Skip to content
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

Bigger output and server setup? #36

Open
Convolutionair opened this issue Apr 20, 2016 · 2 comments
Open

Bigger output and server setup? #36

Convolutionair opened this issue Apr 20, 2016 · 2 comments

Comments

@Convolutionair
Copy link

I am looking into the fork from 'mtyka', as it has a 'hack' or way to deliver bigger output.

  • Can I use a g2.2xlarge Amazon aws for example,
    to render big images: let's say 100 x 200 cm. on 300 dpi?
  • If this is possible, how much time would it take for a render like this on a g2.2xlarge?
    (For calculating the avarage server costs.)

(Most big examples I've seen so far had around the 4000 pix with 96 dpi. So I am thrilled to see bigger outputs.)

Great work on this coding, keep it up!
Thanks in advance

@andersbll
Copy link
Owner

Hi, with the vanilla neural artistic style method I wouldn't expect you to get good results on images of that size. The span of the highest level convolutional features are around 32 pixels. Therefore, the content/style features you merge would be miniscule in an image with a width of 200 cm at 300 dpi (>20k pixels).

@Convolutionair
Copy link
Author

Thanks for the info Anders.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants