fix generate backend to generate "accurate" intermediate images (#2787)

The sample_to_image method in `ldm.invoke.generator.base` was still
using ckpt-era code. As a result when the WebUI was set to show
"accurate" intermediate images, there'd be a crash. This PR corrects the
problem.

- Closes #2784
- Closes #2775

ldm/invoke/generator/base.py

@@ -137,17 +137,9 @@ class Generator:
         Given samples returned from a sampler, converts
         it into a PIL Image
         """
-        x_samples = self.model.decode_first_stage(samples)
-        x_samples = torch.clamp((x_samples + 1.0) / 2.0, min=0.0, max=1.0)
-        if len(x_samples) != 1:
-            raise Exception(
-                f'>> expected to get a single image, but got {len(x_samples)}')
-        x_sample = 255.0 * rearrange(
-            x_samples[0].cpu().numpy(), 'c h w -> h w c'
-        )
-        return Image.fromarray(x_sample.astype(np.uint8))
-
-        # write an approximate RGB image from latent samples for a single step to PNG
+        with torch.inference_mode():
+            image = self.model.decode_latents(samples)
+        return self.model.numpy_to_pil(image)[0]
 
     def repaste_and_color_correct(self, result: Image.Image, init_image: Image.Image, init_mask: Image.Image, mask_blur_radius: int = 8) -> Image.Image:
         if init_image is None or init_mask is None: