search

junyanz / interactive-deep-colorization

Deep learning software for colorizing black and white images with a few clicks.
https://richzhang.github.io/ideepcolor/
MIT License
2.69k stars 447 forks source link

Local hints render area too small #74

Open leochli opened 4 years ago

leochli commented 4 years ago

Hi,

Your repo is amazing and I have been playing around with it. I am wondering is this normal?

I have given a lot of user guided points already. But the colorization seems not working, only a small portion is rendered.

Btw, I am running a GTX 2080 Ti gpu, yet the latency is still very high. Is there anything I was missing?

image

leochli commented 4 years ago

PyTorch model seems not working so well?

FYI, I am using PyTorch backend and your pretrained model. Here's the output of your own test image: image

richzhang commented 4 years ago

Hi, are you using the caffe weights converted into PyTorch?

leochli commented 4 years ago

Hi Richard,

I was using the pytorch_trained.pth fetched from fetch_models.sh

polymesh commented 4 years ago

I'm having the same issue. I provide a hint and the affected area is so small I can barely see anything change. If I cluster up a ton of hints, it looks a little more splotchy with color. Almost seems like this is manual coloring - nothing like what we see in the video showing what it can do.

I'm using the PyTorch model on a GTX 980 Ti on Windows 10. Followed the installation instructions. Using Python 3.7 since that's the latest python that supports Qt4.