding the "polished" look of stock photos.

Clear purpose, the MOSS version has independent training instructions and 12 sample images, but no strength or sampling parameters are provided, rated A.

Images

Sample 1680x944

Reasons for recommendation

  • Port: The same page already has Krea 2 and Chroma versions, with a cumulative 3857 downloads on the model page; the MOSS version was trained on Civitai using the MOSS-Image base and 138 movie stills (up to 2 per film)
  • Serves realistic cinematic shots: no trigger word, use shot terms like Medium shot / Close-up shot / Wide angle / Overhead shot in the prompt to organize the scene
  • 12 sample images; this version has 99 downloads (representing attention only); the author reminds that the 8-step Turbo workflow requires a LoRA designed for Turbo

How to use

  • Load the LoRA, no trigger word; start the prompt with shot types (Medium shot of… / Close-up shot of…) to write cinematic scenes
  • The author states that when using 8-step Turbo, a Turbo LoRA must be paired; strength, steps, and CFG are not provided by the author

Risks and limitations

  • "Avoiding the polished stock photo look" and "looking more like good actors performing" are the author's descriptions of the Krea 2 version; the MOSS version has no separate effect description or comparison
  • Training images come from movie stills, with unclear copyright sources; the license does not allow commercial use of generated images, and attribution is required for reuse
  • 4 sample images are hidden due to rating; this entry has not been tested

[A] Regional Eyes: Umrbium · Moss v1.0

LoRA · Port · Author umbrium2835 · Published 11-09 23:00 · Downloads 38

The MOSS version of a concept LoRA that adds "extra eyes" to body parts: from a single eye on the forehead to clusters of eyes on the back, arms, and legs, used for supernatural characters and bizarre fantasy settings.

Downgraded from automatic S to A: the concept is new, and the trigger words and writing style are clear, but the MOSS version lacks its own parameters and has few samples.

Images

Sample 3328x4992

Reasons for recommendation

  • Port: The same page already has Krea 2, Anima, and Pony versions, with a cumulative 241 downloads on the model page
  • Trigger words are divided by body part, with no general trigger word: eyestyle forehead / face / neck / chest / back / arm / leg / hand / foot, specify where you want them
  • The author provided prompt writing tips: tags and natural language both work, expose the target body part and specify the location clearly, use only one body part at first, and close-ups and medium shots show the effect more easily
  • The license is permissive, allowing image generation, sale, and derivatives

How to use

  • Write body part phrases like eyestyle neck in the prompt, then use a sentence to specify exactly where the eyes should grow and which parts should remain unchanged; start strength at 1.0, use 0.75 for a lighter effect
  • If eyes appear elsewhere, change the seed, slightly lower the strength, or explicitly specify areas that should not change; MOSS steps and CFG are not provided by the author

Risks and limitations

  • The recommended parameters on the model card (Euler / Simple, 30 steps, CFG 4, 832 × 1248) are for the Anima base model; the MOSS version has no specific instructions
  • The author states that positioning is not guaranteed: eyes may spread to nearby skin, clothing, or objects, and high strength may alter pose and art style; the hand and foot parts are still being refined
  • Only 4 visible sample images, with another 2 hidden due to rating; this entry has not been tested

Trigger words: eyestyle hand, eyestyle foot, eyestyle arm, eyestyle back, eyestyle leg, eyestyle chest, eyestyle face, eyestyle forehead

[A] Style: Hanasaki Taru · MOSS-base v1.0

LoRA · New model · Author NerissG · Published 11-10 02:33 · Downloads 40

A LoRA for artist Hanasaki Taru's illustration style, with tags and author prompts pointing to a watercolor illustration feel with paper texture.

The description is short but useful: the style enhancement sentence and the two tips about "not pairing with DMD or er_sde" can be followed directly, but strength and steps are missing, rated A.

Images

Sample 1056x1424

Reasons for recommendation

  • New model, trained on MOSS-base; serves text-to-image generation for watercolor style illustrations
  • No training trigger word; the author claims adding Watercolor illustration with paper texture. to the prompt can strengthen the style
  • The author provided pitfall avoidance tips: DMD acceleration and the er_sde sampler will weaken the LoRA effect
  • 11 visible sample images; also includes a split-qkv version file; the license allows image generation, sale, and derivatives

How to use

  • Load the LoRA, add Watercolor illustration with paper texture. to the prompt to enhance the style; avoid DMD acceleration LoRAs and the er_sde sampler; strength, steps, and CFG are not provided by the author
  • Choose one of the two files; the difference between the split-qkv version and the standard version, and which loader to use for each, is not explained by the author

Risks and limitations

  • The model card has only three sentences, with no parameters or training information
  • Mimics a specific artist's style, with no authorization stated by the author; the license requires attribution
  • 1 sample images are hidden; this entry has not been tested

[A] MOSS WeVisOne image inpainting/edit workflow · v1.0

Workflow · New model · Author hRUSTIN474 · Published 11-10 00:59 · Downloads 25

A MOSS local repaint / edit ComfyUI workflow: uses the WeVisOne-VL vision-language model to find areas to modify based on text and generate masks, then repaints this area according to the edit prompt.

Maintains automatic A: the most practical entry in this section, text-based mask localization saves manual mask drawing, but there are too few samples, so it is recommended to test on your own images on a small scale first.

Images

Sample 1024x1024

Reasons for recommendation

  • New workflow (JSON, not new weights): chains "specify what to change with text → automatically generate mask → local repaint" into one step, saving manual mask drawing
  • WeVisOne can also perform segmentation, detection boxes, point localization, and background removal; also includes a WeVisOne-VL_example.json to separately demonstrate these vision tasks
  • Steps are clearly written: load image → tell WeVisOne what to find → write edit prompt → run; optionally connect LanPaint nodes; the author claims both MOSS and Krea 2 can be used

How to use

  • Import MOSS segmentation Edit, Optional Lan Paint.json; download the text_encoders for WeVisOne-VL from the Comfy-Org Hugging Face repository (currently only supports the 4B version), and install the ComfyUI-WeVisOne-VL custom node
  • After loading the image, write the object to find / mask at the WeVisOne node, and write what to change it to in the edit prompt; if LanPaint is installed, it can be used for repainting; steps, CFG, and repaint strength are not provided by the author

Risks and limitations

  • Only 1 visible sample images (another 2 are hidden), with no mask effect or before-and-after comparison of image edits
  • The author themselves states that the 4B model is small and not to expect miracles; requires additional download of the vision model and installation of custom nodes, with VRAM usage not specified
  • The license does not allow publishing derivative versions; this entry has not been tested

[A] xMyStyle · MOSS Image

LoRA · New model · Author randomimagepotato · Published 11-10 04:02 · Downloads 39

A modern interior style LoRA with dark tones and neon accents, mainly used for generating room / interior scene images.

Maintains automatic A: clear purpose and simple usage, but lacks parameters.

Images

Sample 1472x1920

Reasons for recommendation

  • New model: the author claims that writing the trigger word plus a room description (simple or complex) can generate this dark neon style, serving interior concept art and scene atmosphere images
  • Trigger word xMyStyle; 3 visible sample images, with another 2 hidden

How to use

  • Write xMyStyle + desired room description in the prompt; strength, steps, and sampler are not provided by the author

Risks and limitations

  • The model card has only two sentences, with no parameters; it is not stated whether it can be used for characters or outdoor scenes
  • The license does not allow derivatives or sales; this entry has not been tested

Trigger word: xMyStyle

Worth watching (10)

[B] GTA VI Artstyle (not gamestyle) | MOSS · v1.0

LoRA · New model · Author Fatbuns · Published 11-09 22:13 · Downloads 81

A LoRA mimicking the official promotional illustration style of GTA VI (not the game graphics), trained on currently released game art images.

Images

Sample 2520x1680

[B] Style > gpt-image-2 · MOSS v0.2

LoRA · Version update · Author AgileNumbers · Published 11-10 06:33 · Downloads 30

The MOSS v0.2 of an enhancement LoRA that pushes visuals toward the gpt-image-2 output style: mainly adds details, lighting, and texture, with the author positioning it as image quality / detail enhancement.

Images

Sample 1232x1696

[B] Classic Oil Painting - CE · V03a - MossImage Base/SFT

LoRA · Port · Author CreativeEdge · Published 11-10 01:20 · Downloads 35

The MOSS version (V03a) of a classic oil painting style LoRA: the author claims to pursue an oil painting effect with brushstroke texture, trained on MOSS Base, usable with both Base and SFT.

Images

Sample 768x1024

[B] Upturned Nose aka Pixie Nose - Concept - MOSS Image · MOSS Image

LoRA · New model · Author Kinky_Caterpillar · Published 11-09 23:19 · Downloads 103

A facial feature concept LoRA for realistic portraits: makes characters more consistently feature an upturned "pixie nose".

Images

Sample 2048x3072

【B】Bicep Heart Meme · v1.0

LoRA · New Model · Author cruffsant461 · Published 11-10 01:06 · Downloads 38

Pose meme LoRA: Makes characters perform the "bicep heart" gesture—one arm flexed to show off muscle, while the other hand forms half a heart shape beside it.

Image

Example 896x1184

【B】Patakha Koli Dress Moss Image by Sarcastic TOFU · Patakha_Koli_Moss_v1

LoRA · New Model · Author sarcastictofu · Published 11-10 03:22 · Downloads 30

Traditional Indian attire LoRA: Features a short Koli sari (a style where the fabric is passed between the legs and tucked in for ease of labor) paired with a tube-top style tied Choli blouse, intended for realistic female portraits.

Image

Illustration 1672x941

【B】Dishwasher1910 Style | MOSS Image · v1.0

LoRA · New Model · Author HissingCat · Published 11-10 03:17 · Downloads 59

An anime-style LoRA imitating the art style of artist dishwasher1910, trained on MOSS SFT.

Image

Illustration 1672x941

【B】Dwarfism (achondroplasia) · MOSS

LoRA · Ported · Author dow185 · Published 11-09 22:48 · Downloads 113

MOSS version of the LoRA for the concept of achondroplasia (dwarfism) body types, used to generate characters with this physique.

Image

Illustration 1086x1448

【B】MOSSRang2-lite · nvfp4

Base Model / Checkpoint · New Model · Author DjM_ · Published 11-10 01:01 · Downloads 0 · Paid

The nvfp4 quantized version of DjM_'s MOSS base model MOSSRang2-lite: The author describes it as a streamlined version of MOSS_STutbo_Rang2, featuring built-in Turbo acceleration.

Image

Example 768x1360

【B】Kigurumi Onesie: Puppy [Krea2/MOSS] · MOSS_v1.0

LoRA · Ported · Author AuthentiQ · Published 11-10 08:15 · Downloads 8

MOSS version of the puppy-style kigurumi onesie clothing LoRA, which dresses realistic characters in a puppy-themed onesie.

Image

Example 768x1368

Included Only (13)


Data: Civitai Public API (Base Model MOSS Image), inclusion window 11-09 20:31 → 11-10 09:01 (Beijing Time). Recommendations are reviewed by editors; download counts reflect popularity only.

Civitai

Explore the latest AI Radar updates