Seeking Advice on Building a Custom Virtual Try-On Model Using Pre-Existing Models #10089
Unanswered
4833R11Y45
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hi everyone,
I'm currently working on a custom virtual try-on model and I need some guidance. My goal is to leverage pre-existing models and modules to create a more comprehensive and flexible virtual try-on system. Here are my specific requirements and challenges:
Using Pre-Existing Models and Modules:
I want to utilize pre-existing models such as OpenPose, Detectron2, Stable Diffusion, and IP-Adapter to minimize the amount of heavy lifting required. Has anyone successfully integrated these models for a similar project? Any best practices or tips?
Comprehensive Clothing Support:
Most of the existing virtual try-on models either work with upper clothes or full dresses. However, I need a model that can handle upper clothes, full dresses, and lower body clothes (pants, shorts, skirts). How can I extend the current models to support all these types of clothing in a single system?
Flexible Clothing Analysis:
Is it possible to make the system analyze and adapt the clothing type based on the user's current attire and the clothing item they want to try on? For example, if a person is wearing a shirt and pants and wants to try on a full dress, the model should adapt the dress to fit as a shirt. Conversely, if trying on shorts over trousers, the model should not stretch the shorts to fit like trousers.
Preventing Misalignment:
How can I ensure that certain types of clothing do not get inappropriately stretched or misaligned? Specifically, if a model is wearing full-length pants or trousers and wants to try on shorts, the model should correctly fit the shorts without stretching them. The same should apply when trying on full-length pants over shorts.
Any advice, suggestions, or examples of similar projects would be greatly appreciated. I'm particularly interested in how to integrate these functionalities seamlessly and ensure high-quality, realistic try-on results.
Thanks in advance!
Beta Was this translation helpful? Give feedback.
All reactions