It’s important that we deliberately don’t start by creating a character from scratch. Instead, we start with scaling: first learning how to turn one or two images into dozens or hundreds of new ones, and only then moving on to creating the very first “base” image.
This order is essential. If you don’t understand how to scale an existing look, creating a character from scratch will not produce stable results later on.
Starting Point: We Already Have an AI Character
We begin with a simple and very common situation:
you already have an AI persona with a few existing images.
This could be:
a character you worked with before,
an AI model from social media,
or simply a set of strong images you consider the “canon.”
In the example, a real AI model is used: images are simply downloaded from Instagram and taken as raw input, without any prior editing. The task is to start producing new content while staying within the same character identity.
Tool 1. Carousel — Fast Variations of Existing Photos
The first and simplest tool to start with is Carousel.
Its core idea is straightforward:
from one existing image, generate multiple new angles and variations that would be difficult or time-consuming to describe manually in a prompt.
How We Use Carousel in Practice
We take several existing photos of the character — for example, 3–4 images from different sources. For each one, we set:
the number of variations (for example, 4 per image),
the aspect ratio (if needed),
and run the generation.
In the example, while the current task is generating, you can see results from previous runs — and this clearly shows when this tool is especially useful.
When Carousel Works Best
Carousel is ideal when:
you already have a strong final image,
you want to create an Instagram carousel,
or you need additional angles that are hard to describe in words.
The examples show that from a single image you can get:
side shots,
top-down angles,
close-ups with detail emphasis,
back views,
and even stylized angles like a “fish-eye” look.
It’s important to understand that not every run will be perfect. Sometimes variations are too similar; sometimes errors appear (for example, with legs). This is normal for AI generation — which is why results should always be reviewed. Overall, Carousel is one of the fastest ways to diversify content.
Consistency Is the Key Advantage
In the example generation, you can see that:
the character remains the same,
clothing is preserved,
hairstyle and accessories (such as earrings) repeat across frames,
locations look logically connected.
This makes Carousel an excellent tool for regular content production.
Tool 2. Photoshoot — A Fast Themed Photoset
The second tool is Photoshoot. It’s used when we want not just variations of one image, but a cohesive themed photoset.
Preparing Input Data Correctly
For Photoshoot, we need to upload:
A face image — recommended framing is shoulders or chest-up.
Full-body images should not be used for the face.
A body image — this is critically important if you want consistent proportions.
Even if the face and body aren’t perfectly separated, the tool still works correctly. However, cleaner references always lead to more stable results.
Photoshoot Settings
Next, we choose:
aspect ratio,
resolution,
the photoshoot theme (for example, lingerie, casual, studio, etc.),
the number of images.
Then we start the generation.
The examples show that in just 2–3 clicks, you get a clean, consistent photoset.
These results work well for:
publishing,
testing visual styles,
further processing (for example, sending into Carousel).
It’s also noted that the body reference can be nude — but NSFW specifics are covered in a separate lesson.
Tool 3. Editor / Combiner — The Core Working Tool
Next, we move to one of the key tools of the entire system — Editor & Combiner.
Put simply, Editor & Combiner is NanoBanana without censorship.
However, the tool includes multiple models, and understanding the differences is essential before starting work.
Available Models and Their Differences
Editor & Combiner offers two main models:
1. NanoBanana Pro (Safe for Work)
This is the censored version of NanoBanana.
With this model:
you can’t generate even mildly revealing scenes;
lingerie, semi-nude looks, and similar scenarios are effectively unavailable;
it’s suitable only for strictly SFW content.
2. General Checkpoint (Safe / Not Safe for Work)
This is the main workhorse of the tool.
In practice:
it matches NanoBanana Pro in quality and consistency,
it fully understands scenes and characters,
but it is completely uncensored.
By choosing the General checkpoint and providing character references, you can:
create almost any scene,
change clothing, environment, and actions,
generate images in virtually any style described by prompts or references.
This is why Editor & Combiner is almost always used with the General checkpoint in real workflows, while NanoBanana Pro remains a secondary option for strictly safe content.
Practical Use of Editor & Combiner
Next, the lesson reviews real examples generated earlier.
The same familiar character reference — the image used in previous tools — is provided as input. The task is to generate multiple images within a single photoset.
The result:
a series of images with the same character,
a visually cohesive photoshoot,
frames that are not copies of the original image but remain clearly recognizable.
When working with Editor & Combiner, we can:
fully describe scenes in text (kitchen, pool, home, etc.),
combine text prompts with visual references,
control both appearance and environment.
Why Editor & Combiner Becomes the Main Tool
This tool often becomes the most frequently used for a simple reason:
we upload a character reference,
optionally add pose, location, or other model references,
and ask to show our character in the desired context.
Effectively, Editor & Combiner allows you to:
take one character image,
and turn it into an unlimited number of new images,
across any locations and scenarios.
To use it efficiently, it’s recommended to:
study available templates,
review example tasks,
understand how to structure prompts specifically for this tool.
Tool 4. Image-to-Image — Inspiration Through References
The next step is Image-to-Image, a tool for working with visual references.
A particularly powerful technique is used here:
find another model with a similar body type or style,
use their images for inspiration, not copying,
and generate new images with your own character.
Workflow Overview
We upload:
several reference images,
the face and body images of our character.
With the General model, you can:
preserve the pose,
change or keep the background.
The result is a series of images that:
are clearly inspired by the references,
are not copies,
while maintaining full character consistency.
The example shows how, in just a few clicks, you can generate multiple photosets — different scenes, different moods, the same character.
Combining Tools
An additional advantage:
any successful image can be sent directly into Carousel to generate even more variations from stronger, refined frames.
Sequential Generation: Full Consistency Across a Series
Separately, the lesson covers Sequential Generation.
This mode produces a series of images where:
the character remains the same,
clothing is preserved,
accessories and details don’t shift between frames.
This is especially useful for:
stories,
serial posts,
visual storytelling.
The examples show that with an insufficiently detailed prompt, the model may keep the same pose and only change the location — highlighting once again the importance of detailed action descriptions.
Lesson Summary: Four Practical Ways to Scale Content
At this point, we’ve covered four reliable methods for generating new images from existing ones:
Carousel — fast new angles and variations from finished images; supports batch mode and scales well.
Photoshoot — quick themed photosets based on face and body references.
Editor / Combiner — the main tool for scenes, locations, and flexible reference-based work.
Image-to-Image with references — a way to draw inspiration from other models and create new photosets with your character.
All of these tools can be combined, turning one strong image into dozens or hundreds of new ones without losing consistency.
Final Thought
Even if a prompt isn’t perfect, you can still get usable results — but detailed prompts and the right tool choice are what unlock stable, scalable content.
By using Carousel, Photoshoot, Editor, and Image-to-Image, you get a complete workflow for creating safe-for-work photoshoots based on an existing AI character — a foundation on which you can confidently build an influencer from scratch.