Skip to content

armgit / hypit

Public Imported from GitHub
0

Clone any viral video with AI agents. Not just a script, the whole workflow: swap the face, the words, the B-roll, ship 100 variants in one command, and get your 100M views.

fix: stop the examples teaching a take with no picture

rponebaf63f6c849e
diff --git a/.agents/skills/hypit/references/playbooks/craft/production-gates.md b/.agents/skills/hypit/references/playbooks/craft/production-gates.md
index 96c45e29..ca4c48e3 100644
--- a/.agents/skills/hypit/references/playbooks/craft/production-gates.md
+++ b/.agents/skills/hypit/references/playbooks/craft/production-gates.md
@@ -46,7 +46,8 @@ visual QA.
## Gate 2: generate and review video takes
- Demand only the takes whose reference images have passed Gate 1.
-- Keep Seedance duration inside the declared 4–15 second range; invalid values fail validation rather
+- Keep Seedance duration inside the selected model's declared range, read from that model rather than
+ from memory: the models differ and one accepts far longer takes. Invalid values fail validation rather
than being silently clamped.
- Review each take before assembly: verify the first-frame identity, reference continuity, lip-sync
when applicable, physical motion, stable props/text, intended camera/edit language, and absence of
diff --git a/.agents/skills/hypit/references/playbooks/formats/asmr.md b/.agents/skills/hypit/references/playbooks/formats/asmr.md
index adc4198c..0d8c2edd 100644
--- a/.agents/skills/hypit/references/playbooks/formats/asmr.md
+++ b/.agents/skills/hypit/references/playbooks/formats/asmr.md
@@ -14,7 +14,8 @@ texture sound to create a calm, satisfying micro-film.
place, or another small event that can complete naturally in one take.
4. Use `seedance:FrameVideo` for one controlling first frame or `seedance:ReferenceVideo` for
multiple ordered material references. Set `generate-audio="false"`.
-5. Keep raw Seedance duration at an integer 4–15 seconds. Use explicit media Trim/Retime only after
+5. Keep raw Seedance duration an integer inside the selected model's declared range. Use explicit
+ media Trim/Retime only after
reviewing the generated take.
Avoid rapid edits, large camera moves, busy backgrounds, multiple simultaneous actions, or motion
diff --git a/.agents/skills/hypit/references/playbooks/formats/talking-head.md b/.agents/skills/hypit/references/playbooks/formats/talking-head.md
index c5d348be..d648a658 100644
--- a/.agents/skills/hypit/references/playbooks/formats/talking-head.md
+++ b/.agents/skills/hypit/references/playbooks/formats/talking-head.md
@@ -6,7 +6,7 @@ captions, B-roll, and editorial typography as peer Tracks.
## Author the SVML program
1. Write one Script with a strong opening Segment and 2–3 following Segments for context, evidence,
- mechanism, payoff, or CTA. Keep each generated speaking take within 4–15 seconds; 8–12 seconds is
+ mechanism, payoff, or CTA. Keep each generated speaking take inside the selected model's declared range; 8–12 seconds is
a useful target when the delivery remains natural.
2. Run `estimate:Speech` for each Segment before generation.
3. Vendor `speaker-v1.svs`. Put stable `composition-stability`, `camera-motion`, `edit-rhythm`,
diff --git a/docs/quickstart/composition.md b/docs/quickstart/composition.md
index 17be8d96..facb1c73 100644
--- a/docs/quickstart/composition.md
+++ b/docs/quickstart/composition.md
@@ -116,6 +116,7 @@ The complete data flow from Script to rendered video. This example is based on
<svml>
<import from="@hypit/script@1"/>
<import as="wording" from="@hypit/text@1"/>
+ <import as="gpt" from="@hypit/gpt-image@1"/>
<import as="seedance" from="@hypit/seedance@1"/>
<import as="speech" from="@hypit/speech-spine@1"/>
<import as="whisperx" from="@hypit/whisperx@1"/>
@@ -139,10 +140,22 @@ The complete data flow from Script to rendered video. This example is based on
<wording:Value id="direction">
Locked medium close-up in a quiet daylight studio. Spoken dialogue — say exactly: Meaning becomes the source.
</wording:Value>
- <seedance:TextVideo id="take" model="mini"
- prompt={direction} duration="5" generate-audio="true"/>
- <seedance:TextVideo id="motion" model="mini"
- prompt={direction} duration="5"/>
+ <wording:Value id="scene-look">
+ A photograph with the texture of real iPhone footage. Generate a vertical seated medium
+ close-up, as one frame cut out of video actually shot on an iPhone: genuinely real rather than
+ glossy, carrying the texture of video and not of a posed photograph. The background stays clearly
+ visible, with no depth-of-field blur. Skin texture is fine and real, the light is natural, and no
+ part of the picture is broken. One presenter at a desk in a quiet daylight studio.
+ </wording:Value>
+ <gpt:Image id="studio-scene" prompt={scene-look} aspect-ratio="9:16" resolution="2K"/>
+ <seedance:ReferenceVideo id="take" model="mini"
+ prompt={direction} duration="5" generate-audio="true">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="motion" model="mini"
+ prompt={direction} duration="5">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
<space:Canvas id="vertical" width="1080" height="1920"/>
<space:Frame id="speech-frame" within={vertical}
diff --git a/docs/quickstart/generation.md b/docs/quickstart/generation.md
index e8e177d2..bb088a68 100644
--- a/docs/quickstart/generation.md
+++ b/docs/quickstart/generation.md
@@ -146,6 +146,10 @@ Prompt-only generation. This is the only shape that accepts `web-search`.
<seedance:TextVideo id="ambient" model="mini"
prompt={ambient-direction} duration="5" web-search="false"/>
```
+Craft requires every take to start from an accepted image, so a first frame or ordered references are
+the shapes to author with. This shape is documented because the model has it, not because a take
+should begin without a picture.
+
### seedance:FrameVideo
diff --git a/docs/zh/quickstart/composition.md b/docs/zh/quickstart/composition.md
index bf4a783e..2519e6bd 100644
--- a/docs/zh/quickstart/composition.md
+++ b/docs/zh/quickstart/composition.md
@@ -112,6 +112,7 @@ Build Target,也可以直接接到媒体裁切、音频/帧提取或模型参
<svml>
<import from="@hypit/script@1"/>
<import as="wording" from="@hypit/text@1"/>
+ <import as="gpt" from="@hypit/gpt-image@1"/>
<import as="seedance" from="@hypit/seedance@1"/>
<import as="speech" from="@hypit/speech-spine@1"/>
<import as="whisperx" from="@hypit/whisperx@1"/>
@@ -135,10 +136,22 @@ Build Target,也可以直接接到媒体裁切、音频/帧提取或模型参
<wording:Value id="direction">
Locked medium close-up in a quiet daylight studio. Spoken dialogue — say exactly: Meaning becomes the source.
</wording:Value>
- <seedance:TextVideo id="take" model="mini"
- prompt={direction} duration="5" generate-audio="true"/>
- <seedance:TextVideo id="motion" model="mini"
- prompt={direction} duration="5"/>
+ <wording:Value id="scene-look">
+ A photograph with the texture of real iPhone footage. Generate a vertical seated medium
+ close-up, as one frame cut out of video actually shot on an iPhone: genuinely real rather than
+ glossy, carrying the texture of video and not of a posed photograph. The background stays clearly
+ visible, with no depth-of-field blur. Skin texture is fine and real, the light is natural, and no
+ part of the picture is broken. One presenter at a desk in a quiet daylight studio.
+ </wording:Value>
+ <gpt:Image id="studio-scene" prompt={scene-look} aspect-ratio="9:16" resolution="2K"/>
+ <seedance:ReferenceVideo id="take" model="mini"
+ prompt={direction} duration="5" generate-audio="true">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="motion" model="mini"
+ prompt={direction} duration="5">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
<space:Canvas id="vertical" width="1080" height="1920"/>
<space:Frame id="speech-frame" within={vertical}
diff --git a/docs/zh/quickstart/generation.md b/docs/zh/quickstart/generation.md
index c9f8b351..38621393 100644
--- a/docs/zh/quickstart/generation.md
+++ b/docs/zh/quickstart/generation.md
@@ -133,6 +133,8 @@ Seedance 2.5 复用同样的 Surface,而不是由 Runtime 把别的模型偷
<seedance:TextVideo id="ambient" model="mini"
prompt={ambient-direction} duration="5" web-search="false"/>
```
+创作规则要求每条 take 都从一张已验收的图开始,所以实际创作用的是首帧或有序引用这两种形状。这里记录这个形状是因为模型有它,而不是因为一条 take 应该在没有画面的情况下开始。
+
### seedance:FrameVideo
diff --git a/examples/street-interview-preview/main.svml b/examples/street-interview-preview/main.svml
index 36d68fbd..f4a92693 100644
--- a/examples/street-interview-preview/main.svml
+++ b/examples/street-interview-preview/main.svml
@@ -3,6 +3,7 @@
<svml>
<import from="@hypit/script@1"/>
<import as="wording" from="@hypit/text@1"/>
+ <import as="gpt" from="@hypit/gpt-image@1"/>
<import as="seedance" from="@hypit/seedance@1"/>
<import as="speech" from="@hypit/speech-spine@1"/>
<import as="whisperx" from="@hypit/whisperx@1"/>
@@ -54,14 +55,40 @@
<space:Frame id="full" within={vertical}
left="0%" top="0%" right="100%" bottom="100%"/>
+ <!-- The location is established once and every take references it, so the street stays one street.
+ Prompt-only generation would give each take its own afternoon. -->
+ <wording:Value id="scene-look">
+ A photograph with the texture of real iPhone footage. Generate a vertical wide two-shot, as one
+ frame cut out of video actually shot on an iPhone: genuinely real rather than glossy, carrying
+ the texture of video and not of a posed photograph. The background stays clearly visible, with no
+ depth-of-field blur. Skin texture is fine and real, the light is natural, and no part of the
+ picture is broken. An interviewer and a passer-by stand together on a shopping street in late
+ afternoon light, shopfronts and pedestrians behind them.
+ </wording:Value>
+ <gpt:Image id="scene" prompt={scene-look} aspect-ratio="9:16" resolution="2K"/>
+
<wording:Value id="direction">Handheld street interview, late afternoon, two speakers.</wording:Value>
- <seedance:TextVideo id="take-hook" model="mini" prompt={direction} duration="5" generate-audio="true"/>
- <seedance:TextVideo id="take-price" model="mini" prompt={direction} duration="6" generate-audio="true"/>
- <seedance:TextVideo id="take-rainbow" model="mini" prompt={direction} duration="11" generate-audio="true"/>
- <seedance:TextVideo id="take-payment" model="mini" prompt={direction} duration="9" generate-audio="true"/>
- <seedance:TextVideo id="cutaway-bags" model="mini" prompt={direction} duration="4"/>
- <seedance:TextVideo id="cutaway-fee" model="mini" prompt={direction} duration="4"/>
- <seedance:TextVideo id="cutaway-receipt" model="mini" prompt={direction} duration="5"/>
+ <seedance:ReferenceVideo id="take-hook" model="mini" prompt={direction} duration="5" generate-audio="true">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="take-price" model="mini" prompt={direction} duration="6" generate-audio="true">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="take-rainbow" model="mini" prompt={direction} duration="11" generate-audio="true">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="take-payment" model="mini" prompt={direction} duration="9" generate-audio="true">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="cutaway-bags" model="mini" prompt={direction} duration="4">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="cutaway-fee" model="mini" prompt={direction} duration="4">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="cutaway-receipt" model="mini" prompt={direction} duration="5">
+ <seedance:Reference image={scene.image}/>
+ </seedance:ReferenceVideo>
<speech:Spine id="speech" frame-rate="24"
visual-frame={speech-frame} visual-appearance={studio.speech.visual} visual-z="0">
diff --git a/examples/talking-film-broll-preview/main.svml b/examples/talking-film-broll-preview/main.svml
index 37da004c..18f3fc62 100644
--- a/examples/talking-film-broll-preview/main.svml
+++ b/examples/talking-film-broll-preview/main.svml
@@ -3,6 +3,7 @@
<svml>
<import from="@hypit/script@1"/>
<import as="wording" from="@hypit/text@1"/>
+ <import as="gpt" from="@hypit/gpt-image@1"/>
<import as="seedance" from="@hypit/seedance@1"/>
<import as="speech" from="@hypit/speech-spine@1"/>
<import as="whisperx" from="@hypit/whisperx@1"/>
@@ -27,9 +28,26 @@
left="12%" top="52%" right="88%" bottom="78%"/>
<wording:Value id="direction">Locked medium close-up in a quiet daylight studio. Spoken dialogue.</wording:Value>
- <seedance:TextVideo id="take" model="mini" prompt={direction} duration="10" generate-audio="true"/>
- <seedance:TextVideo id="motion-a" model="mini" prompt={direction} duration="5"/>
- <seedance:TextVideo id="motion-b" model="mini" prompt={direction} duration="5"/>
+ <!-- The studio is established once; every take references that image, so one room stays
+ one room. Prompt-only generation would give each take its own. -->
+ <wording:Value id="studio-look">
+ A photograph with the texture of real iPhone footage. Generate a vertical seated medium close-up,
+ as one frame cut out of video actually shot on an iPhone: genuinely real rather than glossy,
+ carrying the texture of video and not of a posed photograph. The background stays clearly visible,
+ with no depth-of-field blur. Skin texture is fine and real, the light is natural, and no part of
+ the picture is broken. One presenter at a desk in a quiet daylight studio.
+ </wording:Value>
+ <gpt:Image id="studio-scene" prompt={studio-look} aspect-ratio="9:16" resolution="2K"/>
+
+ <seedance:ReferenceVideo id="take" model="mini" prompt={direction} duration="10" generate-audio="true">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="motion-a" model="mini" prompt={direction} duration="5">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="motion-b" model="mini" prompt={direction} duration="5">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
<speech:Spine id="speech" frame-rate="30"
visual-frame={speech-frame} visual-appearance={studio.speech.visual} visual-z="0">
diff --git a/examples/talking-film-graph-check/main.svml b/examples/talking-film-graph-check/main.svml
index c31e680e..a0e9fda9 100644
--- a/examples/talking-film-graph-check/main.svml
+++ b/examples/talking-film-graph-check/main.svml
@@ -3,6 +3,7 @@
<svml>
<import from="@hypit/script@1"/>
<import as="wording" from="@hypit/text@1"/>
+ <import as="gpt" from="@hypit/gpt-image@1"/>
<import as="seedance" from="@hypit/seedance@1"/>
<import as="speech" from="@hypit/speech-spine@1"/>
<import as="whisperx" from="@hypit/whisperx@1"/>
@@ -30,8 +31,23 @@
left="10%" top="20%" right="90%" bottom="70%"/>
<wording:Value id="direction">Locked medium close-up in a quiet daylight studio. Spoken dialogue — say exactly: Meaning becomes the source.</wording:Value>
- <seedance:TextVideo id="take" model="mini" prompt={direction} duration="5" generate-audio="true"/>
- <seedance:TextVideo id="motion" model="mini" prompt={direction} duration="5"/>
+ <!-- The studio is established once; every take references that image, so one room stays
+ one room. Prompt-only generation would give each take its own. -->
+ <wording:Value id="studio-look">
+ A photograph with the texture of real iPhone footage. Generate a vertical seated medium close-up,
+ as one frame cut out of video actually shot on an iPhone: genuinely real rather than glossy,
+ carrying the texture of video and not of a posed photograph. The background stays clearly visible,
+ with no depth-of-field blur. Skin texture is fine and real, the light is natural, and no part of
+ the picture is broken. One presenter at a desk in a quiet daylight studio.
+ </wording:Value>
+ <gpt:Image id="studio-scene" prompt={studio-look} aspect-ratio="9:16" resolution="2K"/>
+
+ <seedance:ReferenceVideo id="take" model="mini" prompt={direction} duration="5" generate-audio="true">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
+ <seedance:ReferenceVideo id="motion" model="mini" prompt={direction} duration="5">
+ <seedance:Reference image={studio-scene.image}/>
+ </seedance:ReferenceVideo>
<speech:Spine id="speech" frame-rate="30"
visual-frame={speech-frame} visual-appearance={studio.speech.visual} visual-z="0">