Skip to content
Kenerate IconKenerate Icon
Kenerate AI
Loading your account

Neta.art Lab

Neta Lumina — Neta.art's open anime model, online

Neta Lumina is an open anime image model from Neta.art Lab, built on Lumina-Image-2.0. On Kenerate it runs in the browser from a plain sentence, a tag list or a mix of both.

Base model
Lumina-Image-2.0 by Alpha-VLLM — a 2-billion-parameter flow-based diffusion transformer
Training data
More than 13 million anime-style images, plus hundreds of thousands of hand-picked images for aesthetics
Prompting
Natural language (Chinese, English and Japanese prioritised) with Danbooru tag support
Weights
Open (Apache 2.0)
On Kenerate
Neta Lumina — text to image, 7 SDXL sizes from 1024×1024 to 1344×768, seed control
Try
  • Neta Lumina
  • Shrine in autumn
  • Pilot and propeller plane
  • Chef cooking ramen
  • Library with a sleeping cat
  • Real outputs
  • Official sources
  • Free to start
  • No watermark

Updated by the

Made with Kenerate

What is Neta Lumina?

Updated

What is Neta Lumina?

Neta Lumina is an open anime image model from Neta.art Lab, fine-tuned from Alpha-VLLM's Lumina-Image-2.0 on more than 13 million anime-style images and released under Apache 2.0. It reads natural language and Danbooru-style tags. On Kenerate AI it runs in the browser at seven SDXL-style sizes.

Key facts

Maker
Neta.art Lab
Based on
Lumina-Image-2.0 (Alpha-VLLM)
Base size
2B-parameter diffusion transformer
Text encoder
Gemma 2B
Training set
13M+ anime-style images
Weights
Open · Apache 2.0
Prompts
Sentences or Danbooru tags
On Kenerate
Text to image · 7 sizes

Models

Neta Lumina on Kenerate

Every version in the generator, and what each is good at.

  • Neta Lumina title card — a clean

    Neta.art Lab

    Neta Lumina

    Output
    7 SDXL sizes

    Best for

    Anime characters, scenery and manga

    Open Neta Lumina

From our tests

Strengths and limits

Good at

  • Rich anime backgrounds — the autumn shrine and the library with a sleeping cat are typical.
  • Follows plain sentences as well as tags: the chef and pilot prompts were ordinary sentences.
  • Can switch to black-and-white manga with screentone.
  • One of the lower-cost models on Kenerate.

Limits

  • “Manga panel” gave a two-panel page with a speech bubble — say “single panel” if you want one image.
  • Simple character prompts can come out flat (see the rooftop); add light, setting and detail.
  • SDXL sizes only, about 1 megapixel, and text to image only, with no negative-prompt field.
  • Not on the free plan — any credit pack unlocks it.

How to

How to use Neta Lumina on Kenerate

  1. Step 1

    Open Neta Lumina

    Pick it in /app/image. Any credit pack unlocks it.

  2. Step 2

    Describe the scene

    A sentence, a tag list or both — e.g. “anime illustration of a chef cooking ramen, steam, warm light”.

  3. Step 3

    Pick a size

    Square, portrait 2:3 or 3:4, landscape 3:2 or 4:3, 9:16 or 16:9.

  4. Step 4

    Download

    No watermark. Reuse the seed to make close variations.

Kenerate AI Image Studio with a text prompt typed into the prompt box

Open Image Studio and type what you want to see — subject, style, lighting and mood.

1 / 5 · Describe your image

Watch the tour · 1:23

Prompts, styles, references, edit, Arena & chat

Chapters

FAQ

Neta Lumina questions

What is Neta Lumina?

Neta Lumina is an open anime image model from Neta.art Lab. It is fine-tuned from Lumina-Image-2.0, Alpha-VLLM's 2-billion-parameter flow-based diffusion transformer, on more than 13 million anime-style images, with a Gemma 2B text encoder. Its weights are released under Apache 2.0, and on Kenerate it runs in the browser.

Link to this answer

Who makes Neta Lumina — what is Neta Art?

Neta Lumina comes from Neta.art Lab, which publishes the model on Hugging Face as neta-art/Neta-Lumina under the Apache 2.0 licence. The base model, Lumina-Image-2.0, is from Alpha-VLLM. Neta.art fine-tuned it on anime images and hand-picked aesthetic data to make it an anime specialist.

Link to this answer

How do I prompt Neta Lumina?

Neta Lumina reads natural language, with Chinese, English and Japanese prioritised, and also understands Danbooru-style tags, so you can write a sentence, a tag list or both. Name the subject, setting, light and style. Kenerate has no negative-prompt field for it, so describe what you want, and say “single panel” for one manga image.

Link to this answer

Can I use Neta Lumina online?

Yes. On Kenerate, Neta Lumina runs in the browser: open /app/image, pick Neta Lumina, write a prompt and choose one of seven sizes, from 1024×1024 square to 1344×768 widescreen. There is no download, no GPU and no watermark. It is text to image only, so you cannot upload a reference image.

Link to this answer

Is Neta Lumina free on Kenerate?

Neta Lumina is not on the Kenerate free plan, but it is one of the lower-cost models. Generations are credit-based, the cost of each image is shown before you generate, and credits come in one-time packs that never expire, with no subscription; see /pricing. Any credit pack unlocks it.

Link to this answer

Neta Lumina vs Pony XL — which should I use?

Pick Neta Lumina if you like writing sentences, want detailed anime scenery or want a newer Lumina-based model. Pick Pony XL if you prefer score tags and Danbooru-style tag lists, or want a western cartoon look. Both run on Kenerate at the same seven sizes, so you can test one prompt on each.

Link to this answer

In depth

Neta Lumina, in depth

Background, how to prompt it, and when to pick something else.

What makes Neta Lumina different

Many popular anime models, such as Pony, are SDXL fine-tunes. Neta Lumina starts instead from Lumina-Image-2.0, a 2-billion-parameter flow-based diffusion transformer [2], and Neta.art Lab fine-tuned it on more than 13 million anime-style images [1].

It uses a Gemma 2B text encoder, and its maker says it follows natural language, with Chinese, English and Japanese prioritised, while keeping Danbooru tag support [1]. Its maker also lists furry, Guofeng (traditional Chinese style) and pets among its strengths.

By Last updated and reviewed

We run these models in the Kenerate generator. Model facts come from the makers' own pages; specs and limits were checked against our code; the examples are our own outputs.

Sources: Neta.art Lab — Neta Lumina (Hugging Face) · Alpha-VLLM — Lumina-Image-2.0 (Hugging Face)

Try Neta Lumina

Free to start. No watermark.

Tried it? Tell us how it went