Credits
el_pato_pato's AnimateDiff LCM+SD WF
Outputs
Description
Simple combine of SD1.5 and SDXL for AnimateDiff. Take about 50 min to create 64 frames (128 with interpolation) on RTX 4060TI. The time can be reduced to 15 - 20 min if used LCM or Turbo model for the SDXL pass, but the quality is not the same. The background video is made from generated hires image and zoomed out in video editing software. The workflow can certainly be improved, but the results are promising.
Built-in nodes
VAEDecode
CLIPTextEncode
EmptyLatentImage
KSampler
CheckpointLoaderSimple
ModelSamplingDiscrete
LoraLoader
VAELoader
ImageCompositeMasked
CLIPVisionLoader
VAEEncode
CLIPSetLastLayer
CLIPTextEncodeSDXL
ControlNetApplyAdvanced
PreviewImage
LoadImage
Custom nodes
VHS_VideoCombine
UltralyticsDetectorProvider
ImpactImageBatchToImageList
SAMLoader
FILM VFI
Mask Invert
ImpactSimpleDetectorSEGS
CR Prompt Text
ADE_AnimateDiffLoaderWithContext
ControlNetLoaderAdvanced
ImageResize+
INTConstant
ADE_LoopedUniformContextOptions
CR Upscale Image
IPAdapterModelLoader
PrepImageForClipVision
DWPreprocessor
CR Seed
ModelMergeBlocks
ColorMatch
VHS_LoadVideo
LineartStandardPreprocessor
VHS_LoadImages
ImpactSEGSToMaskList
MaskListToMaskBatch
Custom files
new
These are separate files that the creator has uploaded for this workflow.
AD_Workflow_Images.zip
Models
person_yolov8m-seg.pt
sam_vit_b_01ec64.pth
film_net_fp32.pt
v3_sd15_mm.ckpt
v3_sd15_adapter.ckpt
vae-ft-mse-840000-ema-pruned.safetensors
4x-UltraSharp.pth
ip-adapter_sd15_light.safetensors
CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors
yolox_l.torchscript.pt
dw-ll_ucoco_384_bs5.torchscript.pt
add-detail-xl.safetensors
hotshotxl_mm_v1.pth
control_v11p_sd15_lineart_fp16.safetensors
lunaMia_v50bakedvae.safetensors