help! is that error cause the black frames? runtime warning for invalid value

[abumolly]

custom_nodes/ComfyUI-VideoHelperSuite/videohelpersuite/nodes.py:95: RuntimeWarning: invalid value encountered in cast return tensor_to_int(tensor, 8).astype(np.uint8) really don't know how to solve the problem, is that error cause the animatediff work flow generated black frames? really appreciate for your reply.

[AustinMroz]

The black outputs are a fairly common issue and are the result of something going wrong earlier in the workflow. If you pass the ouput into a Preview Image node, you'll get the same black output.

These can be quite inconvenient to debug, but I'd suggest you start with the example txt2img workflow and change one thing at a time until it breaks and produces black output.

[abumolly]

thank you for your suggestion, I did used the original txt2img workflow, actually everything was ok before I installed tensorflow and pytorch, which followed chatgpt 4o to enhance the efficience of rendering in mac m3. For now, other workflows were all rendered ok except animatediff ones, and I reinstalled animatediff which was not useful.

[abumolly]

ComfyUI/custom_nodes/ComfyUI-VideoHelperSuite/videohelpersuite/nodes.py:95: RuntimeWarning: invalid value encountered in cast return tensor_to_int(tensor, 8).astype(np.uint8)---------that's the log

[Kosinkadink]

If everything was working before chat gpt blindly told you to edit venv stuff, then your venv is likely messed up. You can create a new venv and set it up from scratch and avoid any and all AI-generated install suggestions.

[abumolly]

If everything was working before chat gpt blindly told you to edit venv stuff, then your venv is likely messed up. You can create a new venv and set it up from scratch and avoid any and all AI-generated install suggestions.

thank you so.... much for your reply. not sure whether it's the problem of the "environment.yaml", since i got the error as below, when I create the new venv and prepare to reinstall the animatediff. do I have to change the code of the "environment.yaml" for Mac system?

conda env create -f environment.yaml Retrieving notices: ...working... done Channels:

• pytorch
• nvidia
• defaults Platform: osx-64 Collecting package metadata (repodata.json): done Solving environment: failed Channels:
• pytorch
• nvidia
• defaults Platform: osx-64 Collecting package metadata (repodata.json): done Solving environment: failed

LibMambaUnsatisfiableError: Encountered problems while solving:

• nothing provides cuda 11.7.* needed by pytorch-cuda-11.7-h67b0de4_0

Could not solve for environment specs The following package could not be installed └─ pytorch-cuda 11.7* is not installable because it requires └─ cuda 11.7. , which does not exist (perhaps a missing channel).

[Kosinkadink]

ComfyUI's readme has a guide for installing ComfyUI on Mac: https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#apple-mac-silicon

ComfyUI does not use conda for its environment for the default install, just pip. I'm not sure what guide you are using, but it does not apply to Mac at all; CUDA is an NVIDIA-card only thing, so it is not a thing for mac installs.

[abumolly]

Really appreciate for your inspiration. while, I canceled conda and reinstalled the animatediff, the problem was still there, black output with the error :

ComfyUI/custom_nodes/ComfyUI-VideoHelperSuite/videohelpersuite/nodes.py:95: RuntimeWarning: invalid value encountered in cast return tensor_to_int(tensor, 8).astype(np.uint8)

seem problem was about the animatediff model, since it's ok if cancel the animatediff related nodes.

[abumolly]

For the simple workflow, there's no errors report but outputs are black.

got prompt [rgthree] Using rgthree's optimized recursive execution. Requested to load SD1ClipModel Loading 1 new model [AnimateDiffEvo] - INFO - Regular AnimateDiff activated - latents passed in (8) less or equal to context_length None. [AnimateDiffEvo] - INFO - Using motion module mm_sd_v15.ckpt:v1. 100%|███████████████████████████████████████████| 20/20 [01:00<00:00, 3.05s/it] Prompt executed in 64.34 seconds

[Kosinkadink]

See if you get black images if you use a V2 or a V3 motion model on that same workflow

[abumolly]

hi, I changed the sampler to dpmpp_2m_sde, and down the denoise value to 0.3, it can generated some images. not sure why but seem problems solved...

denoise 0.5 will be ok, but 0.8, it will turn to black...