academic-presentations

Name: academic-presentations
Author: Boom5426

88installs28forks

How to install academic-presentations?

boom5426/nature-paper-skills/academic-presentations

$npx -y skills add Boom5426/Nature-Paper-Skills --skill academic-presentations

Installs into the current project.

Summary

Create academic presentation slide decks and optionally demo videos from research papers. Use when the user asks to "make slides", "create a deck", "make a presentation", "demo video", "paper slides", "conference talk slides", or wants to turn a paper into a visual presentation.

Files · 1

View on GitHub

SKILL.md

1	# Academic Presentations
2
3	Produce slide decks (and optionally narrated demo videos) from research papers. The human drives all outline and visual decisions — the agent executes.
4
5	## Pipeline
6
7	```
8	[1] Script Draft ──→ [2] Slide Generation ──→ [3] TTS Audio (optional) ──→ [4] Video Assembly (optional)
9	Claude Code nanobanana /edit edge-tts / Kokoro / ElevenLabs ffmpeg
10	```
11
12	Skip stages 3–4 for slide-only output. User can enter at any stage.
13
14	## Stage 1: Script / Outline
15
16	Input: paper + user-provided outline or slide plan
17	Output: `video-scripts.md` or `slide-outline.md` — per-slide content with talking points
18
19	The agent drafts scripts based on the user's outline. The user owns the structure — agent does not decide slide count, order, or what to emphasize.
20
21	## Stage 2: Slide Generation
22
23	> Full reference: [references/slide-generation.md](references/slide-generation.md)
24
25	Tool: nanobanana (Gemini CLI extension)
26
27	Priority order (edit-first):
28	1. Has paper figure → nanobanana `/edit` to wrap into slide frame
29	2. Has existing slide → `/edit` to adapt
30	3. User-provided reference (e.g., from NotebookLM or PPTX the user made) → `/edit` to refine
31	4. Title slide from scratch → generate with academic style prompt
32	5. Content slide from scratch → generate with deck-style preamble
33
34	Key principle: prefer `/edit` on existing HQ paper figures over generating from scratch.
35
36	Deck style: create `deck-style.md` once per deck, prepend to all generate-from-scratch prompts. For `/edit`, style is inherited from the base image.
37
38	Example `deck-style.md`:
39	```markdown
40	- Canvas: 1920x1080, white background
41	- Accent: #2563EB blue, text: #1e293b dark slate
42	- Clean sans-serif, flat design, no gradients/shadows
43	- Bottom bar: blue accent with white affiliation text
44	```
45
46	## Stage 3: TTS Audio (optional)
47
48	> Full reference: [references/tts-engines.md](references/tts-engines.md)
49	> Batch scripts: [scripts/batch_tts_edge.py](scripts/batch_tts_edge.py), [scripts/batch_tts_kokoro.py](scripts/batch_tts_kokoro.py)
50
51	Output: one audio file per narrated slide
52
53	### Engine Selection
54
55	\| Engine \| Quality \| Cost \| Latency \| Best For \|
56	\|--------\|---------\|------\|---------\|----------\|
57	\| edge-tts (default) \| Very good \| Free, unlimited \| ~6s/slide (cloud) \| Quick generation, good male voices \|
58	\| Kokoro \| Very good \| Free, unlimited \| ~1.5s/slide (local) \| Offline use, fast batch, good female voices \|
59	\| ElevenLabs \| Premium \| 10k chars free/mo \| ~3s/slide (cloud) \| Highest quality, voice cloning \|
60
61	Default: Use edge-tts unless user requests offline or premium quality.
62
63	### Quick Start (edge-tts)
64
65	```python
66	import edge_tts, asyncio
67
68	async def tts_slide(text, output, voice="en-US-AndrewNeural"):
69	await edge_tts.Communicate(text, voice).save(output)
70
71	asyncio.run(tts_slide("Your slide text here", "slide_01.mp3"))
72	```
73
74	Voices: AndrewNeural (male, presenter), AriaNeural (female), GuyNeural (male, warm), JennyNeural (female, pro)
75
76	## Stage 4: Video Assembly (optional)
77
78	Tool: ffmpeg
79	Input: slide PNGs + audio files + optional demo recording
80
81	```bash
82	# Use symlink to avoid iCloud path spaces: ln -sfn "long path" /tmp/workdir
83
84	# Slide with audio:
85	ffmpeg -y -loop 1 -i slide.png -i audio.mp3 \
86	-c:v libx264 -tune stillimage -pix_fmt yuv420p \
87	-c:a aac -ar 44100 -ac 2 -shortest seg.mp4
88
89	# Silent slide (N seconds):
90	ffmpeg -y -loop 1 -i slide.png -f lavfi -i anullsrc=r=44100:cl=stereo \
91	-c:v libx264 -tune stillimage -pix_fmt yuv420p \
92	-c:a aac -ar 44100 -ac 2 -t N seg.mp4
93
94	# Concat (always re-encode, never -c copy):
95	printf "file 'seg1.mp4'\nfile 'seg2.mp4'\n..." > concat.txt
96	ffmpeg -y -f concat -safe 0 -i concat.txt \
97	-c:v libx264 -pix_fmt yuv420p -c:a aac -ar 44100 -ac 2 final.mp4
98	```
99
100	All segments MUST share: 44100Hz sample rate, stereo, AAC codec.
101
102	## PPTX Conversion (if needed)
103
104	> Full reference: [references/pptx-conversion.md](references/pptx-conversion.md)
105
106	If starting from an existing PPTX, convert slides to PNG images first:
107	```bash
108	soffice --headless --convert-to pdf --outdir output/ presentation.pptx
109	pdftoppm -png -r 300 output/presentation.pdf output/slide
110	```
111
112	## NotebookLM — Human Reference Only
113
114	The agent must NOT auto-invoke NotebookLM or use its outputs to drive slide/script decisions. The human owns the outline, visual arrangement, and deck direction.
115
116	When to recommend: only when the user says they're unsure what to put on slides or need inspiration.
117
118	## Gotchas
119
120	- iCloud paths with spaces break ffmpeg — symlink to `/tmp/`
121	- **Audio forma

People who liked this also viewed

6 picks

Type

academic-presentations

88installs28forks

View on GitHub

How to install academic-presentations?

boom5426/nature-paper-skills/academic-presentations

$npx -y skills add Boom5426/Nature-Paper-Skills --skill academic-presentations

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Academic Presentations
2
3	Produce slide decks (and optionally narrated demo videos) from research papers. The human drives all outline and visual decisions — the agent executes.
4
5	## Pipeline
6
7	```
8	[1] Script Draft ──→ [2] Slide Generation ──→ [3] TTS Audio (optional) ──→ [4] Video Assembly (optional)
9	Claude Code nanobanana /edit edge-tts / Kokoro / ElevenLabs ffmpeg
10	```
11
12	Skip stages 3–4 for slide-only output. User can enter at any stage.
13
14	## Stage 1: Script / Outline
15
16	Input: paper + user-provided outline or slide plan
17	Output: `video-scripts.md` or `slide-outline.md` — per-slide content with talking points
18
19	The agent drafts scripts based on the user's outline. The user owns the structure — agent does not decide slide count, order, or what to emphasize.
20
21	## Stage 2: Slide Generation
22
23	> Full reference: [references/slide-generation.md](references/slide-generation.md)
24
25	Tool: nanobanana (Gemini CLI extension)
26
27	Priority order (edit-first):
28	1. Has paper figure → nanobanana `/edit` to wrap into slide frame
29	2. Has existing slide → `/edit` to adapt
30	3. User-provided reference (e.g., from NotebookLM or PPTX the user made) → `/edit` to refine
31	4. Title slide from scratch → generate with academic style prompt
32	5. Content slide from scratch → generate with deck-style preamble
33
34	Key principle: prefer `/edit` on existing HQ paper figures over generating from scratch.
35
36	Deck style: create `deck-style.md` once per deck, prepend to all generate-from-scratch prompts. For `/edit`, style is inherited from the base image.
37
38	Example `deck-style.md`:
39	```markdown
40	- Canvas: 1920x1080, white background
41	- Accent: #2563EB blue, text: #1e293b dark slate
42	- Clean sans-serif, flat design, no gradients/shadows
43	- Bottom bar: blue accent with white affiliation text
44	```
45
46	## Stage 3: TTS Audio (optional)
47
48	> Full reference: [references/tts-engines.md](references/tts-engines.md)
49	> Batch scripts: [scripts/batch_tts_edge.py](scripts/batch_tts_edge.py), [scripts/batch_tts_kokoro.py](scripts/batch_tts_kokoro.py)
50
51	Output: one audio file per narrated slide
52
53	### Engine Selection
54
55	\| Engine \| Quality \| Cost \| Latency \| Best For \|
56	\|--------\|---------\|------\|---------\|----------\|
57	\| edge-tts (default) \| Very good \| Free, unlimited \| ~6s/slide (cloud) \| Quick generation, good male voices \|
58	\| Kokoro \| Very good \| Free, unlimited \| ~1.5s/slide (local) \| Offline use, fast batch, good female voices \|
59	\| ElevenLabs \| Premium \| 10k chars free/mo \| ~3s/slide (cloud) \| Highest quality, voice cloning \|
60
61	Default: Use edge-tts unless user requests offline or premium quality.
62
63	### Quick Start (edge-tts)
64
65	```python
66	import edge_tts, asyncio
67
68	async def tts_slide(text, output, voice="en-US-AndrewNeural"):
69	await edge_tts.Communicate(text, voice).save(output)
70
71	asyncio.run(tts_slide("Your slide text here", "slide_01.mp3"))
72	```
73
74	Voices: AndrewNeural (male, presenter), AriaNeural (female), GuyNeural (male, warm), JennyNeural (female, pro)
75
76	## Stage 4: Video Assembly (optional)
77
78	Tool: ffmpeg
79	Input: slide PNGs + audio files + optional demo recording
80
81	```bash
82	# Use symlink to avoid iCloud path spaces: ln -sfn "long path" /tmp/workdir
83
84	# Slide with audio:
85	ffmpeg -y -loop 1 -i slide.png -i audio.mp3 \
86	-c:v libx264 -tune stillimage -pix_fmt yuv420p \
87	-c:a aac -ar 44100 -ac 2 -shortest seg.mp4
88
89	# Silent slide (N seconds):
90	ffmpeg -y -loop 1 -i slide.png -f lavfi -i anullsrc=r=44100:cl=stereo \
91	-c:v libx264 -tune stillimage -pix_fmt yuv420p \
92	-c:a aac -ar 44100 -ac 2 -t N seg.mp4
93
94	# Concat (always re-encode, never -c copy):
95	printf "file 'seg1.mp4'\nfile 'seg2.mp4'\n..." > concat.txt
96	ffmpeg -y -f concat -safe 0 -i concat.txt \
97	-c:v libx264 -pix_fmt yuv420p -c:a aac -ar 44100 -ac 2 final.mp4
98	```
99
100	All segments MUST share: 44100Hz sample rate, stereo, AAC codec.
101
102	## PPTX Conversion (if needed)
103
104	> Full reference: [references/pptx-conversion.md](references/pptx-conversion.md)
105
106	If starting from an existing PPTX, convert slides to PNG images first:
107	```bash
108	soffice --headless --convert-to pdf --outdir output/ presentation.pptx
109	pdftoppm -png -r 300 output/presentation.pdf output/slide
110	```
111
112	## NotebookLM — Human Reference Only
113
114	The agent must NOT auto-invoke NotebookLM or use its outputs to drive slide/script decisions. The human owns the outline, visual arrangement, and deck direction.
115
116	When to recommend: only when the user says they're unsure what to put on slides or need inspiration.
117
118	## Gotchas
119
120	- iCloud paths with spaces break ffmpeg — symlink to `/tmp/`
121	- **Audio forma

People who liked this also viewed

6 picks

Type