How to run Super Resolution code?

[zelenooki87]

I'm really struggling to figure it out. Thanks.

[dome272]

You would just run all the cells in the notebook that are for the SR ControlNet. Did you figure it out already or need more help?

[cersar]

with torch.no_grad(), torch.cuda.amp.autocast(dtype=torch.bfloat16):
    effnet_latents = core.encode_latents(batch, models, extras)
    effnet_latents_up = torch.nn.functional.interpolate(effnet_latents, scale_factor=2, mode="nearest")
    cnet = models.controlnet(effnet_latents_up)
    cnet_uncond = cnet
    cnet_input = torch.nn.functional.interpolate(images, scale_factor=2, mode="nearest")
    # cnet, cnet_input = core.get_cnet(batch, models, extras)
    # cnet_uncond = cnet
    
show_images(batch['images'])
show_images(cnet_input)

@dome272 I am confused by the above code snippet about super resolution.It seems to merely shows the result of direct upsampling.