---
title: Krea-2-OpenCast-LoRA
canonical_url: "https://www.modelscope.ai/models/akatz-ai/Krea-2-OpenCast-LoRA"
md_url: "https://www.modelscope.ai/models/akatz-ai/Krea-2-OpenCast-LoRA.md"
repository: akatz-ai/Krea-2-OpenCast-LoRA
last_updated: 2026-10-30
license: other
pipeline_tag: text-to-image-synthesis
tasks:
- text-to-image-synthesis
base_model:
- krea/Krea-2-Raw
base_model_relation: adapter
parameters: 114.3M
tensor_type:
- BF16
library_name:
- lora
- safetensors
- pytorch
frameworks:
- pytorch
language:
- en
downloads: 17
stars: 1
tags:
- krea-2
- lora
- comfyui
- opencast
- character-design
- portraits
---
# Krea-2-OpenCast-LoRA
> Krea-2-OpenCast-LoRA - An open-source model by akatz-ai on ModelScope. Krea 2 OpenCast LoRA
akatz-ai/Krea-2-OpenCast-LoRA is a 114.3M-parameter text-to-image-synthesis model on ModelScope. licensed under other. derived from krea/Krea-2-Raw.
- **Repository**: akatz-ai/Krea-2-OpenCast-LoRA
- **License**: other
- **Tasks**: text-to-image-synthesis
- **Parameters**: 114.3M
- **Base model**: krea/Krea-2-Raw
- **Tags**: krea-2, lora, comfyui, opencast, character-design, portraits
- **Downloads**: 17
- **Stars**: 1
- **Last updated**: 2026-10-30
Source: https://www.modelscope.ai/models/akatz-ai/Krea-2-OpenCast-LoRA
---
# Krea 2 OpenCast LoRA
[](https://x.com/akatz_ai)
[](https://www.patreon.com/Akatz)
[](https://github.com/sponsors/akatz-ai)
[](https://ko-fi.com/akatz)
Fictional characters with exaggerated faces, sculptural hair, and lifelike skin and clothing. Created by Akatz Labs for OpenCast Studio.
[](assets/opencast-mosaic.png)
Selected outputs at LoRA strength 1.0. The grid reads left to right, top to bottom. Each portrait has an exact prompt, seed, full-resolution image and API workflow in [the example index](examples/README.md). Images are resized for display; the faces have not been retouched.
[LoRA weights](krea2_opencast_v1.safetensors) · [Prompt examples](examples/prompts.json) · [Canvas workflow](examples/Krea-2-OpenCast-Portrait.json) · [Portrait API workflow](workflows/portrait.api.json)
## The look
OpenCast combines photographic materials and studio lighting with deliberately unusual proportions: broad jaw shelves, split chins, long necks, small mouths, oversized hair and sharply defined silhouettes. It also works with prompts for dressed animal characters, as the fox, cat, frog and cockatoo examples show.
This is an image-generation adapter. OpenCast Studio can use the resulting portraits as character references for other workflows; the LoRA itself does not animate or swap a person in a video.
## Use with Krea 2 Turbo in ComfyUI
1. Download `krea2_opencast_v1.safetensors` into `ComfyUI/models/loras/`.
2. Install the base model, text encoder and VAE below. Use a ComfyUI version with native Krea 2 support.
3. Load Krea 2 Turbo, pass its model output through **Load LoRA (Model Only)**, select the OpenCast file, and set strength to **1.0**.
4. Start with one of the saved prompts and the settings below. The examples use ordinary descriptive text, with no special trigger token.
| Component | File | Folder under `ComfyUI/models/` |
| --- | --- | --- |
| Krea 2 Turbo | [krea2_turbo_fp8_scaled.safetensors](https://huggingface.co/Comfy-Org/Krea-2/resolve/ba0c2d43b14d8dff71aa40a992523b2f08425e22/diffusion_models/krea2_turbo_fp8_scaled.safetensors) | `diffusion_models/` |
| Qwen3-VL · image text encoder | [qwen3vl_4b_fp8_scaled.safetensors](https://huggingface.co/Comfy-Org/Krea-2/resolve/ba0c2d43b14d8dff71aa40a992523b2f08425e22/text_encoders/qwen3vl_4b_fp8_scaled.safetensors) | `text_encoders/` |
| Qwen Image VAE | [qwen_image_vae.safetensors](https://huggingface.co/Comfy-Org/Krea-2/resolve/ba0c2d43b14d8dff71aa40a992523b2f08425e22/vae/qwen_image_vae.safetensors) | `vae/` |
| Setting | Example value |
| --- | --- |
| LoRA strength | 1.0 |
| Image size | 1024 × 1536 |
| Steps | 8 |
| CFG | 1 |
| Sampler | `euler` |
| Scheduler | `simple` |
| Denoise | 1.0 |
| Negative conditioning | `ConditioningZeroOut` |
Start with the [OpenCast portrait canvas workflow](examples/Krea-2-OpenCast-Portrait.json): download it and drag the JSON into ComfyUI, or open it from the workflow menu. The five numbered groups follow the image from **User inputs → Models & OpenCast LoRA → Conditioning → Sampling & decoding → Final output**. Markdown notes include model downloads, folder destinations, size and seed guidance, and Akatz links.
The canvas opens with Ramon Volt's exact prompt, seed and portrait settings. Edit **Your character · prompt**, then Run. No custom-node packs are required; use an up-to-date ComfyUI with native Krea 2 support. The [API workflow](workflows/portrait.api.json) remains available for scripts and executors. Base models are separate downloads. Full-body outpainting is not needed for these portrait examples.
[](assets/opencast-portrait-workflow.png)
*Canvas overview. The output preview shows the retained Ramon example; it is not embedded in the JSON. The canvas was checked for import, export and reload in ComfyUI frontend 1.55.18 using retained node schemas. Its resolved generation graph matches the saved portrait recipe; this layout revision was not rerendered.*
## Prompt examples
Describe the geometry you want, then the hair, clothing, expression and light. For example, “jaw corners project beyond the cheekbones” is more specific than “an unusual face.” Keep one subject in the prompt and give prominent features enough space in the frame.
The following are the exact prompts used for selected images above. Seeds and all 18 prompts are also available in [prompts.json](examples/prompts.json).
### 01. Ramon Volt
Seed: `28040000` · [Image](examples/images/01-ramon.png) · [API graph](examples/workflows/01-ramon.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows one original fictional adult man around forty years old. The subject is Latino. warm tan skin with consistent natural coloration. A slender torso and narrow shoulders. a narrow forehead and slim upper face. Enormous jaw corners project far beyond the cheekbones, forming a wide shelf below the narrow upper face. Two enormous rounded chin lobes hang far below the mouth, separated by a deep vertical cleft. Clearly defined cheekbones and lean upper cheeks. A tiny mouth sits high above the chin. a curly mullet with short curls on top and long curls behind the shoulders, black hair. A thin pencil mustache. The subject wears a cobalt technical jacket with a sculpted collar, tailored to the chosen build. The face is turned slightly toward the camera, with a composed self-possessed expression. A playfully absurd sculptural character, with the specified extraordinary facial anatomy as the defining feature. Lifelike human skin and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage. The two chin lobes are absurdly oversized, like two smooth hanging pears beneath a tiny pursed mouth. His serious deadpan gaze makes the impossible proportions funny.
```
### 02. Nala Halo
Seed: `28040822` · [Image](examples/images/02-nala.png) · [API graph](examples/workflows/02-nala.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows one original fictional adult woman around forty years old. The subject is Black. deep brown skin with consistent natural coloration. A slender torso and narrow shoulders. Tall height. Exceptionally long arms and legs. a narrow forehead and slim upper face, an extraordinarily elongated neck, longer than the head, clearly exposed above the collar. A narrow jaw tapering sharply toward the chin. A long pointed chin with a narrow tip. Clearly defined cheekbones and lean upper cheeks. A tiny mouth sits high above the chin. an enormous oval sculptural afro, a continuous cloud of tightly coiled hair with a crisp rounded silhouette, black hair. The subject wears a chartreuse mohair cardigan with a sculptural neckline, tailored to the chosen build. Large brushed-silver sculptural earrings. The face is turned slightly toward the camera, with a composed self-possessed expression. A playfully absurd sculptural character, with the specified extraordinary facial anatomy as the defining feature. Lifelike human skin and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage. Deep brown skin, a narrow elongated angular face, a long thin exposed neck, and an enormous perfectly round black afro. Small sculptural earrings. Chartreuse cardigan with a neat round neckline.
```
### 09. Celeste Loop
Seed: `28041644` · [Image](examples/images/09-celeste.png) · [API graph](examples/workflows/09-celeste.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows one original fictional adult woman around forty years old. The subject is White. fair skin with consistent natural coloration. Broad shoulders and a substantial torso. Short height. a square head with straight sides. A square jaw with sharply defined corners. A short wide flat-bottomed chin with a central cleft. Clearly defined cheekbones and lean upper cheeks. a tall architectural updo, pink hair. The subject wears a chartreuse mohair cardigan with a sculptural neckline, tailored to the chosen build. Large brushed-silver sculptural earrings. The face is turned slightly toward the camera, with a composed self-possessed expression. A distinctive eccentric fashion character with a striking silhouette. Lifelike human skin and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage. A fair-skinned broad-faced woman with a glossy pink hairstyle formed into two enormous curled loops above her head. Brushed-silver earrings, chartreuse mohair cardigan with large balloon sleeves. Dry confident half-smile.
```
### 04. Asha Spire
Seed: `28040548` · [Image](examples/images/04-asha.png) · [API graph](examples/workflows/04-asha.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows one original fictional adult woman in their mid-twenties. The subject is South Asian. warm tan skin with consistent natural coloration. A slender torso and narrow shoulders. an extremely elongated head, much taller than it is wide, a very thin neck, much narrower than the jaw. A narrow jaw tapering sharply toward the chin. An elongated rectangular chin extends far below the lower lip, split by a deep central groove. Clearly defined cheekbones and lean upper cheeks. An exceptionally long nose projects prominently from the face. Prominent ears project sideways from the head. A tiny mouth sits high above the chin. a wide flared bob with a short straight fringe, red hair. The subject wears a tobacco herringbone waistcoat over a cream shirt, tailored to the chosen build. A patterned silk neck scarf with a vivid ochre accent. The face is turned slightly toward the camera, with a composed self-possessed expression. A playfully absurd sculptural character, with the specified extraordinary facial anatomy as the defining feature. Lifelike human skin and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage.
```
### 17. Moss Puddle
Seed: `28043425` · [Image](examples/images/17-frog.png) · [API graph](examples/workflows/17-frog.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows a fictional anthropomorphic frog fashion character, with a clearly recognizable frog head and species-specific face, an upright clothed body and visible natural animal surface texture. The character is an adult around forty years old. Male gender presentation. Broad shoulders and a substantial torso. Short height. A compact torso with short substantial arms and legs. The subject wears a tobacco herringbone waistcoat over a cream shirt, tailored to the chosen build. Distinctive round translucent glasses. The face is turned slightly toward the camera, with a composed self-possessed expression. An individual character with a calm deadpan expression. Lifelike animal surface texture and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage. A green frog with enormous wide-set bulging amber eyes, a broad horizontal mouth and mottled olive skin. Tobacco brown herringbone waistcoat and cream shirt. The frog face remains amphibian, with the same dark sculpted bob-like fringe seen in the original.
```
### 12. Cleo Crumb
Seed: `28043562` · [Image](examples/images/12-cat.png) · [API graph](examples/workflows/12-cat.api.json)
Show exact prompt
```text
Tightly cropped front-facing studio headshot. The complete hairstyle and face occupy nearly all of the image. Only the tops of the shoulders and the garment neckline appear along the lower edge. The photograph shows a fictional anthropomorphic cat fashion character, with a clearly recognizable cat head and species-specific face, an upright clothed body and visible natural animal surface texture. The character is an adult around forty years old. Female gender presentation. A slender torso and narrow shoulders. The subject wears a chartreuse mohair cardigan with a sculptural neckline, tailored to the chosen build. Large brushed-silver sculptural earrings. The face is turned slightly toward the camera, with a composed self-possessed expression. An individual character with a calm deadpan expression. Lifelike animal surface texture and real clothing. Plain white seamless background, soft broad frontal light, gentle contact shadows, individual hair strands, visible pores and detailed fabric. Quiet neutral color. One character, without lettering, logos, borders or collage. A hairless sphynx cat face with gigantic bat-like ears, a long wrinkled narrow muzzle and a judgmental squint. A huge soft chartreuse mohair collar surrounds the tiny pointed head. Keep both ear tips visible.
```
## Notes
These are selected examples, not a success-rate benchmark. The amount of exaggeration varies with the prompt and seed. Hair, ears or accessories can touch the frame, and demanding anatomy can produce artifacts. Matching settings help reproduce a look, but changes in ComfyUI, model precision or kernels can change the exact pixels.
## File identity
- File: `krea2_opencast_v1.safetensors`
- Size: 228,587,488 bytes
- SHA256: `98f8b6d41969b312037cc09090e7283051cdf7686963346e04eb985d4bda84f6`
## License
Use and redistribution are subject to the [Krea 2 Community License Agreement](LICENSE.pdf), including its commercial-use conditions. Downloading or using this adapter requires accepting those terms. See [Notice](Notice) for attribution. Krea 2 OpenCast LoRA is an independent Akatz Labs derivative, not an official Krea product or endorsement.
## 💜 Support Akatz Labs
I build and release AI tools, models, LoRAs, and workflows so others can experiment, create, and build upon them.
If this model has been useful to you, consider helping fund the next experiment!
Your support goes toward GPU compute, software development, and future open-source releases.
**Support the lab:** [GitHub Sponsors](https://github.com/sponsors/akatz-ai) · [Patreon](https://www.patreon.com/Akatz) · [Ko-fi](https://ko-fi.com/akatz)
Thanks for helping keep independent creative AI development moving! 💜
[Explore Akatz Labs →](https://akatz.ai)