The short answer
A virtual try-on model built for South Asian clothing will keep a hijab or head covering exactly where it is, and will render a dupatta as a separate piece of fabric with real weight falling over the shoulder. A general-purpose model trained on Western catalogues will usually do neither.
Why the dupatta is hard
Virtually all mainstream try-on models learned from Western e-commerce photography, where a garment is a single piece of cloth stretched over a body: a t-shirt, a dress, a pair of jeans. The model's whole understanding of "clothing" is fabric conforming to a torso.
A dupatta breaks that assumption completely. It is a separate, unstructured length of cloth that is not fitted to anything. It falls, gathers, doubles over itself, and hangs away from the body. Asked to render one, a model with no concept of a free-hanging second layer does the closest thing it knows: it treats the dupatta as surface decoration and prints it flat onto the kameez.
The result is subtly but unmistakably wrong. The colours are right, the placement is roughly right, and yet it reads as a photograph of a printed shirt rather than a woman wearing a dupatta. Anyone who wears one spots it immediately.
Why head coverings get removed
This failure is less subtle. A hijab, scarf, or dupatta worn over the head partially occludes the head, hairline and shoulders — exactly the landmarks an image model uses to work out body position and proportion. Occlusion makes the generation harder.
Models that have seen very few covered heads in training resolve that difficulty the easy way: they generate the uncovered version, which is what their training data overwhelmingly contains. The head covering does not get preserved and rendered; it gets replaced by the model's default.
For a shopper who covers every day, this is not a minor cosmetic issue. The image she is shown is not a preview of how she will look in the outfit, which is the only thing the tool exists to do.
What a correct result looks like
Three things to check on any try-on image before you trust it:
- The dupatta has edges and depth. You should be able to see where the fabric folds over itself, and where it hangs away from the body. If it has no visible edge and sits perfectly flat against the kameez, it has been painted on.
- Your head covering is still there, and still yours. Not removed, and not swapped for a generic scarf in a different colour or drape.
- Your complexion is unchanged. Compare the try-on result against the photo you uploaded. Skin-tone lightening is common and is the failure most likely to go unnoticed, because the image still looks like a good photograph — just not of you.
Aaina Dekho's model is built to hold all three: the dupatta falls with real weight rather than being flattened onto the kameez, a head covering is preserved rather than removed to make the fit easier, and complexion is preserved exactly.
Where it still struggles
Being straight about the boundaries, because a tool that overstates itself costs you a return:
- Very sheer dupattas — heavy chiffon and voile drape well, but extremely sheer organza and net, where the transparency of the weave is the actual product, remain hard.
- Elaborate styled drapes — a simple over-the-shoulder fall is reliable. A specific pinned or pleated styling is not reproduced faithfully.
- Heavily embellished dupatta borders — printed and lightly embroidered borders transfer well; dense zardozi, mirror work and sequinned borders do not.
Frequently asked questions
Does virtual try-on work if I wear a hijab?
It depends entirely on the tool. Many AI try-on models quietly remove head coverings, because a covered head occludes the landmarks they use to estimate body position and their training data contains very few covered heads, so they generate the uncovered version instead. Aaina Dekho's model is built to preserve a hijab or head covering in place rather than removing it.
Why does the dupatta look flat or painted on in most try-on tools?
Because those models were trained on Western e-commerce photography, where clothing is always fabric stretched over a torso. A dupatta is a separate, unstructured length of cloth that hangs away from the body, which breaks that assumption. Lacking any concept of a free-hanging second layer, the model treats the dupatta as surface decoration and prints it flat onto the kameez.
How can I tell if a virtual try-on result is accurate?
Check three things. First, the dupatta should have visible edges and depth where it folds and hangs away from the body, not sit perfectly flat. Second, your head covering should still be present and unchanged. Third, compare the skin tone against your original photo — lightening is common and easy to miss, because the image still looks like a good photograph, just not of you.
Does virtual try-on work with a sheer or net dupatta?
Partially. Chiffon and voile drape reliably. Extremely sheer organza and net, where the transparency of the weave is itself the product being sold, remain difficult, as do densely embellished borders with zardozi, mirror work or sequins.