# Case 5 · opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042 · 七轮 / Seven-round API 输入
这是实际发送给 API 的逐轮文本输入索引。凭据已脱敏；每一轮都在同一个 context window 中保留之前的 user/assistant 消息。
- **Model:** `gemini-3.5-flash`
- **Media mode:** `images`
- **Round count:** `7`
- **Protocol:** `cs2_reference_multiturn_v2`
## System prompt（每轮共用）
`opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.system.md`
```text
You are a meticulous Counter-Strike 2 first-person gameplay annotator working in a causal multi-turn conversation. The video is divided into five fixed chronological chunks. The conversation begins with one conditioning image, then one visual attachment is added per chunk, and a final text-only turn asks for the production caption.
Use only the visual evidence shown so far. Never infer, describe, or summarize a later chunk during an earlier turn. Keep the same player, held item, viewpoint, environment, and Action identity across turns when the evidence supports continuity. A later turn may correct an earlier draft when stronger evidence appears.
This is a CS2 adaptation of the supplied segment-by-segment protocol. Conditioning and chunk turns return temporary structured drafts; they are evidence for the final turn, not production labels. The final text-only turn follows the existing CS2 production schema: `english_json`, five fixed segments, `environment_detail`, `contained_actions`, `move_description`, and `camera_description`.
Use `<|Subject1|>` for the controllable first-person agent. Hands, gloves, weapons, knives, grenades, C4, and the viewmodel are equipment or body parts of Subject1, not separate subjects. In every draft and in the final object, main-subject actions and state changes belong only in `contained_actions`, wrapped as `...`. Scene and camera prose must not contain Subject tokens or Action tokens. In draft objects, movement is kept in `move_description` and camera motion in `camera_description`; these are separate from action sentences. The final production prompt controls the exact final channel wording.
The supplied chunk boundaries are authoritative. Do not split, merge, reorder, or alter them. DEM facts are evidence for event identity, order, and timing; they do not replace visual evidence. Do not mention frame numbers, HUD, overlays, interface elements, or internal metadata in caption prose. Every turn must return one raw JSON object and no Markdown.
# Adapted in-context examples
The following small examples illustrate the draft separation. They are format examples only; never copy their subjects, actions, timestamps, or scene details.
```json
{
"segment_draft": {
"visual_description_long": "A stone corridor fills the background, with damp blocks, a bright opening in the distance, and a sign on the right wall.",
"description_change": null,
"contained_actions": [
{
"start_time": "00:00.300",
"end_time": "00:00.900",
"action_description": "<|Subject1|> raises the held tool and rotates it through the visible inspection motion."
}
]
}
}
```
For a later chunk, put only scene and camera deltas in `description_change`; put all main-subject movement and state changes in `contained_actions`:
```json
{
"segment_draft": {
"visual_description_long": null,
"description_change": "Compared to the previous segment, the doorway is nearer and the right-wall sign is partly occluded.",
"contained_actions": [
{
"start_time": "00:01.200",
"end_time": "00:01.900",
"action_description": "<|Subject1|> continues the same inspection motion and brings the held item back toward the lower-center viewmodel position."
}
]
}
}
```
```
## Round 0 · `conditioning`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.conditioning.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.conditioning.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.conditioning.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.conditioning.jpg)
```text
# Multi-turn conditioning turn
This turn contains only one conditioning image from frame 0. Do not assume that any chunk has been shown. Establish a provisional global state from this image: `perspective_view`, `Style`, the current complete `subject_list`, the visible initial held item and hand/glove evidence, and a concise baseline environment state. This is a draft that later turns may revise.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current input: conditioning frame 0 (before chunk 0)
- Current range: 00:00.000 - 00:00.062
- Total content chunks: 5
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Required output
Return exactly one raw JSON object with these keys and no final production caption:
```json
{
"turn_type": "conditioning",
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "high",
"evidence": "Concrete evidence visible in the conditioning image",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "player-controlled CS2 agent",
"subject_attribute": "Static description anchored to the conditioning image, or null",
"confidence": "low | medium | high",
"evidence": "Concrete evidence visible in this image",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_summary": {
"initial_held_item": "Visible item or null",
"environment_baseline": "Only the visible baseline environment",
"uncertainties": []
}
}
```
Do not output `segments`, `english_json`, `chinese_json`, future actions, or any detail that is not visible in this image.
```
## Round 1 · `chunk0`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk0.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk0.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk0.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk0.jpg)
```text
# Multi-turn CS2 chunk turn
The newly attached visual contains ONLY the current chunk shown below. It is the immediate continuation of the chunks already shown. Analyze this chunk causally and produce a draft for this chunk only. Do not write the final bilingual production caption yet. The global `perspective_view`, `Style`, and `subject_list` must be the current best cumulative state; revise them when this chunk provides stronger evidence.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current chunk: 1 of 5
- Current exact range: 00:00.062 - 00:01.062
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Current-chunk facts
These facts are restricted to the current chunk. Use them to check event identity and local timing, but still require visible evidence for prose:
{
"chunk": {
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
"actions_visible_in_this_chunk": [],
"view_and_movement_visible_in_this_chunk": [
{
"start_time": "00:00.062",
"end_time": "00:01.062",
"local_start_seconds": 0.0625,
"local_end_seconds": 1.0625,
"time_base": "clip_relative",
"type": "movement",
"movement": "still"
}
]
}
## Causal and revision rules
- Do not describe a future chunk. Do not repeat the prior chunk's prose; describe only this chunk's newly visible scene/camera delta and actions.
- This is the FIRST CONTENT CHUNK, after the conditioning image. Write a complete 80-150 word scene baseline in visual_description_long, and set description_change to null. Include the visible environment in this chunk; do not predict later changes.
- For the first content chunk, `visual_description_long` is the complete subject-free scene baseline and `description_change` is `null`. For later chunks, `visual_description_long` is `null` and `description_change` contains only subject-free scene/background deltas beginning exactly with `Compared to the previous segment, `. Camera motion is recorded separately in `camera_description` for this CS2 schema adaptation.
- Every main-subject action, posture, held-item state change, and visible action consequence belongs in `contained_actions`, uses the exact `<|Subject1|>` token, and is enclosed by one `...` wrapper.
- Use a fresh local action description for each chunk touched by a continuing action. The final harness reconciles deterministic event identity; do not invent an ID from a later unseen interval.
- Use one `move_description` object and one `camera_description` object. Movement uses `<|Subject1|>` as its grammatical subject. Camera prose is subject-free and uses `...` when motion is clearly visible.
- Keep action timestamps inside the current exact range and in chronological order. Copy the current chunk boundaries exactly.
## Required output
Return exactly one raw JSON object with no Markdown:
```json
{
"turn_type": "segment",
"segment_index": 1,
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "low | medium | high",
"evidence": "Cumulative perspective evidence and this chunk's update",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "Current best identity",
"subject_attribute": "Static appearance anchored to earliest clear evidence",
"confidence": "low | medium | high",
"evidence": "Cumulative identity evidence and this chunk's update",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_update": {
"environment_delta": "New environment evidence in this chunk only",
"held_item_state": "Current visible state or null",
"uncertainties": []
},
"segment_draft": {
"start_time": "00:00.062",
"end_time": "00:01.062",
"visual_description_long": "Complete subject-free scene baseline for the first content chunk",
"description_change": null,
"environment_description": "Local environment progression for this chunk",
"contained_actions": [],
"move_description": [{"move_description": "<|Subject1|> ..."}],
"camera_description": [{"camera_description": "The first-person view ..."}]
},
"open_questions_for_later_segments": []
}
```
```
## Round 2 · `chunk1`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk1.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk1.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk1.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk1.jpg)
```text
# Multi-turn CS2 chunk turn
The newly attached visual contains ONLY the current chunk shown below. It is the immediate continuation of the chunks already shown. Analyze this chunk causally and produce a draft for this chunk only. Do not write the final bilingual production caption yet. The global `perspective_view`, `Style`, and `subject_list` must be the current best cumulative state; revise them when this chunk provides stronger evidence.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current chunk: 2 of 5
- Current exact range: 00:01.062 - 00:02.062
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Current-chunk facts
These facts are restricted to the current chunk. Use them to check event identity and local timing, but still require visible evidence for prose:
{
"chunk": {
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
"actions_visible_in_this_chunk": [],
"view_and_movement_visible_in_this_chunk": [
{
"start_time": "00:01.062",
"end_time": "00:02.062",
"local_start_seconds": 1.0625,
"local_end_seconds": 2.0625,
"time_base": "clip_relative",
"type": "movement",
"movement": "still"
}
]
}
## Causal and revision rules
- Do not describe a future chunk. Do not repeat the prior chunk's prose; describe only this chunk's newly visible scene/camera delta and actions.
- This is a FOLLOWING CONTENT CHUNK. Set visual_description_long to null. Write only new scene/background changes in description_change, beginning exactly with 'Compared to the previous segment, '. Do not repeat the full prior inventory.
- For the first content chunk, `visual_description_long` is the complete subject-free scene baseline and `description_change` is `null`. For later chunks, `visual_description_long` is `null` and `description_change` contains only subject-free scene/background deltas beginning exactly with `Compared to the previous segment, `. Camera motion is recorded separately in `camera_description` for this CS2 schema adaptation.
- Every main-subject action, posture, held-item state change, and visible action consequence belongs in `contained_actions`, uses the exact `<|Subject1|>` token, and is enclosed by one `...` wrapper.
- Use a fresh local action description for each chunk touched by a continuing action. The final harness reconciles deterministic event identity; do not invent an ID from a later unseen interval.
- Use one `move_description` object and one `camera_description` object. Movement uses `<|Subject1|>` as its grammatical subject. Camera prose is subject-free and uses `...` when motion is clearly visible.
- Keep action timestamps inside the current exact range and in chronological order. Copy the current chunk boundaries exactly.
## Required output
Return exactly one raw JSON object with no Markdown:
```json
{
"turn_type": "segment",
"segment_index": 2,
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "low | medium | high",
"evidence": "Cumulative perspective evidence and this chunk's update",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "Current best identity",
"subject_attribute": "Static appearance anchored to earliest clear evidence",
"confidence": "low | medium | high",
"evidence": "Cumulative identity evidence and this chunk's update",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_update": {
"environment_delta": "New environment evidence in this chunk only",
"held_item_state": "Current visible state or null",
"uncertainties": []
},
"segment_draft": {
"start_time": "00:01.062",
"end_time": "00:02.062",
"visual_description_long": null,
"description_change": "Compared to the previous segment, ...",
"environment_description": "Local environment progression for this chunk",
"contained_actions": [],
"move_description": [{"move_description": "<|Subject1|> ..."}],
"camera_description": [{"camera_description": "The first-person view ..."}]
},
"open_questions_for_later_segments": []
}
```
```
## Round 3 · `chunk2`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk2.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk2.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk2.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk2.jpg)
```text
# Multi-turn CS2 chunk turn
The newly attached visual contains ONLY the current chunk shown below. It is the immediate continuation of the chunks already shown. Analyze this chunk causally and produce a draft for this chunk only. Do not write the final bilingual production caption yet. The global `perspective_view`, `Style`, and `subject_list` must be the current best cumulative state; revise them when this chunk provides stronger evidence.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current chunk: 3 of 5
- Current exact range: 00:02.062 - 00:03.062
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Current-chunk facts
These facts are restricted to the current chunk. Use them to check event identity and local timing, but still require visible evidence for prose:
{
"chunk": {
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
"actions_visible_in_this_chunk": [
{
"start_time": "00:02.781",
"end_time": "00:03.062",
"local_start_seconds": 2.7812,
"local_end_seconds": 3.0625,
"time_base": "clip_relative",
"event_id": "target",
"type": "switch",
"subtype": "draw",
"weapon": "Desert Eagle",
"from_weapon": "Karambit",
"to_weapon": "Desert Eagle"
}
],
"view_and_movement_visible_in_this_chunk": [
{
"start_time": "00:02.062",
"end_time": "00:03.062",
"local_start_seconds": 2.0625,
"local_end_seconds": 3.0625,
"time_base": "clip_relative",
"type": "movement",
"movement": "still"
}
]
}
## Causal and revision rules
- Do not describe a future chunk. Do not repeat the prior chunk's prose; describe only this chunk's newly visible scene/camera delta and actions.
- This is a FOLLOWING CONTENT CHUNK. Set visual_description_long to null. Write only new scene/background changes in description_change, beginning exactly with 'Compared to the previous segment, '. Do not repeat the full prior inventory.
- For the first content chunk, `visual_description_long` is the complete subject-free scene baseline and `description_change` is `null`. For later chunks, `visual_description_long` is `null` and `description_change` contains only subject-free scene/background deltas beginning exactly with `Compared to the previous segment, `. Camera motion is recorded separately in `camera_description` for this CS2 schema adaptation.
- Every main-subject action, posture, held-item state change, and visible action consequence belongs in `contained_actions`, uses the exact `<|Subject1|>` token, and is enclosed by one `...` wrapper.
- Use a fresh local action description for each chunk touched by a continuing action. The final harness reconciles deterministic event identity; do not invent an ID from a later unseen interval.
- Use one `move_description` object and one `camera_description` object. Movement uses `<|Subject1|>` as its grammatical subject. Camera prose is subject-free and uses `...` when motion is clearly visible.
- Keep action timestamps inside the current exact range and in chronological order. Copy the current chunk boundaries exactly.
## Required output
Return exactly one raw JSON object with no Markdown:
```json
{
"turn_type": "segment",
"segment_index": 3,
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "low | medium | high",
"evidence": "Cumulative perspective evidence and this chunk's update",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "Current best identity",
"subject_attribute": "Static appearance anchored to earliest clear evidence",
"confidence": "low | medium | high",
"evidence": "Cumulative identity evidence and this chunk's update",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_update": {
"environment_delta": "New environment evidence in this chunk only",
"held_item_state": "Current visible state or null",
"uncertainties": []
},
"segment_draft": {
"start_time": "00:02.062",
"end_time": "00:03.062",
"visual_description_long": null,
"description_change": "Compared to the previous segment, ...",
"environment_description": "Local environment progression for this chunk",
"contained_actions": [],
"move_description": [{"move_description": "<|Subject1|> ..."}],
"camera_description": [{"camera_description": "The first-person view ..."}]
},
"open_questions_for_later_segments": []
}
```
```
## Round 4 · `chunk3`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk3.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk3.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk3.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk3.jpg)
```text
# Multi-turn CS2 chunk turn
The newly attached visual contains ONLY the current chunk shown below. It is the immediate continuation of the chunks already shown. Analyze this chunk causally and produce a draft for this chunk only. Do not write the final bilingual production caption yet. The global `perspective_view`, `Style`, and `subject_list` must be the current best cumulative state; revise them when this chunk provides stronger evidence.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current chunk: 4 of 5
- Current exact range: 00:03.062 - 00:04.062
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Current-chunk facts
These facts are restricted to the current chunk. Use them to check event identity and local timing, but still require visible evidence for prose:
{
"chunk": {
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
"actions_visible_in_this_chunk": [
{
"start_time": "00:03.062",
"end_time": "00:03.672",
"local_start_seconds": 3.0625,
"local_end_seconds": 3.6719,
"time_base": "clip_relative",
"event_id": "target",
"type": "switch",
"subtype": "draw",
"weapon": "Desert Eagle",
"from_weapon": "Karambit",
"to_weapon": "Desert Eagle"
}
],
"view_and_movement_visible_in_this_chunk": [
{
"start_time": "00:03.062",
"end_time": "00:04.062",
"local_start_seconds": 3.0625,
"local_end_seconds": 4.0625,
"time_base": "clip_relative",
"type": "movement",
"movement": "still"
}
]
}
## Causal and revision rules
- Do not describe a future chunk. Do not repeat the prior chunk's prose; describe only this chunk's newly visible scene/camera delta and actions.
- This is a FOLLOWING CONTENT CHUNK. Set visual_description_long to null. Write only new scene/background changes in description_change, beginning exactly with 'Compared to the previous segment, '. Do not repeat the full prior inventory.
- For the first content chunk, `visual_description_long` is the complete subject-free scene baseline and `description_change` is `null`. For later chunks, `visual_description_long` is `null` and `description_change` contains only subject-free scene/background deltas beginning exactly with `Compared to the previous segment, `. Camera motion is recorded separately in `camera_description` for this CS2 schema adaptation.
- Every main-subject action, posture, held-item state change, and visible action consequence belongs in `contained_actions`, uses the exact `<|Subject1|>` token, and is enclosed by one `...` wrapper.
- Use a fresh local action description for each chunk touched by a continuing action. The final harness reconciles deterministic event identity; do not invent an ID from a later unseen interval.
- Use one `move_description` object and one `camera_description` object. Movement uses `<|Subject1|>` as its grammatical subject. Camera prose is subject-free and uses `...` when motion is clearly visible.
- Keep action timestamps inside the current exact range and in chronological order. Copy the current chunk boundaries exactly.
## Required output
Return exactly one raw JSON object with no Markdown:
```json
{
"turn_type": "segment",
"segment_index": 4,
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "low | medium | high",
"evidence": "Cumulative perspective evidence and this chunk's update",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "Current best identity",
"subject_attribute": "Static appearance anchored to earliest clear evidence",
"confidence": "low | medium | high",
"evidence": "Cumulative identity evidence and this chunk's update",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_update": {
"environment_delta": "New environment evidence in this chunk only",
"held_item_state": "Current visible state or null",
"uncertainties": []
},
"segment_draft": {
"start_time": "00:03.062",
"end_time": "00:04.062",
"visual_description_long": null,
"description_change": "Compared to the previous segment, ...",
"environment_description": "Local environment progression for this chunk",
"contained_actions": [],
"move_description": [{"move_description": "<|Subject1|> ..."}],
"camera_description": [{"camera_description": "The first-person view ..."}]
},
"open_questions_for_later_segments": []
}
```
```
## Round 5 · `chunk4`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk4.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.chunk4.md)
视觉附件：
- [`assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk4.jpg`](../assets/contact_sheets/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.chunk4.jpg)
```text
# Multi-turn CS2 chunk turn
The newly attached visual contains ONLY the current chunk shown below. It is the immediate continuation of the chunks already shown. Analyze this chunk causally and produce a draft for this chunk only. Do not write the final bilingual production caption yet. The global `perspective_view`, `Style`, and `subject_list` must be the current best cumulative state; revise them when this chunk provides stronger evidence.
## Video metadata
- Whole-video duration: 5.0625 seconds
- Sampled frames: 81 at 16.0 FPS
- Current chunk: 5 of 5
- Current exact range: 00:04.062 - 00:05.062
- Map: de_ancient
- Player side: None
- Initial held item from manifest: Desert Eagle
- Fixed chunk ranges:
[
{
"chunk_index": 0,
"start_time": "00:00.062",
"end_time": "00:01.062",
"start_frame": 1,
"end_frame_exclusive": 17
},
{
"chunk_index": 1,
"start_time": "00:01.062",
"end_time": "00:02.062",
"start_frame": 17,
"end_frame_exclusive": 33
},
{
"chunk_index": 2,
"start_time": "00:02.062",
"end_time": "00:03.062",
"start_frame": 33,
"end_frame_exclusive": 49
},
{
"chunk_index": 3,
"start_time": "00:03.062",
"end_time": "00:04.062",
"start_frame": 49,
"end_frame_exclusive": 65
},
{
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
}
]
## Current-chunk facts
These facts are restricted to the current chunk. Use them to check event identity and local timing, but still require visible evidence for prose:
{
"chunk": {
"chunk_index": 4,
"start_time": "00:04.062",
"end_time": "00:05.062",
"start_frame": 65,
"end_frame_exclusive": 81
},
"actions_visible_in_this_chunk": [],
"view_and_movement_visible_in_this_chunk": [
{
"start_time": "00:04.062",
"end_time": "00:05.047",
"local_start_seconds": 4.0625,
"local_end_seconds": 5.0469,
"time_base": "clip_relative",
"type": "movement",
"movement": "still"
}
]
}
## Causal and revision rules
- Do not describe a future chunk. Do not repeat the prior chunk's prose; describe only this chunk's newly visible scene/camera delta and actions.
- This is a FOLLOWING CONTENT CHUNK. Set visual_description_long to null. Write only new scene/background changes in description_change, beginning exactly with 'Compared to the previous segment, '. Do not repeat the full prior inventory.
- For the first content chunk, `visual_description_long` is the complete subject-free scene baseline and `description_change` is `null`. For later chunks, `visual_description_long` is `null` and `description_change` contains only subject-free scene/background deltas beginning exactly with `Compared to the previous segment, `. Camera motion is recorded separately in `camera_description` for this CS2 schema adaptation.
- Every main-subject action, posture, held-item state change, and visible action consequence belongs in `contained_actions`, uses the exact `<|Subject1|>` token, and is enclosed by one `...` wrapper.
- Use a fresh local action description for each chunk touched by a continuing action. The final harness reconciles deterministic event identity; do not invent an ID from a later unseen interval.
- Use one `move_description` object and one `camera_description` object. Movement uses `<|Subject1|>` as its grammatical subject. Camera prose is subject-free and uses `...` when motion is clearly visible.
- Keep action timestamps inside the current exact range and in chronological order. Copy the current chunk boundaries exactly.
## Required output
Return exactly one raw JSON object with no Markdown:
```json
{
"turn_type": "segment",
"segment_index": 5,
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"perspective_assessment": {
"confidence": "low | medium | high",
"evidence": "Cumulative perspective evidence and this chunk's update",
"could_be_overwritten": true
},
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "Current best identity",
"subject_attribute": "Static appearance anchored to earliest clear evidence",
"confidence": "low | medium | high",
"evidence": "Cumulative identity evidence and this chunk's update",
"could_be_overwritten": true
}
],
"subject_id_revisions": [],
"state_update": {
"environment_delta": "New environment evidence in this chunk only",
"held_item_state": "Current visible state or null",
"uncertainties": []
},
"segment_draft": {
"start_time": "00:04.062",
"end_time": "00:05.062",
"visual_description_long": null,
"description_change": "Compared to the previous segment, ...",
"environment_description": "Local environment progression for this chunk",
"contained_actions": [],
"move_description": [{"move_description": "<|Subject1|> ..."}],
"camera_description": [{"camera_description": "The first-person view ..."}]
},
"open_questions_for_later_segments": []
}
```
```
## Round 6 · `final`
文本指令文件：[`prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.final.md`](../prompts/en/opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042.multiturn.final.md)
视觉附件：无（仅使用已有 context）
```text
# Multi-turn final summary turn
All visual evidence has already been supplied in previous turns. This turn is text-only: do not expect or request a new image or video. Review the conditioning state and every intermediate chunk draft in the accumulated context, correct inconsistencies, and produce the final English-only CS2 production object.
The existing rendered CS2 production prompt is included below. Follow its fixed five-segment schema, authoritative boundaries, Action/Move/Camera rules, deterministic metadata policy, and JSON-only output contract. Intermediate keys such as `turn_type`, `state_summary`, `state_update`, `confidence`, `evidence`, and `open_questions_for_later_segments` must not appear in the final object.
## Final production prompt
You are an expert Counter-Strike 2 (CS2) first-person gameplay annotator. Return
one English JSON object only. The label is used as training metadata, so concrete
visual evidence, event identity, chronological order, and fixed boundaries matter
more than elegant prose. Describe the scene from the controllable player's
first-person gameplay perspective. Use the exact `<|Subject1|>` token for the
controllable agent; in camera prose, make `<|Subject1|>'s view` the grammatical
subject (for example, `<|Subject1|>'s view turns left`). Never write from an
outside observer's viewpoint and never label the agent as `viewer`, `observer`,
or `audience`. Describe directions from the evidence rather than forcing every
change into left/right. Preserve the observed axis and order: yaw can turn left
or right, pitch can tilt up or down, zoom/FOV can zoom in or out or widen/narrow,
and diagonal combinations such as up-left or down-right are valid when
supported. Use plain left/right/up/down wording relative to `<|Subject1|>`'s
current view; do not write `viewer's screen-left/right` or substitute the
agent's body or map/world orientation. The only anatomical left/right terms are
explicit body references such as `left hand` and `right hand`.
# Mission
Watch the complete 81-frame clip and describe the persistent game world, the
first-person viewmodel, visible player models, and every supported player action.
Describe each of the five causal chunks independently. Do not copy a whole-clip
paragraph into every chunk. Use the full/foreground/opening reference sheets and
the complete local chunk videos for appearance, geometry, lighting, hands,
weapons, and effects. Use the supplied DEM facts only for event type,
order, item identity, and timing. Never invent a fact because it is common in CS2.
The request contains a chronological contact sheet, a foreground/viewmodel sheet,
a large opening sheet, and one local video for each chunk. Each local video is the
exact 16-frame window at 16 FPS. Watch each local video from beginning to end and
compare the changes over time. The web viewer may loop the local videos for
inspection; looping is a viewing aid and does not imply that an action repeats.
# Fixed time contract
The normal input is exactly 81 frames at 16 FPS and 832x480. Frame 0 is a
conditioning/opening reference, not a sixth chunk. The five half-open windows are:
| chunk_index | output frames | time |
| ---: | --- | --- |
| 0 | [1, 17) | 00:00.062 - 00:01.062 |
| 1 | [17, 33) | 00:01.062 - 00:02.062 |
| 2 | [33, 49) | 00:02.062 - 00:03.062 |
| 3 | [49, 65) | 00:03.062 - 00:04.062 |
| 4 | [65, 81) | 00:04.062 - 00:05.062 |
Copy all segment start/end strings exactly from `chunk_ranges`. Do not put raw
demo ticks, source frame numbers, coordinates, yaw/pitch degrees, timestamps,
seconds, or algorithmic measurements in prose. `clip_metadata`, `chunk_ranges`,
and the trusted timelines below are authoritative; frame labels are evidence
aids only and must not be copied into a description.
# Subject and appearance
Use `<|Subject1|>` for the controllable first-person agent. The hands, arms,
gloves, viewmodel, weapon, grenade, knife, C4, and defuse kit operated by that
agent are parts/equipment of Subject1, not separate subjects. Do not invent a
face, body, gender, clothing, skin, or anatomy hidden outside the viewport.
Set `perspective_view` to `first-person` and `Style` to
`video game, first-person tactical shooter gameplay`.
Create one Subject1 entry with a concise display name. Its `subject_attribute`
is a global whole-clip field: first state the player's team (`T side` or `CT
side`), then describe only the first/conditioning frame: both visible hands (or
explicitly `not visible`), glove color/material/pattern, the initial held weapon
or item and its visible shape/color/finish, which hand holds or operates it, its
screen anchor, initial stance, and starting viewpoint. Use the trusted initial
item name when provided, but describe appearance only when visible. Do not
mention any later frame, camera turn, movement, action progress, end state,
chunk/segment, or information that appears only after the first frame. Inspect
the large opening sheet before writing this field; if hands, gloves, or the item
are plainly visible, do not replace them with a generic `not resolved` or `not
specified` fallback.
Describe an independently visible other player model in the scene/environment
prose when present, including its left/right/above/below position relative to
`<|Subject1|>`'s current view, pose, equipment, and changes.
Call it `another player` unless identity/team is supplied and visibly supported.
Do not infer enemy/teammate, kills, damage, or intent from color or position.
Inside Action and Move descriptions, refer to the main agent with the exact
`<|Subject1|>` token as the grammatical subject. Every Action must say what
`<|Subject1|>` does; do not make an animation, weapon, or state change the
subject (for example, do not write `the inspect animation begins`). Camera
descriptions must use `<|Subject1|>'s view` as their grammatical subject and
must preserve every observed horizontal, vertical, zoom, or diagonal change.
# Action channels
Keep these channels separate:
- `contained_actions`: concrete player interactions with an object or weapon.
Explain the initial state, which hand/body part manipulates it, contact or
target interaction, transition, and visible result. Locomotion and camera-only
changes do not belong here. The sentence must be player-centered, for example
`<|Subject1|> begins inspecting the Butterfly Knife ...`, rather than
`the Butterfly Knife inspect animation begins`.
- `move_description`: one detailed high-level summary for the whole chunk, with
no `` tags and no `start_time`/`end_time` fields. Use `<|Subject1|>` as
the subject and explicit direction verbs such as `moves forward`, `moves
backward`, `moves right`, `moves left`, or `moves forward-right`. State the
ordered locomotion pattern and add visible positional/context detail, while
preserving every meaningful direction change from the trusted metadata. Do
not force a diagonal, vertical, or backward movement into a generic left/right
label and do not append a coordinate disclaimer.
- `camera_description`: one detailed high-level summary for the whole chunk,
with no `` tags and no `start_time`/`end_time` fields. Use the player's
first-person view as the subject, written as `<|Subject1|>'s view ...`.
Preserve every meaningful yaw/pitch/zoom change from the trusted metadata in
chronological order: express yaw as left/right turns, pitch as up/down tilts,
zoom/FOV as zooming in/out or widening/narrowing, and diagonal combinations
when the evidence supports them. Never rewrite all camera changes as
left/right. Merge adjacent intervals with the same direction into one phrase;
do not mention metadata, intervals, or this merging rule in the output. Do not
expose exact sub-event timestamps or raw control rows. If the trusted timeline
and video do not show a vertical or zoom change, do not invent one.
For weapon switches, state the outgoing and incoming item, the lowering/release,
draw/grip transition, and stable ready hold; mention appearance only when visible.
For reloads, inspect, scope, grenade, objective, fire, and melee, describe the
specific hand/item sequence and before/after state supported by the facts or
frames. Do not infer a hit, kill, damage, detonation, or tactical intent.
## Action identity across chunk boundaries
An Action ID identifies one complete underlying interaction, not one segment row.
If a single reload, inspect animation, scope phase, weapon switch, objective
interaction, or other continuous event crosses a chunk boundary, use the SAME
wrapper ID in every touched segment, for example:
`...first phase...` in chunk 2 and
`...continuing/final phase...` in chunk 3. The prose in each
segment must be freshly written from that segment's local video: do not copy or
lightly paraphrase the preceding Action sentence. Describe the current visible
phase and hand/item state, even when the underlying Action ID is unchanged.
Do not restart numbering at Action1 for each chunk. Assign a new ID only when the
verb, item, target, or independent event genuinely changes. IDs are assigned in
first chronological appearance; repeated IDs for the same cross-boundary event
are intentional. A brief visual occlusion does not split a continuous event.
Each action entry has exactly `start_time`, `end_time`, and `action_description`.
Clip its times to the containing chunk. Action prose should normally be 60-105
English words and must be interaction-focused and player-centered. Move and
Camera entries contain one unwrapped 45-105-word high-level summary each, with no
`start_time`/`end_time` fields; their exact sub-event times remain in deterministic
metadata rather than public prose.
For this detailed labeling pass, expand the four local description channels by
about 30%: write each `environment_description` as approximately 65-80 English
words, each `action_description` as approximately 60-105 English words, and each
`move_description` and `camera_description` as approximately 45-105 English
words. These are descriptive targets, not permission to add unsupported events;
use the extra space for concrete hand/item state, visible geometry, ordered
direction changes, and their visible consequences.
## Camera wording prohibition
Camera prose must be an ordinary concrete observation, not a template disclaimer.
Never write sentences such as:
- `changing the screen framing while revealing or losing only geometry supported by the visual evidence`;
- `the supported view event is recorded`;
- any phrase that says only that evidence is supported without describing the
actual turn, tilt, zoom, or visible result.
Use a factual sentence such as `<|Subject1|>'s view turns left; fixed features
drift right as the turn becomes visible.` when that is what the panels support.
For a vertical change, write `<|Subject1|>'s view tilts up/down`; for zoom, say
that it zooms in/out or widens/narrows when visible. If the view is steady, say
so plainly. Every camera entry has exactly one unwrapped `camera_description`,
and there is exactly one camera entry per chunk.
Every move entry has exactly one unwrapped `move_description`, and there is
exactly one move entry per chunk. Move and Camera entries have no IDs, wrapper
tags, or time fields; Action IDs follow the cross-boundary rule above. Do not
nest wrappers.
# Environment detail
`environment_detail` is the baseline whole-clip environment reference. Fold
the following dimensions into its narrative in a natural order: opening view/FOV
and eye height, foreground/midground/background and left/center/right/above/below
relations relative to `<|Subject1|>`'s view, playable topology and sightlines, concrete
wall/floor/ceiling materials, props/interactables and their state, lighting and
shadows, occlusion or persistent smoke/fire/flash/dust, and any independently
visible player model. Retain the supporting non-HUD environment fields in the
schema and fill them with concrete evidence; write `not visible` instead of
guessing hidden geometry. Do not describe HUD or other interface elements.
The `environment_detail.description` field must be a coherent global summary of
the other environment fields below, not a generic placeholder and not a
chunk-specific timeline. Synthesize the supported map/area, opening view,
topology and sightlines, surfaces/materials, props/interactables,
lighting/visibility, persistent effects, and independently visible player model
information from those fields. Keep it global to the whole clip, avoid
mentioning chunks or later phases, and target roughly 100-140 English words.
Each `environment_description` is a local dynamic-change summary for exactly one
chunk. Treat `environment_detail` as already-known baseline context: do not copy
its static wall/prop/material inventory into every chunk. Instead say what enters,
leaves, appears, disappears, grows nearer, becomes occluded, changes lighting,
or remains stable because of the player's movement, action, or camera. Large
persistent effects must be described consistently in the whole-clip environment,
the relevant chunk description, and the local paragraph. One-frame muzzle
flash/recoil belongs to the action, not a persistent world effect.
# Detail requirement
Each `visual_description_long` must be a distinct, information-dense paragraph
organized as a chronological mini-story: at the start, what changes as the
video progresses, and by the end. In the prose itself, do not write the words
`chunk` or `segment`; use `at the start`, `then`, `as the view progresses`, and
`by the end` instead. Use 220-300 English words in every run (hard minimum 180)
and at least twelve non-redundant concrete clauses.
Interleave, rather than enumerate by category, the required evidence:
foreground/midground/background;
left/center/right/above/below relations relative to `<|Subject1|>`'s view; both
hands, gloves, held item and view anchor;
stance; at least three surfaces, props, or explicit occlusion causes; lighting
and shadows; visible player models; the action before/during/after state; and
every verified movement/camera change. Give the action and camera/movement
progression enough detail that the dedicated channel summaries are not the only
place those changes appear. Do not copy the static environment inventory
verbatim from `environment_detail`. Do not mention frame numbers, timestamps,
seconds, HUD, overlays, or interface elements. Watch each local video
chronologically and do not pad with generic phrases or repeat a neighboring
passage. For a continuous Action crossing multiple chunks, write a fresh,
local-video-specific description in every touched segment; do not copy or
lightly paraphrase the previous Action sentence. Describe the visible phase and
current hand/item state in that local window. If a detail is cropped, occluded, too small, or not legible, say why it is
`not visible`.
# Input facts
The harness substitutes the following blocks before sending the request:
{
"sample_id": "opencs2-switch-static-m2392763-de_ancient-r06-p01-f000020--infer-s0042",
"map_name": "de_ancient",
"profile": "stationary",
"action_reference": "switch",
"player_side": null,
"match_id": "2392763",
"round_id": 6,
"num_frames": 81,
"fps": 16.0,
"duration_seconds": 5.0625,
"width": 832,
"height": 480,
"time_base": "clip_relative",
"conditioning_frame_index": 0,
"content_frame_start": 1,
"content_frame_end_exclusive": 81,
"initial_held_item": "Desert Eagle",
"latent_layout": {
"temporal_stride": 4,
"condition_latent_frames": 1,
"chunk_size_latents": 4,
"num_chunks": 5,
"num_latents": 21,
"chunk_support_tick_offsets": [
2,
66,
130,
194,
258,
322
]
},
"source_fps": 32,
"source_tick_rate": 64,
"source_frame_start": 20,
"source_frame_end_exclusive": 181,
"temporal_resampling": {
"mode": null,
"source_span_frame_count": null,
"source_span_tick_count": null,
"speedup_ratio": null
}
}
[
{
"chunk_index": 0,
"latent_indices": [
1,
2,
3,
4
],
"support_tick_offsets": [
2,
66
],
"start_frame": 1,
"end_frame_exclusive": 17,
"start_seconds": 0.0625,
"end_seconds": 1.0625,
"start_time": "00:00.062",
"end_time": "00:01.062",
"trusted_event_ids": []
},
{
"chunk_index": 1,
"latent_indices": [
5,
6,
7,
8
],
"support_tick_offsets": [
66,
130
],
"start_frame": 17,
"end_frame_exclusive": 33,
"start_seconds": 1.0625,
"end_seconds": 2.0625,
"start_time": "00:01.062",
"end_time": "00:02.062",
"trusted_event_ids": []
},
{
"chunk_index": 2,
"latent_indices": [
9,
10,
11,
12
],
"support_tick_offsets": [
130,
194
],
"start_frame": 33,
"end_frame_exclusive": 49,
"start_seconds": 2.0625,
"end_seconds": 3.0625,
"start_time": "00:02.062",
"end_time": "00:03.062",
"trusted_event_ids": [
"context_00",
"target"
]
},
{
"chunk_index": 3,
"latent_indices": [
13,
14,
15,
16
],
"support_tick_offsets": [
194,
258
],
"start_frame": 49,
"end_frame_exclusive": 65,
"start_seconds": 3.0625,
"end_seconds": 4.0625,
"start_time": "00:03.062",
"end_time": "00:04.062",
"trusted_event_ids": [
"context_00",
"target"
]
},
{
"chunk_index": 4,
"latent_indices": [
17,
18,
19,
20
],
"support_tick_offsets": [
258,
322
],
"start_frame": 65,
"end_frame_exclusive": 81,
"start_seconds": 4.0625,
"end_seconds": 5.0625,
"start_time": "00:04.062",
"end_time": "00:05.062",
"trusted_event_ids": []
}
]
[
{
"event_id": "context_00",
"type": "weapon_switch",
"subtype": "draw",
"weapon": "Desert Eagle",
"from_weapon": "Karambit",
"to_weapon": "Desert Eagle",
"start_seconds": 2.781,
"end_seconds": 3.672,
"start_time": "00:02.781",
"end_time": "00:03.672",
"is_target": true,
"end_reason": "state_change",
"confidence": "high",
"source_segment_id": "s0001",
"time_base": "clip_relative",
"begins_before_clip": false,
"continues_after_clip": false,
"source": "DEM:events.csv"
},
{
"event_id": "target",
"type": "switch",
"subtype": "draw",
"weapon": "Desert Eagle",
"from_weapon": "Karambit",
"to_weapon": "Desert Eagle",
"start_seconds": 2.7812,
"end_seconds": 3.6719,
"start_time": "00:02.781",
"end_time": "00:03.672",
"is_target": true,
"time_base": "clip_relative",
"begins_before_clip": false,
"continues_after_clip": false,
"confidence": "high",
"end_reason": null,
"source_segment_id": "s0001",
"source_segment_ids": [
"s0001"
],
"matched_event_segment_ids": [
"s0001"
],
"start_tick_offset": 178,
"end_tick_offset": 235,
"duration_seconds": 0.890625,
"source": "DEM"
}
]
[
{
"type": "movement",
"movement": "still",
"start_seconds": 0.0,
"end_seconds": 5.0469,
"start_time": "00:00.000",
"end_time": "00:05.047",
"time_base": "clip_relative",
"source": "DEM:ticks"
}
]
Full and foreground contact-sheet images show the chronological clip, and the opening sheet shows the conditioning and first content views at larger scale for both hands, gloves, initial item/weapon, and opening view. Each of the five causal chunks has one exact local MP4 containing all 16 frames in that one-second window. Watch each local video from beginning to end; the web viewer may loop it for inspection, but looping does not imply a repeated action. Any left/right position means the viewer's screen-left/screen-right, not the agent's or map's orientation. Do not mention media frame numbers in prose.
# Required output
Return only one raw JSON object with exactly one top-level key, `english_json`.
Do not return `metadata`, `chunk_captions`, `chinese_json`, Markdown, a code fence,
explanation, confidence prose, or extra keys. The harness restores deterministic
metadata and control arrays after parsing. The object shape is:
```json
{
"english_json": {
"perspective_view": "first-person",
"Style": "video game, first-person tactical shooter gameplay",
"subject_list": [
{
"subject_id": "<|Subject1|>",
"display_name": "player-controlled CS2 agent",
"subject_attribute": "The player is on the T side. At the first frame, describe only the visible hands, gloves, initial item, holding hand, screen anchor, stance, and viewpoint."
}
],
"environment_detail": {
"map_name": "...",
"description": "Merged whole-clip environment and visible player-model reconstruction.",
"area_or_callout": "...",
"opening_view": "...",
"layout_and_sightlines": "...",
"surfaces_and_materials": "...",
"props_and_interactables": "...",
"lighting_and_visibility": "...",
"persistent_world_effects": "..."
},
"segments": [
{
"start_time": "00:00.062",
"end_time": "00:01.062",
"visual_description_long": "Detailed local paragraph with concrete evidence.",
"environment_description": "Merged local subject and environment progression.",
"description_change": null,
"contained_actions": [
{
"start_time": "00:00.234",
"end_time": "00:01.000",
"action_description": "<|Subject1|> performs the supported interaction in detail."
}
],
"move_description": [
{
"move_description": "<|Subject1|> moves forward, then moves right along the visible corridor, with nearby geometry shifting as the position changes."
}
],
"camera_description": [
{
"camera_description": "<|Subject1|>'s view turns left, shifting fixed stonework right, then turns back right and brings the doorway toward center. If the view tilts or zooms, describe that observed up/down or in/out change explicitly."
}
]
}
]
}
}
```
Return exactly five segment objects in chunk order. Copy each fixed chunk time,
set `description_change` to null, and keep all three channel arrays present. Each
segment must contain exactly one unwrapped Move row and one unwrapped Camera row
with no time fields, plus only the contained Action rows supported by the
interaction.
The harness will clip and canonicalize Action times, restore authoritative
metadata, and preserve one Action ID for every complete cross-boundary event.
Before sending, check that all objects and arrays are closed and that the JSON is
complete.
# Multi-round context protocol
The runner delivers the evidence over seven user turns while keeping one context
window. Round 0 contains only one conditioning image and asks for a structured
global-state draft. Rounds 1 through 5 each append exactly one new local chunk
attachment (`chunk0` through `chunk4`) and ask for a structured draft for that
chunk. These drafts may revise the current subject list and viewpoint when later
evidence is stronger; they must not use information from future chunks.
Round 6 is text-only. It carries the complete accumulated history and asks for
the final production object. Reconcile all previous drafts, preserve the fixed
five chunk boundaries, and return only one object with the top-level
`english_json` key. Do not return `READY`, intermediate draft keys, or any new
visual attachment in round 6. Never treat the conditioning draft as a sixth
content segment, and never reset the accumulated context.
Before returning, check that exactly five segment objects are present, every segment boundary is copied exactly, all three channel arrays are present, and the response contains only the top-level `english_json` key.
```