feat/api-nodes/SD2-realhuman: Real human support for Seedance 2 doesn't work

[Gushousekai195]

Custom Node Testing

• I have tried disabling custom nodes and the issue persists (see how to disable custom nodes if you need help)

Expected Behavior

Real human references should have been accepted without error.

Actual Behavior

!!! Exception during processing !!! API Error: The request failed because the input image may contain real person. Request id: 02177689331112365da4203214fbe81b3e8151fb2922fb189c929 (Type: BadRequest)
Traceback (most recent call last):
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 453, in execute
    raise ex
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 449, in execute
    results.append(r.result())
                   ^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 285, in async_wrapper
    return await f(**args)
           ^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 145, in wrapped_async_func
    return await method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1809, in EXECUTE_NORMALIZED_ASYNC
    to_return = await cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 1536, in execute
    initial_response = await sync_op(
                       ^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 112, in sync_op
    raw = await sync_op_raw(
          ^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 242, in sync_op_raw
    return await _request_base(cfg, expect_binary=as_binary)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 766, in _request_base
    raise Exception(msg)
Exception: API Error: The request failed because the input image may contain real person. Request id: 02177689331112365da4203214fbe81b3e8151fb2922fb189c929 (Type: BadRequest)

Prompt executed in 9.89 seconds

Steps to Reproduce

1. git checkout feat/api-nodes/SD2-realhuman
2. start comfyui
3. open seedance 2 template, ref or first-last frame
4. put in reference of realistic human
5. run

Debug Logs

Adding extra search path checkpoints D:\AI\webui_forge_cu121_torch21\webui\models\Stable-diffusion
Adding extra search path configs D:\AI\webui_forge_cu121_torch21\webui\models\Stable-diffusion
Adding extra search path vae D:\AI\webui_forge_cu121_torch21\webui\models\VAE
Adding extra search path loras D:\AI\webui_forge_cu121_torch21\webui\models\Lora
Adding extra search path loras D:\AI\webui_forge_cu121_torch21\webui\models\LyCORIS
Adding extra search path upscale_models D:\AI\webui_forge_cu121_torch21\webui\models\ESRGAN
Adding extra search path upscale_models D:\AI\webui_forge_cu121_torch21\webui\models\RealESRGAN
Adding extra search path upscale_models D:\AI\webui_forge_cu121_torch21\webui\models\SwinIR
Adding extra search path embeddings D:\AI\webui_forge_cu121_torch21\webui\embeddings
Adding extra search path hypernetworks D:\AI\webui_forge_cu121_torch21\webui\models\hypernetworks
Adding extra search path controlnet D:\AI\webui_forge_cu121_torch21\webui\models\ControlNet
WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.
Found comfy_kitchen backend cuda: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
Found comfy_kitchen backend triton: {'available': False, 'disabled': True, 'unavailable_reason': "ImportError: No module named 'triton'", 'capabilities': []}
Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}
Checkpoint files will always be loaded safely.
Total VRAM 24576 MB, total RAM 65405 MB
pytorch version: 2.5.1+cu121
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 3090 : cudaMallocAsync
Using async weight offloading with 2 streams
Enabled pinned memory 26162.0
Torch version too old to set sdpa backend priority.
Using pytorch attention
Unsupported Pytorch detected. DynamicVRAM support requires Pytorch version 2.8 or later. Falling back to legacy ModelPatcher. VRAM estimates may be unreliable especially on Windows
Python version: 3.11.9 (tags/v3.11.9:de54cf5, Apr  2 2024, 10:12:12) [MSC v.1938 64 bit (AMD64)]
ComfyUI version: 0.19.3
comfy-aimdo version: 0.2.12
comfy-kitchen version: 0.2.8
Invalid version format in requirements.txt: 2.0
ComfyUI frontend version: 1.42.14
[Prompt Server] web root: D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
Asset seeder disabled
D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\python_embeded\Lib\site-packages\kornia\feature\lightglue.py:44: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @torch.cuda.amp.custom_fwd(cast_inputs=torch.float32)
Skipping loading of custom nodes
Context impl SQLiteImpl.
Will assume non-transactional DDL.
Starting server

To see the GUI go to: http://127.0.0.1:8188
Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host
got prompt
WARNING: [Errno 2] No such file or directory: 'D:\\AI\\ComfyUI_windows_portable_nvidia\\ComfyUI_windows_portable\\ComfyUI\\input\\sd_last_frame.png'
!!! Exception during processing !!! API Error: The request failed because the input image may contain real person. Request id: 02177689331112365da4203214fbe81b3e8151fb2922fb189c929 (Type: BadRequest)
Traceback (most recent call last):
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 453, in execute
    raise ex
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 449, in execute
    results.append(r.result())
                   ^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 285, in async_wrapper
    return await f(**args)
           ^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 145, in wrapped_async_func
    return await method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1809, in EXECUTE_NORMALIZED_ASYNC
    to_return = await cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 1536, in execute
    initial_response = await sync_op(
                       ^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 112, in sync_op
    raw = await sync_op_raw(
          ^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 242, in sync_op_raw
    return await _request_base(cfg, expect_binary=as_binary)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 766, in _request_base
    raise Exception(msg)
Exception: API Error: The request failed because the input image may contain real person. Request id: 02177689331112365da4203214fbe81b3e8151fb2922fb189c929 (Type: BadRequest)

Prompt executed in 9.89 seconds
Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host
Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host
Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host

Other

No response

[Gushousekai195]

I updated to the latest commit on the master branch, which literally says "add SD2 real human support", and it's still not working!!!

What is happening?

!!! Exception during processing !!! API Error: The request failed because the input video may contain real person. Request id: 021776916995341b0b21285264a6e57d5de0f3c9dff4850565035 (Type: BadRequest)
Traceback (most recent call last):
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 453, in execute
    raise ex
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 449, in execute
    results.append(r.result())
                   ^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 285, in async_wrapper
    return await f(**args)
           ^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 145, in wrapped_async_func
    return await method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1809, in EXECUTE_NORMALIZED_ASYNC
    to_return = await cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 1576, in execute
    initial_response = await sync_op(
                       ^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 112, in sync_op
    raw = await sync_op_raw(
          ^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 240, in sync_op_raw
    return await _request_base(cfg, expect_binary=as_binary)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 759, in _request_base
    raise Exception(msg)
Exception: API Error: The request failed because the input video may contain real person. Request id: 021776916995341b0b21285264a6e57d5de0f3c9dff4850565035 (Type: BadRequest)

Prompt executed in 11.25 seconds

[troycjj]

same，it's comfyui has no test team?lol

[comfyui-wiki]

Please do the real human validation, you shouldn't use a real human image as an input image, you should create an asset_id first.

Use this workflow https://github.com/Comfy-Org/workflow_templates/blob/main/templates/api_seedance2_0_r2v_real_human.json

[Gushousekai195]

Please do the real human validation, you shouldn't use a real human image as an input image, you should create an asset_id first.

Use this workflow https://github.com/Comfy-Org/workflow_templates/blob/main/templates/api_seedance2_0_r2v_real_human.json

I want to use AI-generated realistic human references; humans who should be nonexistent. Why do I need to upload my face to do even that?

[troycjj]

Please do the real human validation, you shouldn't use a real human image as an input image, you should create an asset_id first.
Use this workflow https://github.com/Comfy-Org/workflow_templates/blob/main/templates/api_seedance2_0_r2v_real_human.json

I want to use AI-generated realistic human references; humans who should be nonexistent. Why do I need to upload my face to do even that?

Even platforms like RunningHub, which are based in mainland China, allow users to directly upload real-person images to generate videos.lol

[nftsimmersive]

The create Image Asset node seems to be not working as of 29th April 2026? I am fully up to date front end and backend on this date:

I get an error it cannot connect to server with the correct API key to make a link for the ID verification in the node (as it does in the demo comfy video for this process)

Exception: API Error: {"error": "internal_error", "message": "byteplus CreateVisualValidateSession: AccessDenied: User is not authorized to perform: ark:CreateVisualValidateSession on resource: trn:iam::3000623291:project/*\n\tstatus code: 403, request id: 2026043000235176A9593C3DB8C34C7654"}

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 453, in execute
raise ex

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 449, in execute
results.append(r.result())
~~~~~~~~^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\execution.py", line 285, in async_wrapper
return await f(**args)
^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\internal_init_.py", line 145, in wrapped_async_func
return await method(locked_class, **inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api\latest_io.py", line 1846, in EXECUTE_NORMALIZED_ASYNC
to_return = await cls.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 1962, in execute
resolved_group = await _resolve_group_id(cls, group_id)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 233, in _resolve_group_id
return await _obtain_group_id_via_h5_auth(cls)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\nodes_bytedance.py", line 198, in _obtain_group_id_via_h5_auth
session = await sync_op(
^^^^^^^^^^^^^^
...<3 lines>...
)
^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 112, in sync_op
raw = await sync_op_raw(
^^^^^^^^^^^^^^^^^^
...<19 lines>...
)
^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 242, in sync_op_raw
return await _request_base(cfg, expect_binary=as_binary)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Users\sysadmin\Documents\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\ComfyUI\comfy_api_nodes\util\client.py", line 766, in _request_base
raise Exception(msg)

[Gushousekai195]

COMFYUI, if you want to be seen as the go to AI user interface, please fix this problem immediately.

I am a professional filmmaker and this problem did not persist on the RUNWAY app. In fact I generated many images there and none of them were flagged with this “real face” error. This looks amateur, especially when all of my images were AI creations.

I am reproducing this issue on the latest ComfyUI build with the Seedance 2.0 Reference to Video workflow.

Environment

• OS: Windows 10
• GPU: RTX 3080 10GB
• ComfyUI version: 0.19.1
• Frontend version: 1.42.11

Problem When I run Seedance 2.0 Reference to Video with a realistic AI-generated editorial still, the request fails with:

API Error: The request failed because the input image may contain real person.

This is happening even though the source image is not a photograph of a real person. It is an AI-generated image.

Why this is a serious problem This makes the Seedance 2.0 Reference to Video workflow unreliable for professional visual work. If AI-generated stills that are photorealistic are being blocked as “real person” inputs, then this breaks a major part of the workflow for filmmakers, fashion users, commercial artists, and anyone creating cinematic reference-driven video.

This is especially frustrating because the same kind of image was accepted without issue in Runway.

Expected behavior AI-generated reference images should be accepted, or at the very least the workflow should clearly explain what type of realistic synthetic image is being blocked and why.

Actual behavior The workflow rejects the image with the “may contain real person” API error.

Steps to reproduce

1. Open Seedance 2.0 Reference to Video
2. Load a realistic AI-generated editorial still image
3. Run the workflow
4. The request fails with the “input image may contain real person” error

Notes

• This is not a photo of a real person
• This is an AI-generated still
• The issue is reproducible
• This is a major blocker for professional reference-to-video work

Please clarify whether this is:

1. a ComfyUI bug,
2. a Seedance API limitation,
3. or a missing flag/setting in the current ComfyUI implementation.

If this is a known issue, please give an ETA for a fix.

If you want a shorter, more aggressive version for GitHub, I can write that too.

Bruh... with all due respect, you should remove that last sentence if you want to be taken seriously.

[comfyui-wiki]

Happy to sync with you guys. Now, you can also use the AI-generated portrait with Seedan2.0 in ComfyUI. So I'll close this issue. Feel free to reopen it if you still encounter any issues.

https://x.com/ComfyUI/status/2050256653409505592

[Gushousekai195]

@comfyui-wiki Not working for me, and this person is 100% AI.

[comfyui-wiki]

@Gushousekai195 Please update your ComfyUI.