(Video Tutorial Resources) Kontext, Speed Boosts and updated Loaders — v1.0

This Flux.1 Kontext workflow is for image and text-related processing: it names an image input, text-related inputs, and image-upscaling, preview, and video-combining outputs through LoadImage, CLIPTextEncode, FramePackTimestampedTextEncode, ImageUpscaleWithModel, PreviewImage,よび

The named processing stages include LoadImage, CLIPTextEncode, FramePackTimestampedTextEncode, ImageUpscaleWithModel, PreviewImage, and VHS_VideoCombine.

It uses Flux.1 Kontext and names flux1-dev-kontext_fp8_scaled.safetensors.

LoadImage marks the image input; CLIPTextEncode and FramePackTimestampedTextEncode mark text-related inputs.

ImageUpscaleWithModel, PreviewImage, and VHS_VideoCombine mark the named output stages.

The setup notes specify FP8 (or GGUF) model files under models/checkpoints or unet, Flux VAE under models/vae, and T5 Clip - FP16 under models/clip.

The notes also list NAG (ComfyManager #2706) and Hyper Chroma Speed Booster LORA.

Named model files include: 1x-ITF-SkinDiffDetail-Lite-v1.pth; bbox/face_yolov8m.pt; bj-w4-000007.safetensors; chroma-unlocked-v40.safetensors; clip_l.safetensors; flux1-dev-kontext_fp8_scaled.safetensors; flux1-kontext-dev_fp8_scaled_stochastic-marduk191.safetensors; FramePack_F1

Named nodes include: EmptySD3LatentImage; CLIPVisionLoader; ReferenceLatent; ConditioningZeroOut; ImageScaleBy; Cfg Literal; Image Power Noise; KSampler; VAEDecodeTiled; KSamplerSelect; UpscaleModelLoader; Textbox; Image Perlin Noise; CR Latent Input Switch; JWIntegerSub; mx和

Additional named nodes include: Power Lora Loader (rgthree); FluxGuidance; T5TokenizerOptions; DualCLIPLoader; ImageStitch; PathchSageAttentionKJ; FramePackSampler_F1; GetNode; mxSlider2D; VHS_VideoCombine; KSamplerAdvanced; LatentOperationSharpen; DiffusionModelLoaderKJ; Divide者

The notes cover Kontext Introduction, Kontext Setup, Kontext Speed Boosts, and Updated Flux and Chroma Loaders with Kontext.

Before using the workflow, verify that these named nodes are available in your setup: EmptySD3LatentImage; CLIPVisionLoader; ReferenceLatent; ConditioningZeroOut; ImageScaleBy; KSampler.

Suggestion · not verified

Also verify these named files under their exact names: t5xxl_fp16.safetensors; 1x-ITF-SkinDiffDetail-Lite-v1.pth; flux1-kontext-dev_fp8_scaled_stochastic-marduk191.safetensors; hunyuan_video_vae_bf16.safetensors; sigclip_vision_patch14_384.safetensors.

Suggestion · not verified