Neta Lumina is Neta.art Lab's anime-illustration model, a fine-tune of Alpha-VLLM's open Lumina-Image-2.0 with a Gemma 2B text encoder. The publisher lists illustration, posters, storyboards and character design as its strengths.

Should I write tags or sentences?

Either, or both in one prompt. Tags worked for the pianist and three plain sentences worked for the neon-window portrait. Use tags for the look and a sentence for camera position, lighting direction and how two characters relate.

Do artist and character tags work?

Published guidance for the model describes artist tags written as @artistname:1.0 and character tags written as character:#name. We did not test either on this endpoint, so check a few seeds before you build a pipeline on them.

Does the fixed prefix help?

The model's prompt book suggests an optional prefix sentence, “You are an assistant designed to generate anime images based on textual prompts.” On the same seed, images with and without it were nearly identical in our test, so you can skip it.

Can I prompt in other languages?

The publisher recommends Chinese, English and Japanese prompts first, and describes native multilingual support. All of our tests were in English, so test your own language on a few seeds.

How fast is it, and how is it billed?

Across 31 runs model time was 13 to 75 seconds, median 40, and none finished in under ten. Billing is per image at one flat price, whatever the ratio or the format. Current prices are on the model page.

Neta Lumina API

Q: Is it really uncensored?

Nothing is filtered or reviewed on our side, and our strongest content probe rendered as written on the first try. Rights and consent for everything you generate stay with you.

neta-art/neta-lumina/text-to-image

Neta Lumina text-to-image draws one anime illustration from Danbooru-style tags, plain sentences or both, on Neta.art Lab's fine-tune of Lumina-Image-2.0. Nine aspect ratios at about one megapixel, JPEG, PNG or WebP output, and no content filter.

From $0.012/image

Run model View API

Input

Text to image

Prompt494/5000

Aspect ratio

Output format

JPEGPNGWebP

Seed

Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run

Example result

Animate

Edit

Split into layers

{
  "model": "neta-art/neta-lumina/text-to-image",
  "input": {
    "aspect_ratio": "3:4",
    "output_format": "jpeg",
    "seed": 203,
    "prompt": "masterpiece, best quality, anime illustration, 1girl, solo, mature female, adult woman in her early thirties, tall, long straight black hair with a silver streak, violet eyes, black velvet off-shoulder gown, high-collared cape, long black gloves, standing between towering gothic bookshelves, floating open grimoires, glowing violet runes lighting her face from below, shafts of cold moonlight from a high oculus, dust motes, low angle, looking at viewer, faint smile, indigo and violet palette"
  }
}

Introduction

Neta Lumina Text to image draws one anime illustration from a written prompt. Neta Lumina is Neta.art Lab's fine-tune of Alpha-VLLM's open Lumina-Image-2.0, trained on a large set of anime images and multilingual tag data, with Google's Gemma 2B as its text encoder. Prompt it with Danbooru-style tags, plain sentences or both, pick one of nine aspect ratios and get JPEG, PNG or WebP back. The platform adds no content filter to the prompt or the image.

Key features & highlights

Sentences, not just tags

Because the text encoder is a language model, the prompt can be a description: who is in the frame, what they wear, where the camera is and how the scene is lit. Tags still work and can be mixed into the same prompt.

Poses and angles with energy

Foreshortening, dynamic angles and motion lines are part of its vocabulary. A flying kick aimed at the viewer came out as a dramatic vertical frame in our tests.

Nine shapes, one price

Every ratio from 21:9 to 9:21 is about one megapixel: 1024 x 1024, 960 x 1280, 1536 x 864. The shape and the output format never change the price.

Uncensored

Our strongest content probe came back as written on the first try. Nothing is flagged, softened or reviewed on our side.

Request parameters

Request parameters for Neta Lumina · Text to Image
Parameter	Required	Description
`prompt`	Yes	Tags, sentences or both, up to 5,000 characters. Open with "masterpiece, best quality", put the head count ("1girl", "1boy") next, then outfit, pose, camera and light. at least 1 characters · at most 5000 characters
`aspect_ratio`	No	Nine options from 21:9 to 9:21; 1:1 by default. 1:1 is 1024 x 1024, 3:4 is 960 x 1280 and 16:9 is 1536 x 864. `1:1` · `16:9` · `9:16` · `4:3` · `3:4` · `3:2` · `2:3` · `21:9` · `9:21` · Default: `1:1`
`output_format`	No	JPEG by default, PNG or WebP if you need them. Does not change the price. `jpeg` · `png` · `webp` · Default: `jpeg`
`seed`	No	Keeps a set of images comparable while you change one tag at a time. It does not guarantee the exact same pixels on every run. `≥ 0` · `≤ 2147483647`

Pricing

$0.012 / image

Billed per image at one flat price for every aspect ratio and output format.

Use cases

Character art and key visuals

Portraits, full-body art and posters from a list of tags or a short description. Fix the seed and change one detail at a time to explore outfits and lighting for the same character.

Action scenes

Fights, jumps and chase shots with a strong camera angle, in tall frames for phone screens or wide ones for banners.

Uncensored art for your platform

There is no content flag to set and no second review of the result, so what your users write is what they get back.

Explore key use-case tutorials

Prompt guide & pro tips

Lead with quality and head count

Start with "masterpiece, best quality", then "1girl", "1boy" or both, then the character, outfit, pose, camera, light and atmosphere. Lowercase tags with spaces read best.

Write a real sentence for the hard parts

Camera position, lighting direction and what two characters are doing to each other are easier to say in a sentence than in tags.

Budget time, not money

Each image takes about 40 seconds of model time. Run variations in parallel and keep the seed fixed while you compare them.

Reroll the odd corner

The model occasionally adds a signature-like scribble or a pixelated patch learned from its training art. A new seed is the fix; there is no negative prompt.

API quick start

# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $SPICY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
  -d '{
  "model": "neta-art/neta-lumina/text-to-image",
  "input": {
    "prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette",
    "aspect_ratio": "1:1",
    "output_format": "jpeg"
  }
}'

# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $SPICY_API_KEY"

See the API reference for authentication, the complete schema, polling, and error handling.

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

Anime key art from Danbooru tags or plain sentences, with real action poses

Neta Lumina is Neta.art Lab's anime-illustration fine-tune of Alpha-VLLM's Lumina-Image-2.0, with a Gemma 2B text encoder. It reads Danbooru-style tags, plain sentences or both and draws nine aspect ratios at about one megapixel. Across 31 runs an image took 13 to 75 seconds of model time, a median of 40 and none under ten, so plan for batches rather than live typing. Nothing filters the prompt or the result: our strongest content probe rendered as written on the first try.

Open the playground

Anime key art from Danbooru tags or plain sentences, with real action poses

Model: Neta Lumina by Neta.art Lab, a fine-tune of Lumina-Image-2.0 with a Gemma 2B text encoder
Speed: 13 to 75 seconds of model time per image in our 31 runs, median 40; none finished in under 10
Sizes: 9 aspect ratios at about one megapixel: 1024 x 1024 at 1:1, 960 x 1280 at 3:4, 1536 x 864 at 16:9, 1536 x 656 at 21:9
Prompt: Up to 5,000 characters of tags, sentences or both; no negative prompt
Output: One image per request as JPEG (default), PNG or WebP, with an optional seed
Price: Billed per image at one flat price for every ratio and format
Content filter: None; our strongest content probe rendered as written on the first try

Prompting: Tags and sentences both workQuality tags, a head count and outfit tags drew the candlelit pianist, and three plain sentences drew a man at a rain-streaked window with the scar, the shirt and the neon reflections all present. Mix them: tags for the look, a sentence for camera and lighting direction.
Speed: Budget time, not moneyImages took 13 to 75 seconds, median 40, and picture size does not explain it: a 656 x 1536 run took 13 seconds while 960 x 1280 runs took 14 to 75. Run variations in parallel and keep the seed fixed while you compare them.
Limits: What it will not doThere is no negative prompt, image input, LoRA slot, steps or guidance setting, and one image per request. Action shots are the least reliable: a second seed of our flying-kick prompt tangled a leg. The model also occasionally adds a signature-like scribble or a pixelated patch learned from its training art, and a new seed is the fix.

Getting the most from Neta Lumina

Lead with quality and head count

Open with “masterpiece, best quality, anime illustration”, then “1girl”, “1boy” or both, then the character, outfit, pose, camera, light and atmosphere. Lowercase tags with spaces read best. The pianist in the first picture used exactly this order.

Write a sentence for the hard parts

Camera position, lighting direction and how two characters relate are easier to say in a sentence than in tags. The neon-window man in the second picture came from three plain sentences, and the scar, the unbuttoned shirt and the cyan and magenta reflections all made it into the picture.

Give each character their own phrase
For two characters, put the head-count tags first and then give each one an outfit in a phrase of their own. In the beach pair, her white sundress and his open linen shirt stayed with the right person in our runs.
Ask for foreshortening and motion
“Foreshortening, dynamic angle, motion lines” in a tall 9:16 frame produced a flying kick aimed at the viewer. Action is the least reliable corner, so expect to reroll: our second seed of the same prompt tangled a leg.
Choose the frame
Nine ratios, all around one megapixel: 1:1 is 1024 x 1024, 3:4 is 960 x 1280, 3:2 is 1344 x 896, 16:9 is 1536 x 864 and 21:9 is 1536 x 656, with the tall ratios mirrored. The ratio and the output format never change the price.

Who uses Neta Lumina

Key-visual and poster artists
A moonlit key visual from one prompt
A black-haired adult woman in a flowing crimson coat on a moonlit rooftop of an old European city, a huge full moon behind her, rose petals in the wind, wide shot. Tags set the look and the lighting; the same prompt on other seeds gave other compositions of the same idea.
Comic and storyboard artists
Tall action frames for phones
Two seeds of a rooftop flying kick in 9:16, and the neon-window portrait at 2:3. Tall frames suit phone screens and comic panels; reroll the seed when a limb tangles.
Character designers
One character, many takes
Hold the tags and change only the seed to see the same pianist in different poses. Then change one tag at a time to explore outfits and lighting for the same character.
Banner and wide-format designers
Wide scenes for banners
A couple dancing barefoot at the edge of the sea at sunset in 16:9, with a glittering sky and waves. For banners go wider with 21:9, which is 1536 x 656.

From first tag to finished art with Neta Lumina

Start from the standard opening
“masterpiece, best quality, anime illustration”, then the head count, then the character.
Add outfit, pose, camera and light
Tags for the look; one sentence for camera position and lighting direction.
Pick the ratio and run once
Choose the frame for where the picture will live and fix the seed.
Change one tag at a time
Keep the seed and swap a single detail to explore outfits and light.
Reroll the odd corner
A stray scribble, a pixelated patch or a tangled limb is cured by a new seed, since there is no negative prompt.

Neta Lumina prompts worth copying

Candlelit pianist

Pure tags: quality, head count, character, gown, room and light. Seed 5 at 3:4.

masterpiece, best quality, anime illustration, 1girl, solo, mature female, adult woman in her late twenties, long wavy crimson hair, golden eyes, black off-shoulder evening gown with a high leg slit, sitting on the lid of a grand piano, legs crossed, one hand resting on the keys, candlelit ballroom, crystal chandelier, warm rim light, looking at viewer, slight smile

Neon window in sentences

No tags at all. Seed 2 at 2:3.

An anime illustration of a handsome adult man in his early thirties with messy silver hair and a small scar through one eyebrow. He wears an unbuttoned black dress shirt and leans against a rain-streaked window in a neon-lit city at night, glancing over his shoulder at the viewer. Cyan and magenta reflections slide across his face.

Sunset dance for two

Head-count tags first, then each outfit in its own phrase. Seed 2 at 16:9.

masterpiece, best quality, anime illustration, 1girl, 1boy, hetero, mature female, mature male, tall adult couple in their thirties, she has long wavy black hair and wears a white sundress with thin straps, he has short dark hair and stubble and wears an open linen shirt, barefoot, dancing at the edge of the sea at sunset, she laughs with her head tilted back, he spins her by the hand, golden sky, glittering waves, wide shot

Workbench

Input

Output

Introduction

Key features & highlights

Sentences, not just tags

Poses and angles with energy

Nine shapes, one price

Uncensored

Request parameters

Pricing

Use cases

Character art and key visuals

Action scenes

Uncensored art for your platform

Prompt guide & pro tips

Lead with quality and head count

Write a real sentence for the hard parts

Budget time, not money

Reroll the odd corner

API quick start

Related models and next tasks

Anime key art from Danbooru tags or plain sentences, with real action poses

Getting the most from Neta Lumina

Lead with quality and head count

Write a sentence for the hard parts

Give each character their own phrase

Ask for foreshortening and motion

Choose the frame

Who uses Neta Lumina

A moonlit key visual from one prompt

Tall action frames for phones

One character, many takes

Wide scenes for banners

From first tag to finished art with Neta Lumina

Start from the standard opening

Add outfit, pose, camera and light

Pick the ratio and run once

Change one tag at a time

Reroll the odd corner

Neta Lumina prompts worth copying

Candlelit pianist

Neon window in sentences

Sunset dance for two

More Neta Lumina questions

What is Neta Lumina?

Should I write tags or sentences?

Is it really uncensored?

Do artist and character tags work?

Does the fixed prefix help?

Can I prompt in other languages?

How fast is it, and how is it billed?