[Bug]: Z-Image turbo Black images

### ComfyUI and Frontend Versions

I am using ComfyUI Desktop Version, currently the latest version 0.3.76.

### Description

When I open the template for Z-Image and start generating, I get a completely black image.
I have checked the model, clip, and vae several times, and I have re-downloaded the model, clip, and vae, but nothing has changed; the problem persists, and the images are completely black.
I noticed the problem with my workflow, which worked perfectly in version 0.3.75, but with the current version of ConfyUI it gives black images, so I restarted ComfyUI and opened the template for Z-Image, as mentioned before, completely black image.

### Steps to Reproduce

I open the template and start the generation.

### Debug Logs

```powershell
got prompt
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:1, offload device: cpu, dtype: torch.bfloat16
loaded partially; 9728.80 MB usable, 9501.73 MB loaded, 2237.82 MB offloaded, 225.00 MB buffer reserved, lowvram patches: 0
100%|██████████| 9/9 [00:24<00:00, 2.73s/it]
Requested to load AutoencodingEngine
Unloaded partially: 3557.98 MB freed, 5943.75 MB remains loaded, 225.00 MB buffer reserved, lowvram patches: 0
loaded completely; 248.25 MB usable, 159.87 MB loaded, full load: True
Prompt executed in 26.64 seconds
```

### Browser Logs

It is not the Portable version. I am using the Desktop version.

### Setting JSON

[comfy.settings.json](https://github.com/user-attachments/files/23997641/comfy.settings.json)

### What browsers do you use to access the UI ?

_No response_