Choose a base model
Krea 2, FLUX, Qwen Image and Z-Image suit different looks. Your LoRA is fine-tuned on the one you pick.
Skip to content
Prepare a dataset
LoRA fine-tuning for image and video models
LoRAs, fine-tuned on your images.
Loras AI: fine-tuned models from your own images. Train a LoRA, keep every image consistent.
Loras AI turns 4–30 of your images into a LoRA: a lightweight fine-tuned model that keeps your character, product or style consistent in every new image. No GPU, no code.
Try:
Quick answer: Train supported image and video LoRAs with configured services, sufficient credits and a compatible dataset. Each LoRA must use its paired generator; catalog listings do not establish compatibility.
Start from an example
Fine-tune and generate with
- Krea 2
- FLUX.1 Dev
- FLUX.2 Dev
- Qwen Image 2512
- Z-Image Turbo
- Wan 2.2
- LTX-2 19B
- MOSS
- Anima
Everything you need to fine-tune
4–30 images
A small, focused dataset is enough to train a LoRA.
Characters, products, styles
Presets for the three things people fine-tune most.
Image and video models
FLUX, Krea 2, Qwen Image and Z-Image, plus Wan and LTX video families.
Caption editor
Review every caption and the trigger word before training.
My LoRAs
Follow training progress and reuse finished models.
Pay per task
Credit costs are shown before you submit anything.
No GPU needed
Training and generation run in the cloud.
Four languages
English, 中文, 日本語 and 한국어.
Made with Loras AI
Images and videos generated with Loras AI Image and Loras AI Video. Click one to use its prompt.
Fine-tuned models for characters, styles and products
Explore examplesThese original images were AI-generated as concept references. They are not Loras AI training outputs, before/after comparisons or evidence of model performance.
Built for creators who need consistency
- Comics and storybooksKeep the same face and outfit across every panel and page.Try it
- Online storesTrain on your product once, then place it in new scenes without a studio shoot.Try it
- Brands and illustratorsTurn an art style or brand look into a model your whole team can reuse.Try it
- Video LoRAsPlan motion datasets for Wan and LTX video models.Try it
- AI image generatorCreate and edit images from a prompt, no LoRA required.Try it
- AI video generatorTurn text or an image into a short video clip.Try it
How fine-tuning works on Loras AI
Three steps from your images to a fine-tuned model you can reuse.
- 01
Upload
Add 4–30 clear images of one character, product or style. Review each image and caption before you train.
- 02
Fine-tune
Pick a base model and start fine-tuning. Your LoRA trains in the cloud; follow its progress in My LoRAs.
- 03
Create
Generate new scenes with your fine-tuned model while your subject stays the same.
Why fine-tune with Loras AI
Everything you need to fine-tune a model, without setting up hardware or writing code.
| Feature | Loras AI | Train locally | Developer APIs |
|---|---|---|---|
| Needs your own GPU | No | Yes, high VRAM | No |
| Needs coding | No | Environment setup | Yes |
| Dataset and caption tools | Built in | Separate tools | Build your own |
| Generate after training | Same workspace | Separate setup | Build your own |
| How you pay | Credits per task | Hardware and power | Per-request billing |
From dataset preparation to LoRA training and generation
Krea 2 connects cloud training to image generation. MOSS and Anima offer separate generation workflows, alongside video model exploration and draft tools. Training and generation require sign-in and sufficient credits.
Train a custom Krea 2 LoRAInspect images, edit captions and submit training. Follow task progress in My LoRAs, then use the trained weights to generate images.Explore video models and workflowsCompare Wan and LTX families and plan motion datasets. The video workspace organizes references and start/end-frame drafts; cloud video training and generation are not available.
Each model needs compatible LoRA weights; adapters are not interchangeable. Resolution options and credit costs appear in the generation workspace.
Training capabilities and requirements
Costs and limits below follow the application configuration. Integration describes the code, not verified provider uptime or a successful live training run.
| Model | Integration | Cloud input limits | Credits and settings | Output and compatibility |
|---|---|---|---|---|
| Krea 2 | Integrated; configuration required | 1–30 JPEG, PNG or WebP images with captions and a trigger phrase. Up to 15 MiB per image; 25 MiB total. | Training: 100 credits / task. Generation: 10 credits / image. Fixed training steps: 100. | LoRA weights after success; use the paired Krea 2 generator. Compatibility with other base models is not implied. |
| FLUX.1 Dev | Integrated; configuration required | 4–30 JPEG, PNG or WebP images with captions and a trigger phrase. Up to 15 MiB per image; 25 MiB total. | Training: 167 credits / task. Generation: 10 credits / image. Fixed training steps: 500. | LoRA weights after success; use the paired FLUX.1 Dev generator. Compatibility with other base models is not implied. |
| FLUX.2 Dev | Integrated; configuration required | 10–30 JPEG, PNG or WebP images with captions and a trigger phrase. Up to 15 MiB per image; 25 MiB total. | Training: 2134 credits / task. Generation: 14 credits / image. Fixed training steps: 1000. | LoRA weights after success; use the paired FLUX.2 Dev generator. Compatibility with other base models is not implied. |
| Qwen Image 2512 | Integrated; configuration required | 4–30 JPEG, PNG or WebP images with captions and a trigger phrase. Up to 15 MiB per image; 25 MiB total. | Training: 167 credits / task. Generation: 10 credits / image. Fixed training steps: 500. | LoRA weights after success; use the paired Qwen Image 2512 generator. Compatibility with other base models is not implied. |
| Z-Image Turbo | Integrated; configuration required | 4–30 JPEG, PNG or WebP images with captions and a trigger phrase. Up to 15 MiB per image; 25 MiB total. | Training: 209 credits / task. Generation: 10 credits / image. Fixed training steps: 500. | LoRA weights after success; use the paired Z-Image Turbo generator. Compatibility with other base models is not implied. |
| Wan 2.2 | Integrated; configuration required | 3–10 MP4 clips with a trigger phrase. Up to 8 MiB per clip; 24 MiB total. | Training: 1667 credits / task. Generation: 67 credits / 5-second video. Fixed training steps: 100. | LoRA weights after success; use the paired Wan 2.2 generator. Compatibility with other base models is not implied. |
| LTX-2 19B | Integrated; configuration required | 3–10 MP4 clips with a trigger phrase. Up to 8 MiB per clip; 24 MiB total. | Training: 584 credits / task. Generation: 34 credits / 5-second video. Fixed training steps: 500. | LoRA weights after success; use the paired LTX-2 19B generator. Compatibility with other base models is not implied. |
Cloud tasks require sign-in and sufficient credits. Krea 2 needs Fal; WaveSpeed image and video trainers need WaveSpeed access and R2 storage.
Cloud training sends your selected media and captions to the provider. Image and video dataset limits differ. Confirm account readiness before submitting.
A focused workflow
From a few images to a fine-tuned model
A LoRA is a lightweight fine-tuned model: instead of retraining a whole base model, it learns one character, style or product from a small set of images. Loras AI handles the dataset, captions and training, so a few images become a fine-tuned model you can reuse.
Prepare 4–30 images
Check dimensions and repeats, and keep useful variety in angle, lighting and background.
Review the captions
Describe what changes between images so the fine-tuned model learns only your subject. Check the trigger word before training.
Generate with your LoRA
When training finishes, open it from My LoRAs and generate new scenes with your fine-tuned model.
Ideas worth teaching
What fine-tuned LoRAs can do: characters, styles and products
Explore three ways to define a training concept. Each reference links to a dedicated workflow, with dataset advice and evaluation prompts.
A style that travels.
Use this coastal illustration as a visual-language reference. A style dataset should vary the subject while keeping its materials, colors and mark-making coherent.
Caption exampleA coastal village in ohwx_style, gouache texture, warm flat colors and fine ink contours.
Explore style LoRA trainingOne character. New settings.
An original ceramic explorer gives a dataset a clear identity. Vary the angle, pose and surroundings instead of repeating one composition.
Caption exampleohwx_robot, a ceramic explorer with blue glass eyes, standing beside an alpine lake in daylight.
Explore character LoRA trainingKeep the product recognizable.
A blank amber bottle makes shape and materials easy to discuss. Include alternate views of its glass body and cap rather than teaching a single background.
Caption exampleohwx_bottle, an amber glass bottle with a round walnut cap, on a travertine surface in window light.
Explore product LoRA trainingThese original images were AI-generated as concept references. They are not Loras AI training outputs, before/after comparisons or evidence of model performance.
From intent to evaluation
How to fine-tune a model with LoRA
Prepare the image or video dataset required by the selected model, review its trigger and captions, then submit training. Use the completed weights with their matching generator.
- 1
Choose one concept
Decide whether you want a recognizable character, a transferable style, a distinctive product or a specific motion.
Read the note - 2
Collect useful examples
Keep the intended concept consistent while varying details you do not want the model to memorize.
Read the note - 3
Review captions & export
Inspect file dimensions, edit each caption and export the local configuration before leaving the workspace.
Read the note - 4
Plan your test prompts
Test completed, compatible LoRA weights with new prompts. Record the model, settings, successes and failures, and compare with the base model to evaluate the concept.
Read the note
Dataset & compatibility
Images that make a good fine-tuned model
A LoRA is fine-tuned on one base model family and only works there. A FLUX LoRA will not load in Krea 2, Qwen Image or Wan, so choose the base model before you prepare images.
Read the dataset preparation note- Fine-tune one subject per LoRA.
- Use images you have the rights to train on.
- Remove corrupted files and near-duplicates.
- Test the fine-tuned model on new scenes.
Prompt notebook
Prompts to test your fine-tuned model
Save these ideas for evaluating a compatible trained adapter later. They are prompt examples, not generated results or a promise of consistent output.
Architecture
A hillside library in ohwx_style, warm flat colors, textured paper and delicate ink contours.
An everyday object
A vintage desk lamp in ohwx_style, simple composition, warm ochre and indigo palette.
A natural scene
A quiet forest path after rain in ohwx_style, layered shapes, visible gouache texture.
A different subject
A small sailboat in ohwx_style, bold simplified forms, soft paper grain, no lettering.
A character study
An original fictional gardener in ohwx_style, full-body pose, simple pale background.
A new palette test
A night market in ohwx_style, restrained blue and amber palette, hand-painted details.
Related notes
A little preparation goes a long way.
Practical reading on image selection, captions, training settings and evaluating what a model actually learned.
How many images do you need?Choose useful variation rather than relying on one universal image count.A practical LoRA captioning guideSeparate the concept from changing details, with character, style and product examples.Training steps without the guessworkUnderstand steps, learning rate and overfitting before changing advanced settings.How to evaluate a trained LoRATest new prompts, compare a baseline and document useful failures.
FAQ
LoRAs and fine-tuned models: FAQ
Model-specific guidance, with the current availability stated clearly.
What are LoRAs, and how do they relate to fine-tuned models?
A LoRA is a type of fine-tuned model. Full fine-tuning retrains an entire base model; a LoRA trains a small add-on file instead, so it needs far fewer images and less compute, and it loads on top of its base model.
How do I create a fine-tuned model on Loras AI?
Open Train, pick a supported base model, upload 4–30 images of one subject and review the captions. The credit cost appears before you submit, and finished LoRAs show up in My LoRAs, ready to generate with.
Can I submit a paid training job on this page?
Krea 2, FLUX.1, FLUX.2 Dev, Qwen Image, Z-Image, Wan 2.2 I2V and LTX-2 19B have training and paired generation integrations. Paid tasks require configured services, sign-in and sufficient credits. Other catalog models remain previews.
What happens to the images I add?
Local draft preparation processes images in the current browser tab; exported JSON excludes source files. Refreshing or leaving clears local images. Submitting cloud training sends the selected dataset and captions to the provider; WaveSpeed training also uses configured R2 storage for datasets and archived artifacts. Storage privacy depends on bucket access settings. Generation sends prompts and selected LoRA references to the relevant service. Only submit material you have permission to use.
Can I reuse one LoRA with every image or video model?
Adapters are not interchangeable. Match the model architecture, base version and loading workflow. Krea 2, MOSS and Anima need their own compatible weights; similar names do not imply compatibility. Video workflows may also require multiple adapter files.
Keep exploring
Base models you can fine-tune
Compare related model families before choosing a training workflow. Adapter formats and base-model versions must remain compatible.
Turn your images into a fine-tuned model
Choose a compatible model, prepare images and captions, then submit an available training or generation workflow. You can also export a dataset configuration draft for free.
Your first fine-tuned model starts with a few images
Bring 4–30 images. Leave with a LoRA that keeps them consistent.