skyreels-hunyuan-I2V-workflow — v1.0

Use skyreels-hunyuan-I2V-workflow — v1.0 in ComfyUI to create a video from a selected image and text prompt.

Named components include LoadImage, CLIPTextEncode, EmptyHunyuanLatentVideo, SamplerCustom, VAEDecodeTiled, and VHS_VideoCombine.

It is listed as a Workflows item from AbdallahAlswa80, using Flux.1 D as its base model.

Use LoadImage for the image input and CLIPTextEncode for text input; VHS_VideoCombine is named for combining video.

Update ComfyUI to the latest version before use.

Suggestion · non vérifié

Before use, check that these model files are available: clip_l.safetensors, e5m2-skyreels_hunyuan_i2v_bf16.safetensors, hunyuan_video_vae_bf16.safetensors, llava_llama3_fp8_scaled.safetensors, and skyreels-hunyuan-I2V-Q4_K_S.gguf.

Suggestion · non vérifié

Check for these named nodes: ApplyFirstBlockCachePatch, BasicScheduler, CLIPTextEncode, ColorMatch, ConditioningZeroOut, DualCLIPLoader, EmptyHunyuanLatentVideo, FluxGuidance, ImageNoiseAugmentation, ImageResize+, and InstructPixToPixConditioning.

Suggestion · non vérifié

Also check for KSamplerSelect, LoadImage, ModelSamplingSD3, PatchModelPatcherOrder, SamplerCustom, SplitSigmas, UNETLoader, UnetLoaderGGUF, VAEDecodeTiled, VAELoader, and VHS_VideoCombine.

Suggestion · non vérifié

Start each prompt with FPS-24.

Suggestion · non vérifié

Choose the image carefully, and keep the prompt short and close to the image.

Suggestion · non vérifié

To use LoRAs, enable the node shown for that option.

Suggestion · non vérifié

No license is listed for this workflow.

Do not use GPTs with this workflow.

Suggestion · non vérifié