Virtual Mirror

[kenkja]

Anyone know if there's an add on for a virtual mirror which would allow users to upload a photo of themselves to see how they look in say sunglasses or jewelry

 

ken

[Guest]

Ken,

 

 

Although there are contributions that allow customers to upload images, there currently isn't anything to show them wearing a product.

 

 

 

 

Chris

[MrPhil]

An interesting idea, but the processing needed to recognize landmarks on the face to determine scale and orientation would be well beyond most servers. For example, to show sunglasses in the right place and at the right size, and accounting for a face not quite directly facing the camera, would be quite difficult. Now, it's much simpler to let the user manually scale and position the item, and even account for their face not quite looking straight into the camera. Some virtual dials or scrollers could be used to set size and position. Once it's been done the first time, that information could be saved along with the image for later use, or trying on different pairs. Either the product image could be manipulated to match the photo, or the photo might be transformed to a standard orientation and scale to fit the product image. It sounds like a fun project for somebody!

[kenkja]

Hi Phil/Chris

 

I've done a bit of research on this looking at other online glasses shops.

 

I'm no programming expert but by using Chrome->inspect element, it appears that (at least the ones I've looked at) use the flashgordan.js and another overlay.js

 

I see your point about the customer being able to manually adjust the spec overlay onto their image, and have always worked on the basis that if it can be done manually then eventually it ought to be able to be done by code, its just a question of being sharp enough at code to do it - I know for sure I'm not !!

 

How about you have a base image which is profile from say shoulders up, the glasses images are presized to the width of the profile at eyelevel at scripted to land at those co-ordinates, when selected, the trick would then be for a script to select from their own image, their profile and resize it to base profile and merge both profiles

 

My photo manipulation program has a "snap to" function, might be the way forward.

 

Anyway, just sort of thinking out loud here, if anyone has more ideas please throw them in

 

ken