Get direct support to install Instamodel & more!: Instara Discord
Our first Open-Source Character LoRA!
The response to Instagirl V2 has been incredible. Since its launch, we've been grinding and investing a massive effort to tackle one of the biggest challenges in AI art: character consistency.
Today, we're not just releasing a new model but committing to giving back. As a thank you to the community that has supported us, we are proud to make our first consistent character LoRA fully open source.
Meet Lucy, Your Consistent, Open-Source Character
Lucy is a breakthrough in our workflow. She is an original AI-native character, created through a unique synthesis pipeline we developed. She is not based on any real-world person. Our goal was to create a versatile, realistic, and consistent face that can serve as a canvas for your creativity.
This LoRA is designed to give you the power to tell stories, build series, and create a cohesive body of work with a recurring character, while maintaining the hyper-realism you expect from Instara.
How to Use
This is a character LoRA. For the best results and intended aesthetic, we strongly recommend using it with our Instagirl V2.0 style LoRA, along with the included Low-Noise LoRA and workflow in the ZIP.
Prompt Structure
To get the most out of this LoRA, follow this general structure. Start with the character trigger, then layer in details about her appearance, the scene, and the overall style. This gives you precise control over the final image.
Template:
Instamodel_1, [Character Details], [Scene & Composition], [Style & Quality Tags]
* [Character Details]: Describe her expression, physical features, clothing, and action.
* Example: a woman with medium‑light skin and clear freckles, shoulder‑length straight black hair, hazel‑green eyes narrowed into a sharp, challenging glare, a cold, confident smirk on her lips, wearing a black silk button-down shirt with top buttons undone, revealing her black cord necklace with a small round white pendant, leaning forward over the table, holding a hand of cards
* [Scene & Composition]: Describe the location, background, lighting elements, and shot type.
* Example: medium shot in a dark, smoke-filled room at a private poker table, looking directly at the camera from an opponent's POV, lit by a single low-hanging lamp above the green felt table
* [Style & Quality Tags]: Add keywords for mood, realism, and camera/film effects. For best results, use our foundational Instagirl trigger here.
* Example: Instagirl, dangerous charm, kept delicate noise texture, amateur cellphone quality, visible sensor noise, heavy HDR glow, amateur photo, blown-out highlight from the lamp, deeply crushed shadows
Putting It All Together
Instamodel_1, medium shot in a dark, smoke-filled room at a private poker table, a young woman with medium‑light skin and clear freckles, her shoulder‑length straight black hair is sleek, her hazel‑green eyes are narrowed into a sharp, challenging glare, with a cold, confident smirk on her lips, she wears a black, silk button-down shirt with the top buttons undone, revealing her black cord necklace with a small round white pendant, leaning forward over the green felt table, holding a hand of cards, looking directly at the camera (from an opponent's POV), the scene is lit by a single low-hanging lamp above the table, Instagirl, kept delicate noise texture, dangerous charm, amateur cellphone quality, visible sensor noise, heavy HDR glow, amateur photo, blown-out highlight from the lamp, deeply crushed shadows.
Truly Open Source - Our Gift to You
We believe in empowering creators. This model is released with a permissive license to give you maximum creative freedom.
✅ Merge & Adapt: You are free to merge this model into your own creations and release them.
✅ Sell Your Generations: You are free to use images generated with this model for commercial purposes, including prints, commissions, and other projects.
✅ Use on Generation Services: You are permitted to use this model on generation services.
✅ No-Strings-Attached: This LoRA is yours to use.
Credit is not required, but is highly appreciated. If you enjoy our work, a shout-out to Instara and a link back to this page helps us a lot.
⚖️ Ethical Use Policy:
While the license is open, we ask that you use this model responsibly.
No Impersonation: Do not use this model to create images intended to impersonate or defame real individuals.
No Illegal Content: Do not use this model to create hateful, illegal, or harmful content.
Community & Support
This is a big step for us, and we want to see what you create. Share your work, get direct support, and hang out with the team in our official Discord server!
Description
FAQ
Comments (12)
great work, could you share training data?
Superb creation! While I really appreciate you thoughtful gesture, there is absolutely nothing here that is open source, you're not even sharing the dataset?
To make it "Truly" open source would be to share training scripts / dataset / tools used etc etc....
Other than that, to be totally honest, without cracking it open and sharing the gory content its just another Lora, a good one yeah, but Loras of this quality are easy to make with Wan 2.2 and Ai-Toolkit. The model is the best ever made for Images and video.
There is nothing open source here, and this is not giving anything back, except a Lora and a prompt.
Simply saying people can use it in XYZ| generation does not make it open source, people will use it in XYZ generation if you grant them permission or not,
Just like they do for every other Lora....
Thanks for sharing though nice work.
will you produce a wan 5B version?
Does anyone knows where to run this LoRA, I mean any cloud inference platform as I don't see anyone providing wan 2.2 lora
People here with Wan as their first model gazing at a character lora with open mouth:)
Lucy is a breakthrough in our workflow. She is an original AI-native character, created through a unique synthesis pipeline we developed
Dude, it's just a character LoRa. There are thousands of such LoRas on Civitai. I don't get what is so unique about it...
What's with the unnecessary zip folder?
How much time it takes generate one image?
Another goth mommy added to the collection
Could you share the tutorial you followed to train the Lora?
how did you create the dataset
Would it be possible to make a Lora for her in Illustrious or Pony or Flux , So she could be inpainted into multiple worlds, animation styles 3D, and other interesting Image to Video fun effects?