ð H3 Timeline 4 (v1.2): lip-synced music videos on your own song with MiniMax H3
A simple ComfyUI workflow that turns your own song and 1â4 pictures into a lip-synced music video with MiniMax H3. Play the song on a timeline and tap the keys where the scenes should change and where the singing starts and stops. No LLM, no YuE2, no score, and only one custom node folder to install.
âš Features
ð Timeline editor at the bottom of the ComfyUI page: drop your song and pictures, then play it back with the waveform
â Live cutting: press C while the song plays and each scene ends exactly where you press (up to 4 scenes, 1â15 s each)
ð€ Lip sync: each scene's piece of the song goes to H3 unchanged as
<Audio 1>. Press S where the singing starts and stops, and H3 sings only inside those spansðŒ First frame lock: each scene starts exactly on its picture (ComfyUI's Add Guide); switch it off per scene to use the picture only as a reference
â Continue the previous scene: one click at a cut makes the frame right after the previous scene's cut the next scene's first frame, so the shot carries on
ð§ Characters (up to 4): a picture and a name for each person. Choose who is in each scene and who sings (one character, or everyone); their pictures go to H3 as
<Picture 2>,<Picture 3>, ⊠so faces, hair and clothes stay the same. Call them by name in your text, and add a role if you like (e.g. lead vocalist)ð¥ Group photo: one photo of everyone together, so faces, heights and clothes match side by side. Used in the scenes where everyone appears (or in / out per scene)
ð Audio lock: each scene's piece of the song is anchored on H3's audio track in time while it generates (ComfyUI's own Add Guide), so the lips and the beat stay with the music. Per scene; the finished video keeps the original song
Transitions: cut, crossfade, dip to black or dip to white. The crossfade blends the previous scene while it is still moving
ð§© Redo one scene: change a scene's text, picture or timing (or press ð² Another take) and only that scene is made again; the others stay cached
Low memory: each scene is saved to disk as soon as it is made and joined at the end. The workflow needs about one scene's worth of RAM, not the whole video
Prompts are written for you: the
<Picture 1>/<Audio 1>definitions and the lip-sync line (with exact times) are added to your textðŸ Civitai metadata (optional): one switch saves the video with ComfyUI-Video-Saver, including the prompt H3 got for each used scene, model, seed and steps
Japanese and English workflows are included
â¶ How to use
Press ð Timeline at the bottom left of the canvas
Drop your song (wav, flac, mp3, âŠ) on the blue audio track
Drop a picture on each clip (+ adds scenes)
Press Space to play and C where each scene should end
Press S where the singing starts and S again where it stops. With no spans, H3 sings whenever it hears vocals
Write each scene's text below the timeline (English works best)
Press Run. The video is saved in
output/video/
âš Timeline controls
Space: play / stop · C: cut here · S: singing starts / stops
Drag a clip's right edge for its length (Shift = 0.1 s steps). Drag the waveform to choose where in the song the video starts
ð€ spans: drag the ends or the middle, and double-click to delete one. Double-clicking an empty spot adds one
Remove a scene's picture: â on the clip, or select the clip and press Delete over the timeline
â at a cut: continue the previous scene (leave that scene's picture empty for a seamless continuation) · â: transition · mouse wheel: zoom
+ð§ / +ð¥ at the top: add a character or the group photo (drop a picture, type a name and role) · below the timeline: ð§ In this scene and ð€ Who sings for the selected scene
Below the timeline: ð² Another take re-rolls only the selected scene (⺠= back to the shared seed) · ð: audio lock for that scene · ð§© Per scene at the top: redo only the scenes you changed (on by default)
ðŠ Requirements
A recent ComfyUI with the built-in MiniMax H3 nodes
models/diffusion_models/:minimax_h3_ref2va_pruned_int8_convrot.safetensorsmodels/loras/:minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors(Turbo, 4 steps)models/text_encoders/:qwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsmodels/vae/:minimax_h3_video_vae_fp16.safetensors,minimax_h3_audio_vae_fp32.safetensorsOptional: ComfyUI-Video-Saver, only to save with Civitai metadata. Without it the workflow runs as usual (ComfyUI may list its two nodes as missing; you can ignore that)
ð§ Install
Copy
custom_nodes/ComfyUI-H3Timelineinto ComfyUI'scustom_nodes/Restart ComfyUI and reload the page (Ctrl+F5)
Drop a workflow from
workflows/onto the canvas
ð Notes
One scene is at most 15 s (H3's limit)
The timeline preview shows the stills with the song. H3 makes the real motion
av_shifton the board delays the finished video's sound against the picture (default 0; about 0.2 s if the motion lags without ð)The temporary scene files (up to about 0.9 GB per scene) are kept in
temp/h3tl/and cleared every time ComfyUI starts. If one goes missing, it is decoded again automatically, and a full disk stops the run with a clear messageð§© Redoing one scene uses ComfyUI's normal cache (not
--cache-none); restarting ComfyUI forgets the scenes it remembered. The shared ð² seed is fixed: change it to re-roll every sceneThe ð attention node is for Nvidia / AMD GPUs. Set it to
pytorch attentionif it doesn't fit your GPU
ð H3 Timeline 4ïŒv1.2ïŒïŒææã¡ã®æ²ã§ MiniMax H3 ã®å£ã㯠MV
ææã¡ã®æ²ãš 1ã4 æã®ç»åãããMiniMax H3 ã§å£ãã¯ã®ãã¥ãŒãžãã¯ãããªãäœãã·ã³ãã«ãª ComfyUI ã¯ãŒã¯ãããŒã§ããã¿ã€ã ã©ã€ã³ã§æ²ãåçããªãããã·ãŒã³ãåãæãšæãæãããŒã§æ±ºããã ããLLMã»YuE2ã»æ¥œèã¯äœ¿ãããå ¥ããã«ã¹ã¿ã ããŒãã 1 ãã©ã«ãã ãã§ãã
âš ã§ããããš
ð ã¿ã€ã ã©ã€ã³ïŒComfyUI ã®ç»é¢ã®äžã«éãç·šéç»é¢ãæ²ãšç»åãããããããŠã波圢ãèŠãªããåçã§ããŸã
â åçããªããã«ããïŒæ²ãæµããªãã C ãæŒããšãæŒããæã§ã·ãŒã³ãåããŸãïŒæå€§ 4 ã·ãŒã³ã»1 ã·ãŒã³ 1ã15 ç§ïŒ
ð€ å£ãã¯ïŒåã·ãŒã³ã®åºéã®æ²ããã®ãŸãŸ H3 ã®
<Audio 1>ã«æž¡ããŸããæãå§ããšæãçµããã§ S ãæŒããšããã®éã ãæããŸãðŒ æåã®ã³ããç»åã«åºå®ïŒåã·ãŒã³ãç»åã©ããã«å§ãŸããŸãïŒæ¬äœã® Add GuideïŒãã·ãŒã³ããšã«ãªãã«ããã°ãç»åã¯åèã ãã«
â åã®ã·ãŒã³ããç¶ããïŒåãç®ãã¯ãªãã¯ããã ãã§ãåã®ã·ãŒã³ã®åãç®ã®çŽåŸã®ã³ããæ¬¡ã®ã·ãŒã³ã®æåã®ã³ãã«ãªããæ åãã€ãªãããŸã
ð§ ãã£ã©ïŒæå€§ 4 人ïŒïŒäººç©ããšã«ç»åãšååãç»é²ãã·ãŒã³ããšã«ãåºã人ããšãæã人ãïŒ1 人ïŒã¿ããªïŒãéžã¹ãŠãåºã人ã®ç»åã
<Picture 2>ã»<Picture 3>⊠ãšããŠæž¡ããé¡ã»é«ªã»æãããããŸããæç« ã®äžã§ã¯ååã§åŒã¹ãŠã圹ã»èª¬æãä»ããããŸãïŒäŸïŒlead vocalistïŒð¥ å šäœåçïŒå šå¡äžç·ã®åçã 1 ææž¡ããŠã䞊ãã æã®é¡ã»èäžã»æãããããŸããå šå¡ãåºãã·ãŒã³ã§äœ¿ããŸãïŒã·ãŒã³ããšã«å ¥ããïŒå€ããå¯ïŒ
ð æ²ãåºå®ïŒåã·ãŒã³ã®åºéã®æ²ããçæäžã« H3 ã®é³å£°ãã©ãã¯ãžæéã©ããåºå®ïŒæ¬äœã® Add GuideïŒãå£ãããŒããæ²ããããã«ãããªããŸããã·ãŒã³ããšã«åæ¿ã宿åç»ã®é³ã¯å ã®æ²ã®ãŸãŸ
ã€ãªãæ¹ïŒcutïŒcrossfadeïŒdip to blackïŒdip to whiteãcrossfade ã¯åã®ã·ãŒã³ãåãããŸãŸéãªããŸã
ð§© ã·ãŒã³ã ãäœãçŽãïŒã·ãŒã³ã®æç« ã»ç»åã»ã¿ã€ãã³ã°ãå€ããïŒãŸã㯠ð² å¥ãã€ã¯ïŒãšããã®ã·ãŒã³ã ãäœãçŽããã»ãã®ã·ãŒã³ã¯ãã£ãã·ã¥ã®ãŸãŸ
çã¡ã¢ãªïŒåã·ãŒã³ã¯ã§ããããããã£ã¹ã¯ã«ä¿åããŠãæåŸã«ã€ãªããŸããã¡ã¢ãªã¯åç»å šäœã§ã¯ãªãã1 ã·ãŒã³ã¶ããã§æžã¿ãŸã
ããã³ããã¯èªåã§æŽããŸãïŒ
<Picture 1>ã»<Audio 1>ã®å®çŸ©ãšãæå»å ¥ãã®å£ãã¯ã®æç€ºãæžããæç« ã«è¶³ããŸãðŸ Civitai ã¡ã¿ããŒã¿ïŒä»»æïŒïŒã¹ã€ãã 1 ã€ã§ãComfyUI-Video-Saver ã䜿ã£ãŠã䜿ã£ãã·ãŒã³ã¶ãã® H3 ã®ããã³ããã»ã¢ãã«ã»seedã»steps ä»ãã§ä¿å
æ¥æ¬èªçã»è±èªçã®ã¯ãŒã¯ãããŒå ¥ã
â¶ äœ¿ãæ¹
ãã£ã³ãã¹å·Šäžã® ð ã¿ã€ã ã©ã€ã³ ãæŒã
æ²ïŒwavã»flacã»mp3 ãªã©ïŒãéãæ®µã«ãããã
åã¯ãªããã« ç»å ãããããïŒïŒ ã§ã·ãŒã³ã远å ïŒ
Space ã§åçããŠãã·ãŒã³ãåãããæã§ C
æãå§ãã§ Sãæãçµããã§ããäžåºŠ SïŒä»ããªããã°ã声ãèããããæãïŒ
ã¿ã€ã ã©ã€ã³ã®äžã®æ¬ã«ãåã·ãŒã³ã®æç« ãæžãïŒè±èªãããããïŒ
å®è¡ã
output/video/ã«ä¿åãããŸã
âš ã¿ã€ã ã©ã€ã³ã®æäœ
SpaceïŒåçïŒåæ¢ã»CïŒããã§åãã»SïŒæãå§ãïŒæãçµãã
ã¯ãªããã®å³ç«¯ããã©ãã°ã§ç§æ°ïŒShift ã§ 0.1 ç§å»ã¿ïŒã波圢ãå·Šå³ã«ãã©ãã°ã§æ²ã®ã©ããã䜿ãã
ð€ 垯ïŒç«¯ãçãäžããã©ãã°ãããã«ã¯ãªãã¯ã§åé€ïŒç©ºããæã®ããã«ã¯ãªãã¯ã§è¿œå ïŒ
ã·ãŒã³ã®ç»åãå€ãïŒã¯ãªããã® âããŸãã¯ã¯ãªãããéžãã§ã¿ã€ã ã©ã€ã³ã®äžã§ Delete
åãç®ã® âïŒåã®ã·ãŒã³ããç¶ããïŒãããã«ç¶ãããæã¯ããã®ã·ãŒã³ã®ç»åã¯ç©ºã®ãŸãŸïŒã»âïŒã€ãªãæ¹ã»ãã€ãŒã«ïŒæ¡å€§çž®å°
äžã® ïŒð§ïŒïŒð¥ïŒãã£ã©ãå šäœåçã远å ïŒç»åãããããããŠååã»åœ¹ãå ¥åïŒã»äžã®æ¬ïŒéžãã ã·ãŒã³ã® ð§ åºã人 ãš ð€ æã人
äžã®æ¬ã® ð² å¥ãã€ã¯ïŒéžãã ã·ãŒã³ã ãåŒãçŽãïŒâº ã§å ±éã® seed ã«æ»ãïŒã»ðïŒãã®ã·ãŒã³ã®æ²ã®åºå®ã»äžã® ð§© ã·ãŒã³åäœïŒå€ããã·ãŒã³ã ãäœãçŽãïŒæšæºã§ãªã³ïŒ
ðŠ å¿ èŠãªãã®
MiniMax H3 ã®ããŒããæšæºã§å ¥ã£ãŠãããæ°ãã ComfyUI
models/diffusion_models/ïŒminimax_h3_ref2va_pruned_int8_convrot.safetensorsmodels/loras/ïŒminimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensorsïŒTurboã»4 stepïŒmodels/text_encoders/ïŒqwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsmodels/vae/ïŒminimax_h3_video_vae_fp16.safetensorsã»minimax_h3_audio_vae_fp32.safetensorsä»»æïŒComfyUI-Video-SaverïŒCivitai ã¡ã¿ããŒã¿ä»ãã§ä¿åããæã ãïŒãç¡ããŠãæ®éã«åããŸãïŒãã® 2 ã€ã®ããŒãããèŠã€ãããªãããšåºãããšããããŸãããç¡èŠããŠOKïŒ
ð§ å ¥ãæ¹
custom_nodes/ComfyUI-H3Timelineã ComfyUI ã®custom_nodes/ã«ã³ããŒComfyUI ãåèµ·åããŠãããŒãžãåèªã¿èŸŒã¿ïŒCtrl+F5ïŒ
workflows/ã®ã¯ãŒã¯ãããŒããã£ã³ãã¹ã«ãããã
ð æ°ãã€ããããš
1 ã·ãŒã³ã¯ 15 ç§ãŸã§ïŒH3 ã®äžéïŒ
ã¿ã€ã ã©ã€ã³ã®åçã¯ã鿢ç»ãšæ²ã§ã®ç®å®ã§ãïŒå®éã®åã㯠H3 ãäœããŸãïŒ
ããŒãã®
av_shiftã§å®æåç»ã®é³ãæ åã«å¯ŸããŠé ãããããŸãïŒæšæº 0ãð ãªãã§åããé ãããªã 0.2 ç§ãããïŒéäžã®ãã¡ã€ã«ïŒ1 ã·ãŒã³ãããæå€§çŽ 0.9GBïŒã¯
temp/h3tl/ã«çœ®ããComfyUI ãèµ·åãçŽããšæ¶ããŸããæ¶ããŠããæã¯èªåã§ãã³ãŒããçŽãã空ã容éãè¶³ããªãæã¯åãããããã¡ãã»ãŒãžã§æ¢ãŸããŸãð§© ã·ãŒã³ã ãäœãçŽãã®ã¯ ComfyUI ã®æ®éã®ãã£ãã·ã¥ã䜿ããŸãïŒ
--cache-noneã§ã¯å¹ããŸããïŒãComfyUI ãåèµ·åãããšèŠããŠããã·ãŒã³ã¯æ¶ããŸããå ±éã® ð² seed ã¯åºå®ãªã®ã§ãå šã·ãŒã³ãåŒãçŽãæã¯ seed ãå€ããŠãð attention ããŒã㯠NvidiaïŒAMD çšã§ããåããªãæã¯
pytorch attentionã«ããŠãã ãã
Description
ð H3 Timeline 4 (v1.1): lip-synced music videos on your own song with MiniMax H3
A simple ComfyUI workflow that turns your own song and 1â4 pictures into a lip-synced music video with MiniMax H3. Play the song on a timeline and tap the keys where the scenes should change and where the singing starts and stops. No LLM, no YuE2, no score, and only one custom node folder to install.
âš Features
ð Timeline editor at the bottom of the ComfyUI page: drop your song and pictures, then play it back with the waveform
â Live cutting: press C while the song plays and each scene ends exactly where you press (up to 4 scenes, 1â15 s each)
ð€ Lip sync: each scene's piece of the song goes to H3 unchanged as
<Audio 1>. Press S where the singing starts and stops, and H3 sings only inside those spansâ Continue the previous scene: one click at a cut makes the frame right after the previous scene's cut the next scene's first frame, so the shot carries on
ð§ Characters (up to 4): a picture and a name for each person. Choose who is in each scene and who sings (one character, or everyone); their pictures go to H3 as
<Picture 2>,<Picture 3>, ⊠so faces, hair and clothes stay the same. Call them by name in your text, and add a role if you like (e.g. lead vocalist)ð¥ Group photo: one photo of everyone together, so faces, heights and clothes match side by side. Used in the scenes where everyone appears (or in / out per scene)
ð Audio lock: each scene's piece of the song is anchored on H3's audio track in time while it generates (ComfyUI's own Add Guide), so the lips and the beat stay with the music. Per scene; the finished video keeps the original song
Transitions: cut, crossfade, dip to black or dip to white. The crossfade blends the previous scene while it is still moving
Low memory: each scene is saved to disk as soon as it is made and joined at the end. The workflow needs about one scene's worth of RAM, not the whole video
Prompts are written for you: the
<Picture 1>/<Audio 1>definitions and the lip-sync line (with exact times) are added to your textðŸ Civitai metadata (optional): one switch saves the video with ComfyUI-Video-Saver, including the prompt H3 got for each used scene, model, seed and steps
Japanese and English workflows are included
â¶ How to use
Press ð Timeline at the bottom left of the canvas
Drop your song (wav, flac, mp3, âŠ) on the blue audio track
Drop a picture on each clip (+ adds scenes)
Press Space to play and C where each scene should end
Press S where the singing starts and S again where it stops. With no spans, H3 sings whenever it hears vocals
Write each scene's text below the timeline (English works best)
Press Run. The video is saved in
output/video/
âš Timeline controls
Space: play / stop · C: cut here · S: singing starts / stops
Drag a clip's right edge for its length (Shift = 0.1 s steps). Drag the waveform to choose where in the song the video starts
ð€ spans: drag the ends or the middle, and double-click to delete one. Double-clicking an empty spot adds one
Remove a scene's picture: â on the clip, or select the clip and press Delete over the timeline
â at a cut: continue the previous scene (leave that scene's picture empty for a seamless continuation) · â: transition · mouse wheel: zoom
+ð§ / +ð¥ at the top: add a character or the group photo (drop a picture, type a name and role) · below the timeline: ð§ In this scene and ð€ Who sings for the selected scene
ðŠ Requirements
A recent ComfyUI with the built-in MiniMax H3 nodes
models/diffusion_models/:minimax_h3_ref2va_pruned_int8_convrot.safetensorsmodels/loras/:minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors(Turbo, 4 steps)models/text_encoders/:qwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsmodels/vae/:minimax_h3_video_vae_fp16.safetensors,minimax_h3_audio_vae_fp32.safetensorsOptional: ComfyUI-Video-Saver, only to save with Civitai metadata. Without it the workflow runs as usual (ComfyUI may list its two nodes as missing; you can ignore that)
ð§ Install
Copy
custom_nodes/ComfyUI-H3Timelineinto ComfyUI'scustom_nodes/Restart ComfyUI and reload the page (Ctrl+F5)
Drop a workflow from
workflows/onto the canvas
ð Notes
One scene is at most 15 s (H3's limit)
The timeline preview shows the stills with the song. H3 makes the real motion
The finished video's sound plays 0.2 s after the picture, because H3's motion lags the sound slightly. You can change this with
av_shifton the boardThe temporary scene files (up to about 0.9 GB per scene) are kept in
temp/h3tl/and cleared every time ComfyUI starts. If one goes missing, it is decoded again automatically, and a full disk stops the run with a clear messageThe ð attention node is for Nvidia / AMD GPUs. Set it to
pytorch attentionif it doesn't fit your GPU
ð H3 Timeline 4ïŒv1.1ïŒïŒææã¡ã®æ²ã§ MiniMax H3 ã®å£ã㯠MV
ææã¡ã®æ²ãš 1ã4 æã®ç»åãããMiniMax H3 ã§å£ãã¯ã®ãã¥ãŒãžãã¯ãããªãäœãã·ã³ãã«ãª ComfyUI ã¯ãŒã¯ãããŒã§ããã¿ã€ã ã©ã€ã³ã§æ²ãåçããªãããã·ãŒã³ãåãæãšæãæãããŒã§æ±ºããã ããLLMã»YuE2ã»æ¥œèã¯äœ¿ãããå ¥ããã«ã¹ã¿ã ããŒãã 1 ãã©ã«ãã ãã§ãã
âš ã§ããããš
ð ã¿ã€ã ã©ã€ã³ïŒComfyUI ã®ç»é¢ã®äžã«éãç·šéç»é¢ãæ²ãšç»åãããããããŠã波圢ãèŠãªããåçã§ããŸã
â åçããªããã«ããïŒæ²ãæµããªãã C ãæŒããšãæŒããæã§ã·ãŒã³ãåããŸãïŒæå€§ 4 ã·ãŒã³ã»1 ã·ãŒã³ 1ã15 ç§ïŒ
ð€ å£ãã¯ïŒåã·ãŒã³ã®åºéã®æ²ããã®ãŸãŸ H3 ã®
<Audio 1>ã«æž¡ããŸããæãå§ããšæãçµããã§ S ãæŒããšããã®éã ãæããŸãâ åã®ã·ãŒã³ããç¶ããïŒåãç®ãã¯ãªãã¯ããã ãã§ãåã®ã·ãŒã³ã®åãç®ã®çŽåŸã®ã³ããæ¬¡ã®ã·ãŒã³ã®æåã®ã³ãã«ãªããæ åãã€ãªãããŸã
ð§ ãã£ã©ïŒæå€§ 4 人ïŒïŒäººç©ããšã«ç»åãšååãç»é²ãã·ãŒã³ããšã«ãåºã人ããšãæã人ãïŒ1 人ïŒã¿ããªïŒãéžã¹ãŠãåºã人ã®ç»åã
<Picture 2>ã»<Picture 3>⊠ãšããŠæž¡ããé¡ã»é«ªã»æãããããŸããæç« ã®äžã§ã¯ååã§åŒã¹ãŠã圹ã»èª¬æãä»ããããŸãïŒäŸïŒlead vocalistïŒð¥ å šäœåçïŒå šå¡äžç·ã®åçã 1 ææž¡ããŠã䞊ãã æã®é¡ã»èäžã»æãããããŸããå šå¡ãåºãã·ãŒã³ã§äœ¿ããŸãïŒã·ãŒã³ããšã«å ¥ããïŒå€ããå¯ïŒ
ð æ²ãåºå®ïŒåã·ãŒã³ã®åºéã®æ²ããçæäžã« H3 ã®é³å£°ãã©ãã¯ãžæéã©ããåºå®ïŒæ¬äœã® Add GuideïŒãå£ãããŒããæ²ããããã«ãããªããŸããã·ãŒã³ããšã«åæ¿ã宿åç»ã®é³ã¯å ã®æ²ã®ãŸãŸ
ã€ãªãæ¹ïŒcutïŒcrossfadeïŒdip to blackïŒdip to whiteãcrossfade ã¯åã®ã·ãŒã³ãåãããŸãŸéãªããŸã
çã¡ã¢ãªïŒåã·ãŒã³ã¯ã§ããããããã£ã¹ã¯ã«ä¿åããŠãæåŸã«ã€ãªããŸããã¡ã¢ãªã¯åç»å šäœã§ã¯ãªãã1 ã·ãŒã³ã¶ããã§æžã¿ãŸã
ããã³ããã¯èªåã§æŽããŸãïŒ
<Picture 1>ã»<Audio 1>ã®å®çŸ©ãšãæå»å ¥ãã®å£ãã¯ã®æç€ºãæžããæç« ã«è¶³ããŸãðŸ Civitai ã¡ã¿ããŒã¿ïŒä»»æïŒïŒã¹ã€ãã 1 ã€ã§ãComfyUI-Video-Saver ã䜿ã£ãŠã䜿ã£ãã·ãŒã³ã¶ãã® H3 ã®ããã³ããã»ã¢ãã«ã»seedã»steps ä»ãã§ä¿å
æ¥æ¬èªçã»è±èªçã®ã¯ãŒã¯ãããŒå ¥ã
â¶ äœ¿ãæ¹
ãã£ã³ãã¹å·Šäžã® ð ã¿ã€ã ã©ã€ã³ ãæŒã
æ²ïŒwavã»flacã»mp3 ãªã©ïŒãéãæ®µã«ãããã
åã¯ãªããã« ç»å ãããããïŒïŒ ã§ã·ãŒã³ã远å ïŒ
Space ã§åçããŠãã·ãŒã³ãåãããæã§ C
æãå§ãã§ Sãæãçµããã§ããäžåºŠ SïŒä»ããªããã°ã声ãèããããæãïŒ
ã¿ã€ã ã©ã€ã³ã®äžã®æ¬ã«ãåã·ãŒã³ã®æç« ãæžãïŒè±èªãããããïŒ
å®è¡ã
output/video/ã«ä¿åãããŸã
âš ã¿ã€ã ã©ã€ã³ã®æäœ
SpaceïŒåçïŒåæ¢ã»CïŒããã§åãã»SïŒæãå§ãïŒæãçµãã
ã¯ãªããã®å³ç«¯ããã©ãã°ã§ç§æ°ïŒShift ã§ 0.1 ç§å»ã¿ïŒã波圢ãå·Šå³ã«ãã©ãã°ã§æ²ã®ã©ããã䜿ãã
ð€ 垯ïŒç«¯ãçãäžããã©ãã°ãããã«ã¯ãªãã¯ã§åé€ïŒç©ºããæã®ããã«ã¯ãªãã¯ã§è¿œå ïŒ
ã·ãŒã³ã®ç»åãå€ãïŒã¯ãªããã® âããŸãã¯ã¯ãªãããéžãã§ã¿ã€ã ã©ã€ã³ã®äžã§ Delete
åãç®ã® âïŒåã®ã·ãŒã³ããç¶ããïŒãããã«ç¶ãããæã¯ããã®ã·ãŒã³ã®ç»åã¯ç©ºã®ãŸãŸïŒã»âïŒã€ãªãæ¹ã»ãã€ãŒã«ïŒæ¡å€§çž®å°
äžã® ïŒð§ïŒïŒð¥ïŒãã£ã©ãå šäœåçã远å ïŒç»åãããããããŠååã»åœ¹ãå ¥åïŒã»äžã®æ¬ïŒéžãã ã·ãŒã³ã® ð§ åºã人 ãš ð€ æã人
ðŠ å¿ èŠãªãã®
MiniMax H3 ã®ããŒããæšæºã§å ¥ã£ãŠãããæ°ãã ComfyUI
models/diffusion_models/ïŒminimax_h3_ref2va_pruned_int8_convrot.safetensorsmodels/loras/ïŒminimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensorsïŒTurboã»4 stepïŒmodels/text_encoders/ïŒqwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsmodels/vae/ïŒminimax_h3_video_vae_fp16.safetensorsã»minimax_h3_audio_vae_fp32.safetensorsä»»æïŒComfyUI-Video-SaverïŒCivitai ã¡ã¿ããŒã¿ä»ãã§ä¿åããæã ãïŒãç¡ããŠãæ®éã«åããŸãïŒãã® 2 ã€ã®ããŒãããèŠã€ãããªãããšåºãããšããããŸãããç¡èŠããŠOKïŒ
ð§ å ¥ãæ¹
custom_nodes/ComfyUI-H3Timelineã ComfyUI ã®custom_nodes/ã«ã³ããŒComfyUI ãåèµ·åããŠãããŒãžãåèªã¿èŸŒã¿ïŒCtrl+F5ïŒ
workflows/ã®ã¯ãŒã¯ãããŒããã£ã³ãã¹ã«ãããã
ð æ°ãã€ããããš
1 ã·ãŒã³ã¯ 15 ç§ãŸã§ïŒH3 ã®äžéïŒ
ã¿ã€ã ã©ã€ã³ã®åçã¯ã鿢ç»ãšæ²ã§ã®ç®å®ã§ãïŒå®éã®åã㯠H3 ãäœããŸãïŒ
宿åç»ã®é³ã¯æ åãã 0.2 ç§é ãããŠããŸãïŒH3 ã®åãã¯é³ããå°ãé ãããããããŒãã®
av_shiftã§å€ããããŸãïŒéäžã®ãã¡ã€ã«ïŒ1 ã·ãŒã³ãããæå€§çŽ 0.9GBïŒã¯
temp/h3tl/ã«çœ®ããComfyUI ãèµ·åãçŽããšæ¶ããŸããæ¶ããŠããæã¯èªåã§ãã³ãŒããçŽãã空ã容éãè¶³ããªãæã¯åãããããã¡ãã»ãŒãžã§æ¢ãŸããŸãð attention ããŒã㯠NvidiaïŒAMD çšã§ããåããªãæã¯
pytorch attentionã«ããŠãã ãã

