AMD: input must be 4-dimensional

[naive cairn]

So i tried following this guide
https://comfyui-wiki.com/en/tutorial/advanced/video/wan2.1/wan2-1-video-model?utm_source=chatgpt.com

specifically: "2. Wan2.1 Image-to-Video Workflow"

I use
wan2.1_i2v_480p_14B_bf16.safetensors
umt5_xxl_fp8_e4m3fn_scaled.safetensors

and chose those in the workflow.
I basically just followed the guide, downloaded the files, added them where it said (the "text:encoders/" folder is the "clip" folder)
i didnt change any other parameters and did all of this in the comfy workflow tab.
I hit run and get:

**"WanImageToVideo

input must be 4-dimensional "**

with these logs:

https://paste.denizenscript.com/View/137410

my server backends: "--directml --lowvram"
i tried removing those and just adding "--cpu" instead, which resulted in the error not appearing, but also the server simply disconnecting and restarting (i have 32gb of ram) from what i saw the cpu was only having a 50% workload and the ram only used liked 10gb

Graphics card
Amd Radeon RX 5700 XT 8GB
Sapphire Nitro+ special edition
Drive
Samsung 990 PRO NVMe M.2 SSD
Processor
Amd Ryzen 9 3900x 12 core prozessor
Ram
Corsair Dimm 32GB DDr4-3600 kit
Mainboard
Asus Rog Strix B550-F Gaming

Let me know if any other information is needed

[icy trail]

• You followed the most complicated instructions, not the easy ones. Easy instructions here https://github.com/mcmonkeyprojects/SwarmUI/discussions/716
• note amd gpu on windows is unstable at the best of times https://github.com/mcmonkeyprojects/SwarmUI/blob/master/docs/Troubleshooting.md#amd-on-windows