How to use Lora

LoRA (Low-Rank Adaptation) are specialized "micro-models" that overlay the main model to enhance images with additional styles, unique elements, or to amplify specific features.

Why Use LoRA?

• Styling: Change visual styles (anime, realistic, comic, etc.) without loading a different model.
• Unique Elements: Add specific details like "elf ears" or "special costumes."
• Combination: Merge multiple LoRA to create unique mixes.

How to Use LoRA in Our Bot

1. Enable Pro Mode:
2. Access Image Creation:
3. Select "Create Image (Pro)" or a similar option to access full functionality.
4. Choose LoRA Category:
5. Click "Choose LoRA category" to see categories like "Anime," "Realistic," "Style," "Cosplay," etc.
6. Select LoRA:
7. Click on a desired LoRA name. The bot may provide a preview, description, and options to select or go back.
8. Set Weight:
9. After selecting, choose a weight between 0.1 (minimal influence) to 1.0 (strong influence). Higher weight means a more pronounced effect.
10. Add Multiple LoRA (Optional):
11. You can add several LoRA, e.g., DarkGothicStyle:0.4, ElfEars:0.5. Note that each additional LoRA increases the generation cost by +1 token.
12. Manage LoRA:
13. Use "Manage LoRA" to view or remove selected LoRA. Removing a LoRA excludes its influence from the final image.
14. Enter Prompts:
15. Positive Prompt: Describe what you want to see (in English).
16. Negative Prompt: Specify what to exclude (also in English).
17. Generate Image:
18. The bot will check your token balance (remember, each LoRA adds to the cost), queue the task, and deliver the generated image upon completion.

Example Usage

1. Click "Create Image (Pro)."
2. Navigate to:
3. Choose LoRA category → select "Gothic / Dark."
4. Select LoRA g0th1cPXL.
5. The bot sends a preview; click "Select this LoRA."
6. Set weight to 0.4.
7. To add another LoRA, repeat the steps.
8. In "Manage LoRA," you might see:
9. g0th1cPXL:0.4
10. incase-ilff-v3-4:0.4
11. Enter Positive Prompt:
12. 1girl, red hair, gothic dress, standing in dark church, ...
13. Click "Generate Image" — done!

FAQs

• Can I set the weight above 1.0?
• The menu suggests a maximum of 1.0. Higher values might be ignored or produce overemphasized results.
• Will selecting 5 LoRA improve the result?
• Not necessarily. Multiple LoRA can blend styles, but too many may clutter the image. Experiment cautiously.
• What is the cost of using LoRA?
• Each additional LoRA (beyond the first) adds +1 token to the generation cost.

Conclusion

LoRA offers a quick way to add stylistic enhancements to your model.

In this bot, the process is intuitive:

1. Enter Pro Mode.
2. Select LoRA (via categories).
3. Set the weight.
4. Optionally, add more.
5. Enter positive/negative prompts.
6. Click Generate — and you're done!

Happy creating!

Examples:

Using the standard LoRA installed in the bot and the following positive prompt:

detailed face, perfect face, A corrupted draenei mage with glowing green tattoos and sharp fel-infused horns. Her glowing green eyes and fiery aura radiate chaos as she wields a staff crackling with fel energy. Her revealing fel-forged armor exposes her flawless, glowing skin, her tail swishing as she walks confidently through the shattered Outlands.

Adding LoRA FNTS:0.6

Append the activation tag (provided in the LoRA description) FNTS to the positive prompt.

Adding LoRA LVTHN:0.6

Append the activation tag (provided in the LoRA description) LVTHN to the positive prompt.

Adding LoRA ARKN2:0.6

Append the activation tag (provided in the LoRA description) ARKN2 to the positive prompt.

Adding LoRA Concept Art Style SDXL_LoRA_Pony Diffusion V6 XL:0.6

This LoRA does not have activation tags, so no additional text is appended to the prompt.

Adding LoRA Watercolor Anime Style LoRA_Pony XL v6:0.6

Append the activation tag (provided in the LoRA description) watercolor \(medium\), traditional media to the positive prompt.

Adding LoRA Sketch Illustration Style [LoRA]:0.6

This LoRA does not have activation tags, so no additional text is appended to the prompt.

As you can see, LoRA allows you to change the style of the generated image (and much more) while using the same prompt and without switching the model. They open up nearly limitless possibilities for expressing your imagination in all possible variations.