www.tryndbuy.com
Mix & Match
Transforming electronics retail through immersive product visualization and Try on
Transforming electronics retail through immersive product visualization and Try On
Existing solutions in the market
www.tryndbuy.com
Mix & Match
Missing Virtual Try on
Most of the retailer brands suggest what to try with the apparel but there is no virtual try on available on the website.
www.tryndbuy.com
Mix & Match
Missing Virtual Try on
Most of the retailer brands suggest what to try with the apparel but there is no virtual try on available on the website.
www.tryndbuy.com
Comparision of Tryndbuy’s Try-On Apparel Solution with Google Doppl
Parameters
Output Image
Output capabilities
TAT
Upload user's selfie to create digital twin
Prepares body avatar with user's body measurement in real-time
Flat images of products with mannequins required for creating catalogue
Allows users to try mixing and matching of different products - With 80 different apparel categories in their portfolio
Fit and size recommendation basis body measurement to try-on digital twin
• ~4 seconds
Support western dresses (top & bottom dresses), limited support for Indian ethnic wear like sarees and long suits, misidentifying them or not rendering them - Lacks intelligent cloth recognition, often mistakes long garments for short ones
Generate inaccurate clothing shapes and textures, with missing patterns, blur and color mismatches
Generate moderate resolution output, lacking in visual richness
No download option for try-on outputs
• ~18 seconds
www.tryndbuy.com
Incorrect Output by Google Doppl
INPUT 1:
User Uploaded Image
First Step is to generate a 3D Avatar
INPUT 2:
User Uploaded Image
First Step is to generate a 3D Avatar
Different Garment
Length and Cut
Unable to recreate the
garment image shared by the brand. This is because Doppl is built on GAN.
It doesn’t create a generative output. It finds the closest output in its result to generate an output, which can’t be used by brand.
www.tryndbuy.com
Incorrect Output by Google Doppl
INPUT 1:
User Uploaded Image
First Step is to generate a 3D Avatar
INPUT 2:
User Uploaded Image
First Step is to generate a 3D Avatar
Result: Unusable Output
Many ethnic designs are outside the sample set of outputs of GAN based models.
The output needs to simulated, rather than
generated from existing data set.
www.tryndbuy.com
Incorrect Output by Google Doppl
INPUT 1:
User Uploaded Image
First Step is to generate a 3D Avatar
INPUT 2:
User Uploaded Image
First Step is to generate a 3D Avatar
Different Garment
Length and Cut
For high AOV items, low try-on accuracy leads to significant customer drop-off. If the output doesn’t match brand quality, shoppers lose confidence quickly.”
“Since results are user-generated, manual QC isn’t feasible. That’s why extreme accuracy is the north star for driving trust and conversions.
www.tryndbuy.com
Incorrect Output by Google Doppl
Our Solutions
www.tryndbuy.com
Mix & Match
Mix. Match. Perfect.
Tryndbuy allows online shoppers to mix and match different clothes and accessories on their bodies. They can try the same trousers with different tops and shoes until they find the perfect match for them.
www.tryndbuy.com
Mix & Match
www.tryndbuy.com
Mix & Match Comparison
Try ND Buy operates on a one time per style modelling cost instead of per API hit cost.
Once the style is modelled on Try ND Buy platform, it can be used unlimited number of times without any extra cost.
Tryndbuy Modelling cost per style is $11
Thank You
Connect with us at
nitinvats@tryndbuy.com