min_size and max_size in random_samples.py?

[peterk]

trafficstars

What is the intended use of [--min_size MIN_SIZE] and [--max_size MAX_SIZE] in random_samples.py? Setting min_size larger than 256 seems to impact image generation but does not result in larger images.

[tamarott]

min_size and max_size effect the training. These are the minimal/maximal image dimensions over the model's levels. The number of level is adjusted accordingly.

[peterk]

Ah ok!, thank you! Is there a way to output higher resolution images from random_samples.py?

[JonathanFly]

Your examples used 250px size training images, though in some cases it was 1000px wide by 250px high. If I want to maximize quality at the expense of training time, should I use the highest resolution input that can still fit into GPU memory?

[tamarott]

yes, you can adjust max_size according to the desire output size.

[peterk]

Ah so training on a larger image I can use this?

python main_train.py --input_name <my 2000x800 image.png> --max-size 2000

[tamarott]

yes, but you may encounter some computation limitation

[JonathanFly]

With 16GB of GPU memory, the highest resolution output I have achieved is 667 x 413 (scale 9 or 10). Does this seem about right? I'd like to fit more information in there so I will experiment with different aspect ratios for training images.

[warrider301]

python main_train.py --input_name <my 2000x800 image.png> --max-size 2000 I tried this but I got an error stating; main_train.py: error: unrecognized arguments: --max-size 2000 Can anyone please help to solve this issue

[tamarott]

you need to use max_size instead of max-size