Open wilhelmli opened 1 year ago
You may ask the author of that thread.
You may ask the author of that thread.
I am the author lol. I experimented with a few deepfake models(render videos), but the quality is not yet ideal for photogrammetry.
Is there an option for batch swap similar to Upscayl?
What is the ideal resolution for metahuman?
What is the ideal resolution for metahuman?
The resolution of the current model on MJ is already sufficient. What we need is video input, or batch process the images frame by frame from input video.
I see, right now it's not possible. Do you think it's a very effective tool for creating metahuman, if it supports video pipeline?
I see, right now it's not possible. Do you think it's a very effective tool for creating metahuman, if it supports video pipeline?
If achieved, we can push out mesh WITHOUT video captures of the actual person. 我觉得insight模型挺屌的,有很大空间。
I see, right now it's not possible. Do you think it's a very effective tool for creating metahuman, if it supports video pipeline?
If achieved, we can push out mesh WITHOUT video captures of the actual person. 我觉得insight模型挺屌的,有很大空间。
你这个评价很屌
Here is another interesting example. Reversed process!
/saveID from this 3D clay mesh
/swapid input
/swapid image output
sharpness is even higher than the original image
sharpness is even higher than the original image
Particularly, Insight model did a really good job for Eyes, which is something quite valuable. 眼睛是心灵的窗户哈哈
For this image does it look similar to the source identity?
For this image does it look similar to the source identity?
I created a new id with an image of young Keanu.
Then this one with Keanu at present day, the facial matches really well~
looks good
Any solutions for creating MetaHumans using a deepfake version of InsightFaceSwap?![Screenshot_2023-05-27_at_10 40 16_ins](https://github.com/deepinsight/insightface/assets/37603769/d861e024-c1c4-43ba-8fe3-02364313aed7)
Reddit post: https://www.reddit.com/r/unrealengine/comments/13sbl5u/am_i_the_only_one_using_insightfaceswap_for/