Fix resolution | None | Add chedule_setting for init_image

[NeuralAIM]

Fix for resolution <64px

[NeuralAIM]

It will be easier if the user immediately finds out what kind of video card he has:

p.s. And also simplified the output.

---

It was inconvenient to change the settings every time you add or delete init_image. I think it needs to be added to the config settings:

#Maximum step is 25, larger - create artifacts
custom_schedule_setting = [
 [200,1000,8],
 "gfpgan:1.5","scale:.9","noise:.55",
 [1,200,5],
]

#For init image
custom_schedule_setting_init = [
 [100,500,8], #It consists of 50% init_image and 50% model generation
 [1,100,5],
]

---

Added a fix for colab when None was considered as string

def NoneToNone(par): #fix for None(str)
    if(par == '' or par.lower() == 'none'):
        par = None
    return par

I tried to translate the text into English correctly, but maybe there are mistakes somewhere. I'm not a programmer, but I decided to try to help a little, don't judge strictly :)

[apolinario]

Thanks so much for those improvements @NeuralAIM! One thing I didn't get for the init_image is why did you need to add those xtra steps? I currently modify the first step to skip the steps directly in the schedule, could you go over a bit why you did it that way?

[NeuralAIM]

Thanks so much for those improvements @NeuralAIM! One thing I didn't get for the init_image is why did you need to add those xtra steps? I currently modify the first step to skip the steps directly in the schedule, could you go over a bit why you did it that way?

starting_timestep - It didn't work for me. I fixed it :)