-
Notifications
You must be signed in to change notification settings - Fork 25
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Generating instructions #7
Comments
@Lylo01 Did you generate "vector graph" (aka vectorization of the lineart)? |
Do you have a method to generate a vector graph? |
@hyw-dev In the paper, the author used https://github.com/MarkMoHR/virtual_sketching. I'm currently finding a way to run it on Colab. |
I'm thinking about can virtual_sketching directly generate JSON files containing three parameters: vertex, connection, and original index? |
@hyw-dev Afaik, the original index isn't necessary for inference. It can generate vertex and connection but it can't generate JSON output directly. |
Thank you for your reply. I'll also try it out |
hi,i have vector graph picture,then how can i use it? |
hi,i have vector graph picture,then how can i use it? |
After I used virtual_sketching to generate the svg file(vector graph), I found that the svg contained quadratic Bezier curves. I calculated many points through formulas to replace them, and converted them into vertices and connections.Next it takes time to make some modifications to animeinbet's code to allow input of one's own data. |
thanks,Waiting for your good news |
@hyw-dev I think use this one to generate svg file should be a better choice: https://github.com/ivanpuhachov/line-drawing-vectorization-polyvector-flow, but this code depend on a library Gurobi, that is not open source |
I also am curious about how to reproduce inference for custom data. I tried converting SVGs to a seemingly matching format with a script I wrote to no avail. The output of the model is just a blank image. It might also be the case that I introduced some bugs in the repository when changing the logic to accomodate custom inputs that do not rely on 3D models (which, as far as I understood, is something possible to do). Would be nice if someone shared a guide on how to consistently do this! @hyw-dev, do you have any update on the code you were working on? Thank you 🙏 |
We are also attempting to perform inference using our own dataset. If possible, we would like to establish a communication channel to collaborate on improving the code together. Alternatively, if you have already implemented this functionality, we would be interested in obtaining this portion of the code for compensation. |
For what it's worth, I have code + small patches for feeding data to AnimeInbet. I wouldn't call it pretty or otherwise great, but might be better than nothing for some uses. I'm here to answer any questions. |
Hi and thank you so much for your work! It is truly breathtaking!
I've been trying to test this myself but I'm running into many issues with the pretrained weight, mainly I cannot get any video input to work. I've debugged and can read the video file just fine, but afterwards it does not generate anything.
I'm getting frames and labels lenghts 0 as well so I was just wondering if there are any tips on running this locally with Python?
Cheers!
The text was updated successfully, but these errors were encountered: