floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.
floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.

Vertical Video Lighting & Mood Shift Using Seedream + Wan

235

Generates in about 26 secs

Nodes & Models

SeedreamV4Edit_floyo
Label (rgthree)
GetNode
Note
PrimitiveInt
UpscaleModelLoader
4x-UltraSharp.pth
LoadImage
CLIPVisionLoader
clip_vision_h.safetensors
LoadAudio
Fast Groups Bypasser (rgthree)
PreviewImage
SetNode
ImageScaleBy
ImageUpscaleWithModel
ImageBlend
ImageScale
SaveImage
PreviewAny
WanVideoLoraSelect
Wan/wan_i2v_720_l1v3w4llp4p3r_e50_with_trigger.safetensors
lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors
LoadWanVideoClipTextEncoder
open-clip-xlm-roberta-large-vit-huge-14_fp16.safetensors
WanVideoVAELoader
wan_2.1_vae.safetensors
Wan2_1_VAE_bf16.safetensors
WanVideoModelLoader
Wan2.1-Fun-Control-14B_fp8_e4m3fn.safetensors
Wan2_1-I2V-14B-480P_fp8_e4m3fn.safetensors
LoadWanVideoT5TextEncoder
umt5_xxl_fp16.safetensors
WanVideoTorchCompileSettings
WanVideoBlockSwap
WanVideoLoraSelect
Wan/wan_i2v_720_l1v3w4llp4p3r_e50_with_trigger.safetensors
lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors
LoadWanVideoClipTextEncoder
open-clip-xlm-roberta-large-vit-huge-14_fp16.safetensors
WanVideoVAELoader
wan_2.1_vae.safetensors
Wan2_1_VAE_bf16.safetensors
WanVideoModelLoader
Wan2.1-Fun-Control-14B_fp8_e4m3fn.safetensors
Wan2_1-I2V-14B-480P_fp8_e4m3fn.safetensors
LoadWanVideoT5TextEncoder
umt5_xxl_fp16.safetensors
WanVideoSLG
WanVideoTeaCache
DownloadAndLoadWav2VecModel
WanVideoTorchCompileSettings
WanVideoBlockSwap
WanVideoTextEncodeCached
umt5-xxl-enc-bf16.safetensors
MultiTalkModelLoader
Wan2_1-InfiniteTalk-Multi_fp8_e4m3fn_scaled_KJ.safetensors
WanVideoTextEncode
MultiTalkWav2VecEmbeds
WanVideoEncode
WanVideoControlEmbeds
WanVideoClipVisionEncode
WanVideoImageToVideoEncode
WanVideoImageToVideoMultiTalk
WanVideoSampler
WanVideoDecode
easy mathFloat
easy float
easy showAnything
VHS_BatchManager
VHS_LoadVideo
VHS_VideoInfo
VHS_VideoCombine
DownloadAndLoadWav2VecModel
MultiTalkModelLoader
Wan2_1-InfiniteTalk-Multi_fp8_e4m3fn_scaled_KJ.safetensors
MultiTalkWav2VecEmbeds
WanVideoImageToVideoMultiTalk
DownloadAndLoadWav2VecModel
ImageResizeKJ
ImageResizeKJ
AddLabel
ImageResizeKJv2
GetImageRangeFromBatch
GetImageSizeAndCount
ImageConcatMulti
ImageResizeKJ
AddLabel
ImageResizeKJv2
GetImageRangeFromBatch
GetImageSizeAndCount
ImageConcatMulti
MiDaS-DepthMapPreprocessor
Scribble_XDoG_Preprocessor
DWPreprocessor
MiDaS-DepthMapPreprocessor
Scribble_XDoG_Preprocessor
DWPreprocessor
DWPreprocessor
WanVideoEncode
ImageResize+
RIFE VFI
rife47.pth
FinalFrameSelector
FinalFrameSelector

Mood & Lighting Rework – Simple Explanation

This workflow transforms your original video by enhancing its mood, lighting, and emotional tone—without reshoots or complex color-grading tools.
Every original motion, expression, camera movement, and performance remains untouched while the scene’s visual atmosphere is dramatically upgraded.


How It Works (Step-by-Step)

1. Edit a Key Frame with Seedream

Choose one important frame from your video and open it in Seedream.

Here, you adjust lighting, shadows, color tones, glow, atmosphere, or overall mood—similar to doing high-end color grading on a single still.

This single edited frame becomes the reference blueprint for the entire video.


2. Mask or Select Areas (Optional)

If only certain parts of the scene need mood changes—like facial highlights, background glow, rim lighting, haze, or ambience—you can mask those areas on the first frame.

This ensures the mood adjustments apply exactly where you want them.


3. Describe the Mood & Lighting in Your Prompt

Write a simple mood/lighting prompt such as:

  • “soft warm cinematic lighting across the face”

  • “dramatic backlight with golden glow”

  • “rich blue nighttime mood with soft contrast”

  • “warm sunset ambience with soft haze”

Seedream edits the frame accordingly while keeping identity, expressions, and details intact.


4. Send the Edited Frame to Wan 2.1 FunControl

Wan compares:

  • your Seedream-edited frame

  • your original unedited video

Wan 2.1 FunControl understands exactly what changed in the frame, then tracks those changes through the entire video.

It locks your new lighting/mood to:

  • real motion

  • natural shadows

  • camera movement

  • scene depth

  • facial expressions

This makes the scene feel like it was lit that way originally.


5. Upload Your Original Video

Wan analyzes the full video frame-by-frame to understand:

  • motion

  • lighting variations

  • camera angles

  • facial expressions

This ensures stable, flicker-free results.


6. Activate Wan 2.1 FunControl

Wan generates a new version of the video where your lighting and mood adjustments are applied naturally while preserving:

  • all performances

  • gestures

  • timing

  • expression accuracy

  • camera movement

Only the mood/lighting changes—nothing else is altered.


7. LipSync Pass

The video then goes through a LipSync workflow.

This keeps:

  • mouth movements

  • dialogue sync

  • micro-expressions

  • timing

all stable and correct—especially important for talking shots.


8. Final Upscaling

The final video is upscaled to enhance:

  • sharpness

  • texture clarity

  • skin detail

  • lighting transitions

  • overall resolution

The result looks clean, polished, and professional.


9. Download Your Final Video

You receive a final MP4 where:

  • the mood and lighting feel natural and cinematic

  • the actor’s appearance stays the same

  • all motion is preserved

  • lipsync is accurate

  • the shot looks intentionally lit and professionally graded


Behind the Scenes (Simple Terms)

You edit one frame in Seedream.
Wan tracks your mood/lighting edits through the entire video using the original motion and lighting logic.
LipSync refines facial accuracy.
Upscaling adds final sharpness and detail.


Final Result

A beautifully re-lit, mood-enhanced video that keeps every natural movement, expression, and performance intact—without any traditional VFX software, color grading tools, or manual editing.

Read more

N