yael-vinker / live_sketch Goto Github PK

Hi, thank you for your open-source code and great work! I wanted to ask if it's normal for two GeForce 4090s (RAM: 16Gib for each) to run out of memory while using animate_svg.py?
Are there any tips on how to save memory space that you could recommend? Thanks!

Hi, thank you for great works.
I think there is some minor typo, it should be self.frames_rigid_scale rather than self.frames_rigid_shear

Thank you for your nice work. Is there no train code released?

cool tool, I would be grateful for help with the following

environment.yml mentions

• python=3.8
• torch==1.12.1+cu116
• xformers

but upon installation on ubuntu 20.04

• xformers requires -> pytorch=2.0.1
  and
• xformers -> python[version='>=3.10,<3.11.0a0|>=3.9,<3.10.0a0'] (although it may accept 3.8.18 when installed via pip xformers rather than conda)

I reckon torch 1 is because of diffvg. I jumped through a lot of hoops to get diffvg compiled (e.g. downgrading to gcc 7 is a must) via DIFFVG_CUDA=1 python setup.py install

script CUDA_VISIBLE_DEVICES=0 python animate_svg.py runs but throws this error BachiLi/diffvg#3 and does nothing
It would be good to know what the min/max requirements exactly are. thanks in advance

Hi Rinon.

In this function, there is only points_delta_optimizer.zero_grad(). Why there is no global_delta_optimizer.zero_grad(). Did I miss something? Thanks.

Hello, thanks for your amazing work! I'd like to know which diffusers the project use?

Hi, this is truly a great piece of work!
One of the things I noticed is that the inference or optimization time per video generation is pretty large (I tried an example that took about 35mins). Is this expected?