What should be my ideal input should look like for the best result?

[mishra-anubhav]

I have been using the product for quite a few days and I love the product-

But I am facing some issues with few images, output is extremely bad. I want to reach out and ask what is the best pixel quality, dimensions, lighting required for the input image.

[AjayJoshi-Chicmic]

well i have also used it a lot of times and i can only tell you what i have observed. It's better to add images having clear backgrounds and for the garment image if you upload an only garment with no human its better and the garment should be as big as possible like filling whole image frame. Apart from that pixel wise its trained on 768 * 1024 images so it gives better result on them but i also gave like high resolution images and it gave not so good results

[mishra-anubhav]

Yes, I am planning to finetune this model.

[AjayJoshi-Chicmic]

That sounds great! If you get it working, please share it with me—and if you run into any issues, I’d love to hear about those too. I’m really fascinated and interested in this!