From lookbook generation to image expansion and asset management.
This guide walks you through choosing the right feature for the image you want,
and carrying each result forward into your next piece of content.
What Images Should You Upload?
LaonGEN reads your uploaded images to interpret the product’s shape, color, and details.
As with any AI generation, not every result will be perfect —
but the clearer the product information in your image, the more accurate the output.
With images that meet the recommended conditions, LaonGEN’s success rate is over 80%.
We recommend images like these
1. The full product is visible
Use images where sleeves and hems are not cut off,
so the entire shape of the product can be seen.
2. True colors and proportions
Avoid images where strong lighting, shadows, or heavy retouching
distort the product’s color or shape.
3. Crop to just the parts you’ll use
More images do not mean higher accuracy.
If the model’s face will be replaced, crop the original face out of the image;
if you’re only using the top, crop the bottoms away.
Leftover information you don’t intend to use can confuse the AI’s read of the product.
| 👍 GOOD | 👎 BAD |
|---|---|
| Full product visible, nothing cropped | Sleeves or hems cut off by the frame |
| Even lighting, true-to-life color | Colors distorted by harsh light or over-editing |
| Each product clearly separated | Multiple products overlapping with unclear boundaries |
Can I put several products in one image?
Yes, you can.
You can upload up to 2 outfit images.
When an outfit has many pieces — top, bottoms, outerwear, shoes — and you run out of slots,
you can combine several products into a single image.
Keep in mind that when one image contains several products,
the AI may misread how they relate to each other:
- Reading each product’s length differently than intended
- Getting the layering order wrong
- Treating the front and back of one product as two different items
When uploading multiple products together, make sure each one is clearly distinguishable.
TIP — If you upload several products together, describe how they should be styled in the instructions.
Example: “Style the denim jacket worn open over the white T-shirt.”
Remember: the standard for a good input image is simple.
Minimize anything that could confuse the AI’s interpretation of the product.
Generating Lookbooks
This feature creates images of a model wearing your product, based on the outfit image.
Depending on the kind of shot you need, choose between
Standard Lookbook, Horizon Lookbook, and Reference Lookbook.
Which lookbook should I use?
Want a newly staged shot that captures the concept of a background image? → Standard Lookbook
Need clean studio shots showing the front, side, and back? → Horizon Lookbook
Want an editorial that keeps the exact composition and lighting of a reference photo? → Reference Lookbook
Standard vs. Reference Lookbook — which one?
Both take a single photo as input, but they treat that photo very differently.
Input
Results from the same two inputs
Both results were made from the same outfit, model, and single photo.
Standard Lookbook reads the concept from the photo and stages it anew;
Reference Lookbook reproduces the photo’s direction as-is.
1-1. Standard Lookbook
For building your core brand lookbook.
| Input | Model · outfit · background image |
| How it works | Reads concept-related elements from the background — location, pose — and reinterprets them |
| Output | Full-body and half-body shots with the model at the center |
Choose a model and outfit, add a background image,
and LaonGEN generates natural on-model shots informed by the place and mood of that background.
Since the results are mostly full-body or half-body compositions centered on the model,
it works well for the kind of core brand lookbook that shows products naturally.
Background (1) → Results
The seated pose on the sofa, the table lamp, the side table with books,
and the window light all carried over into the results.
The sofa’s color and the lamp’s material are not copied one-to-one —
the elements that made up the scene are restaged instead.
Background (2) → Results
The lean against the window corner, the greenery outside, and the bright wall carried over.
Background (3) → Results
The stone-building street, the parked bicycle, the iron railing, and the cobblestones came through.
These results keep the same model and outfit and change only the background.
Pose, expression, space, and props transfer from the background photo — while the composition is newly created each time.
Recommended for
- Core lookbooks centered on the model and the outfit
- Natural on-model shots across a variety of locations
- Full-body and half-body images for product pages and social media
- Exploring many results around the product rather than one specific staging
1-2. Horizon Lookbook
For showing the product’s fit and shape clearly.
| Input | Model · outfit (front required, back optional) · background color |
| How it works | Generates clean studio-style shots with tidy backgrounds and poses |
| Output | Multiple angles — front, side, back — 4 shots per background color |
Horizon Lookbook creates images that present the product and outfit
from multiple directions against a clean, uniform background.
Since it can generate front, left, right, and back views,
it works well when you want to present fit and shape in a consistent format,
rather than mood or staging.
Standard pose
Free pose
What sets it apart: you choose the background as a color, not a photo.
Pick several colors to render the same outfit side by side per color,
and choose between standard and free poses.
Recommended for
- On-model shots where the overall fit reads clearly
- Front, left, right, and back views of a single product
- Presenting multiple products in an identical background and format
- Baseline on-model images for product pages and listings
1-3. Reference Lookbook
For editorials that carry a reference’s staging and mood.
| Input | 1 base lookbook image · a reference image |
| How it works | Inherits the reference’s composition, lighting, color, and mood |
| Output | Editorial shots recreated in the same staging as the reference |
Reference Lookbook doesn’t just borrow the location in a photo —
it generates a new lookbook informed by the reference’s overall direction:
composition, lighting, color grading, and mood.
It’s the right tool when you have an editorial or campaign image you love
and want to bring that feeling into your own brand’s lookbook.
Reference (1) → Results
Reference (2) → Results
Reference (3) → Results
Compare — the same inputs through Standard Lookbook
Unlike Reference Lookbook, the staging is not followed.
Only the concept of each background is borrowed, and every shot is staged anew.
Recommended for
- Fashion editorials where a specific composition and angle carry the shot
- Images that adopt a reference’s color grading and lighting mood
- Brand campaigns and seasonal editorials where staging matters
- Applying one creative direction across multiple outfits
TIP — Save frequently used backgrounds and your own references to My Assets and reuse them.
Writing Instructions
What matters in an instruction is not its length,
but how well it fits the images you uploaded and the settings you chose.
Rather than repeating information already conveyed by images or settings,
focus on what you want to add or change from the defaults.
Use instructions when you want to:
1. Change the hairstyle — adjust the model’s default hair.
Example: “Give her a low bun with bangs.”
2. Add shoes or accessories — specify styling elements not in the outfit image.
Example: “Add black loafers and silver earrings.”
3. Direct a specific pose or staging — describe the exact pose or shot you want.
Example: “Have the model lean naturally against a wall.”
Don’t repeat what’s already provided
There’s no need to restate information the AI can already see in your uploads,
like the model’s face or the outfit itself.
For example, if you uploaded a model image directly, avoid instructions like this:
| ✕ Not recommended | Why |
|---|---|
| ”Make the uploaded model’s face appear exactly as-is.” | This information was already conveyed by the image |
When images and instructions duplicate the same information or contradict each other,
the AI can get confused about which source to trust.
TIP — Instructions don’t have to be short.
If a detailed instruction produces the style you want, save it and reuse it.
What matters is that the instruction and the input images never conflict.
Expanding Images
This feature takes one image you love and builds out
a series of new poses and compositions in the same mood.
Pick a strong shot from your lookbook, and LaonGEN creates new images
that carry its color, atmosphere, and sense of space forward.
It’s ideal for building an editorial series that feels like
many frames shot at a single photo session.
| Input | 1 image to expand — a lookbook result, an asset, or a direct upload |
| How it works | Keeps the original’s tone and mood, restages the scene by expansion strength |
| Output | An editorial series that flows in the same mood |
Original → Expanded results
How is this different from generating lookbooks?
Want to explore many directions first? → Generate lookbooks
Creates images across varied backgrounds and compositions, based on your outfit and model.
Want to develop one favorite further? → Expand the image
Builds new poses and compositions from a single chosen image, carrying its mood forward.
Used together: find your direction in the lookbook,
then expand the best result into a cohesive series.
See the image expansion guide for hands-on examples.
2-1. Expansion Strength
Choose how much of the original to keep.
The same input produces different results depending on the strength.
Not sure where to start? Get a feel for it with “New composition.”
Almost identical — keeps the same place and composition, changing only pose and angle slightly.
New composition — varies the camera position and framing within the same space.
New space — keeps a similar atmosphere while rebuilding the spatial layout.
Mood only — keeps only lighting, color, and style; the location changes freely.
2-2. Lock Outfit · Lock Model
Use these options to keep the base image’s outfit and model
as consistent as possible across the expanded results.
If you’re building one editorial series with the same model and look,
we recommend turning both on.
Locks on — same model, same outfit throughout
Locks off — model and outfit change freely
2-3. Additional Instructions
For when you want to steer where the results go.
Expansion strength only decides how far from the original to move —
not where to go. The higher the strength, the more the results scatter.
If you have a concept in mind, narrow the results with an instruction.
Below: the same image expanded at “Mood only,” with and without an instruction.
Without instructions
With “daytime outdoors, a garden”
Without instructions, only the lighting and color survive — the location is newly invented each time.
With an instruction, results converge in one direction even at the same strength.
Try instructions like these
- “Daytime outdoors, a garden”
- “Give each shot a different pose.”
- “Add a vintage bicycle as a prop.”
The same principle as lookbook instructions applies here —
don’t repeat information already conveyed by the image.
Recommended for
- Growing one strong shot into a multi-cut editorial series
- Varied poses and compositions with the same model and look
- Scaling up content volume while holding one brand mood
- When the concept is set and you just need more shots
My Assets
A space to save and manage the models, backgrounds, references,
and outfits your brand uses most.
Save your favorite results from lookbooks and image expansion,
or create brand-fit AI models, backgrounds, and references from scratch.
With the right assets saved ahead of time, there’s no need to start over each time —
you can reuse the same model and background across seasons and products,
building a consistent brand image.
What you can do in My Assets
1. Save your best results
Save any result you like from lookbooks or image expansion straight to My Assets.
You can also upload images you already own and manage them as assets.
2. Create models, backgrounds, and references
Even without existing images, you can generate brand-fit AI models,
backgrounds, and references and save them as assets.
3. Reuse them in your next production
Saved assets load directly into lookbook generation and image expansion.
Mark frequent assets as favorites to find them quickly as your library grows.
What kinds of assets can I save?
My Assets is organized into four types: models, backgrounds, references, and outfits.
Each type plugs back into its generation feature as an input image.
| Asset | Used as |
|---|---|
| Model | The model in Standard and Horizon Lookbooks |
| Background | The background in Standard Lookbooks |
| Reference | The reference in Reference Lookbooks |
| Outfit | The outfit in lookbooks |
The All tab shows every asset type at once,
and each type’s tab can filter down to favorites only.
TIP — Favorite the models and backgrounds you use most.
You’ll find what you need quickly even as your asset library grows.
3-1. Creating a Model
For building a dedicated AI model for your brand.
Generate and save AI models that match your brand’s image.
Once created, a model can be reused across all your lookbooks —
keeping your brand imagery consistent even as seasons and products change.
Model settings
| Required | Age range · finish · look direction · gender |
| Optional | Ethnicity · vibe · face shape · eyes · nose · lips · skin · cheeks (AI decides if unset) |
| Face & hair references | Upload reference images for the face or hairstyle you want. For faces, you can also set how strongly the reference is reflected |
Finish offers natural · polished · editorial;
look direction offers striking · pretty · cute.
These two settings strongly shape the model’s overall impression,
so choose them to match where you want the brand to go.
Always test with a real lookbook.
Rather than judging a model from its standalone image, apply real outfits and generate lookbooks before deciding.
Standalone model shots and actual on-model results can differ — try several outfits and confirm the model truly fits your brand.
Recommended flow: create the AI model → apply outfits → review lookbook results → keep as your brand model
3-2. Creating a Background
For backgrounds your brand will reuse again and again.
Create and save backgrounds that match your seasonal concept or brand mood.
A well-fitting saved background can serve lookbooks across many products,
helping every piece of content hold the same brand atmosphere.
Background settings
| Product type | Clothing · bags · shoes — the default framing changes to flatter each type |
| Creation mode | Direct setup · match my outfit · follow a reference |
| Direct setup options | Location · people (1–4) · style mood (AI decides if unset) |
For location, just set the broad direction of the space you want.
Season, time of day, lighting, and weather are handled by the AI.
For clothing, a frontal full-body framing is the default —
use instructions if you want a specific composition or staging.
3-3. Creating a Reference
For building your own shooting direction in advance.
If you don’t have the right image for a Reference Lookbook,
generate a reference with the composition and mood you want, and save it.
Generated references are saved to My Assets
and can be loaded directly when creating a Reference Lookbook.
Reference settings
| Product type | Clothing · bags · shoes |
| Shooting direction | Style: safe · balanced · bold / Purpose: mood catalog · social content |
| Detail settings | Location · light · mood · technique · color (AI decides if unset) |
The shooting style decides how boldly the result is staged.
If you know exactly what you want, set only those items;
if you’re exploring, leave some fields open and see what comes back.
Generating Multiple Combinations at Once
Model, background, and reference creation all share a “set” feature.
Build one configuration, press Add Set to save that combination,
then adjust the settings and add again —
generating up to 4 different combinations in a single run.
Each set can have its own image count.
Generate without adding sets, and you get 1 image from the current settings.
Use it to compare several directions side by side under similar conditions.
Your best lookbook shots can also become short videos.
Usage scenarios and an FAQ will be added to this guide in upcoming updates.