Stop Copying Faceless YouTube Channels! Build Their System Instead — Transcript

Full transcript

1. 0:00Look at this faceless YouTube channel,
2. 0:02Neo. Instead of copying its topics or
3. 0:04visual style, I'm going to extract the
4. 0:07system underneath it and use that to
5. 0:09build a completely original faceless
6. 0:11video. And once it's built, we can reuse
7. 0:14the same system for the next video, too.
8. 0:16We'll use Neo's latest videos to learn
9. 0:18the storytelling structure, fresh web
10. 0:21research for the facts, then turn the
11. 0:23script into a custom visual plan, and
12. 0:25batch produce the scenes in Flow.
13. 0:27[music] And this is the result we're
14. 0:29building.
15. 0:30The reference videos shape the
16. 0:31structure, not the facts. And this kind
17. 0:34of video engine can be customized for
18. 0:36any channel you want to study. Let's
19. 0:38build it. Start by creating a new
20. 0:40notebook LM notebook. I'm naming this
21. 0:42one Neolike script engine. Now open Neo
22. 0:46and take the latest five videos. I'm
23. 0:49adding all five YouTube URLs to this
24. 0:51notebook as sources. These videos have
25. 0:54one job. Teach the engine how this kind
26. 0:56of channel structures a story.
27. 0:59Notebook LM can use the YouTube sources
28. 1:01to understand things like the hook,
29. 1:03pacing, information density, section
30. 1:06flow, curiosity, resets, transitions,
31. 1:09and how the story resolves. But I do not
32. 1:12want it using those videos as factual
33. 1:15sources for our new video. So, open
34. 1:17configure chat, switch to custom, and
35. 1:20paste the reference structured script
36. 1:22engine prompt. The important rule inside
37. 1:25the prompt is simple. The YouTube
38. 1:27sources shape how the story is told. The
39. 1:30web sources determine what factual
40. 1:33content can be said. Now, type, give me
41. 1:36five original topic ideas with relevant
42. 1:38web search keywords. Notebook LM gives
43. 1:41us five original directions based on the
44. 1:44broad storytelling patterns in the
45. 1:45reference videos. Each idea includes a
46. 1:48working title, a promise, a central
47. 1:51curiosity gap, why it fits the format,
48. 1:54and search keywords we can use for
49. 1:55research. For this example, I'm choosing
50. 1:58idea 2. The island that changes
51. 2:01countries every 6 months. [music] The
52. 2:04topic is pheasant island, and notebook
53. 2:06LM gives us several research queries.
54. 2:09I'm taking the first one, Feeasant
55. 2:12Island, France, Spain sovereignty
56. 2:14treaty. Now use Notebook LM's web search
57. 2:17with fast research. Paste the query. Let
58. 2:20notebook LM discover relevant sources.
59. 2:23Review the results and import the useful
60. 2:25ones. Now the same notebook contains two
61. 2:28different kinds of sources. The five Neo
62. 2:31videos for structure and the new web
63. 2:33sources for facts. With the research
64. 2:36added, I type write a 750word script for
65. 2:39the idea too. Notebook. LM now writes
66. 2:43the original narration. It can use the
67. 2:45reference videos for highle storytelling
68. 2:47structure, but the actual facts in this
69. 2:50new script have to come from the web
70. 2:51sources we just added. And at the end,
71. 2:54it gives us a source check showing which
72. 2:56factual sources support the major parts
73. 2:58of the script. So, step one gives us a
74. 3:02researched original script while still
75. 3:04learning from the structure of a proven
76. 3:06format. Quick flash forward. This is
77. 3:09where the engine is taking us. By the
78. 3:11end, these research notes become a
79. 3:13finished faceless documentary with maps,
80. 3:16sectional scenes, and motion graphics
81. 3:18like this. Let's continue with the
82. 3:20visual plan. I've already built the
83. 3:23visual planner, but there are two
84. 3:24sections that change depending on the
85. 3:26reference channel. Open the prompt. The
86. 3:30first is called visual types for this
87. 3:32project. For Neo, I defined the
88. 3:35recurring styles I saw while reviewing
89. 3:37the videos. cinematic infrastructure, 3D
90. 3:40map and geography, 3D sectional
91. 3:42reconstruction, diagram or mechanical
92. 3:45cutaway, and infographic motion. And I'm
93. 3:48not just giving chat GPT the names. For
94. 3:51every type, I explain when to use it,
95. 3:54what the visual should look like, the
96. 3:56typical motion, and what to avoid. For
97. 3:59example, for sectional views, I use two
98. 4:02simple options. C1 shows the whole
99. 4:04structure so you can understand how
100. 4:06everything fits together. C2 zooms in on
101. 4:10one hidden area so you can clearly see
102. 4:12what's inside. That gives the planner a
103. 4:15much clearer target than just asking for
104. 4:17a 3D cutaway. The second customizable
105. 4:20section is the optional target scene
106. 4:22type mix. For this neo inpired version,
107. 4:26I'm using a rough balance between
108. 4:27cinematic scenes, maps, sectional
109. 4:30reconstructions, mechanical cutaways,
110. 4:33and infographic motion. These
111. 4:35percentages are only a guide. If you are
112. 4:38building around another reference
113. 4:40channel, this is the part you change.
114. 4:43Watch two or three of their videos,
115. 4:45identify the recurring visual types,
116. 4:47[music] describe them in the same
117. 4:49format, and adjust the mix if necessary.
118. 4:52Everything else in the planner can stay
119. 4:54the same. Now I copy the complete visual
120. 4:57planner prompt and open a new chat GPT
121. 4:59conversation. Before I run it, I go back
122. 5:02to notebook LM and copy the complete
123. 5:05script we created in step one. Paste
124. 5:08that narration into the script section
125. 5:09at the bottom of the prompt and submit
126. 5:11it. [music] Chat GPT first creates an
127. 5:14adapted visual bible so the different
128. 5:17scene types still belong to the same
129. 5:19visual world. Then it creates the batch
130. 5:22visual plan. For every scene, I get the
131. 5:24matching voice over, the visual type,
132. 5:27framing mode, a complete flow, batch
133. 5:29video prompt, and an optional editor
134. 5:32overlay for exact text or labels. And if
135. 5:35you look down the visual type column,
136. 5:37you can see the different scene types
137. 5:39spread throughout the script. Maps in
138. 5:41some rows, sectional views in others,
139. 5:44cinematic scenes, cutaways, and
140. 5:46infographic motion. That is exactly what
141. 5:49the scene type mix was supposed to do.
142. 5:51Give us variety across the video instead
143. 5:53of repeating the same kind of visual
144. 5:55over and over. So once this table is
145. 5:58finished, the next step is simply
146. 6:00turning these prompts into scenes. Now
147. 6:03open Google Flow. This is where the
148. 6:05batch ready plan starts to pay off. For
149. 6:08this demonstration, I'm using scenes 31
150. 6:11through 37 from the visual plan. I
151. 6:14picked these seven scenes on purpose
152. 6:16because they include different visual
153. 6:18types. [music] Cinematic infrastructure,
154. 6:20infographic motion, and a 3D map so we
155. 6:24can see whether the planner's visual mix
156. 6:26also works in production. Instead of
157. 6:28generating them one by one, I'm going to
158. 6:31use flow's agent. To activate it, I
159. 6:34click the agent button in flow. Then I
160. 6:36copy the flow prompts for scenes 31 to
161. 6:3937 and type create these videos and
162. 6:43paste all seven scene prompts
163. 6:45underneath. Notice that every scene is
164. 6:48completely standalone. There are no
165. 6:50character reference images. None of
166. 6:53these clips depends on the previous
167. 6:54scene and there is no first frame or end
168. 6:57frame chain. Each prompt already
169. 7:00contains the subject style, framing,
170. 7:02camera movement and motion it needs. Now
171. 7:06send the request. Flo recognizes that
172. 7:09I'm asking for seven separate video
173. 7:11generations and asks me to approve the
174. 7:13batch. I approve it and all seven scenes
175. 7:17start generating together. This is the
176. 7:19big advantage of planning the video this
177. 7:21way. Instead of manually building one
178. 7:24scene, waiting then moving to the next
179. 7:27one. I can send a whole group of scenes
180. 7:29into production at once. Once the
181. 7:32generations are finished, I download the
182. 7:34finished clips. Now we can edit and
183. 7:37finish our video. Before we make the
184. 7:39final edits and see the finished video,
185. 7:41I wanted to share something quickly. My
186. 7:44faceless YouTube engine guide walks you
187. 7:46through the complete nine-stage
188. 7:48production pipeline built specifically
189. 7:50for demonetization safe faceless videos.
190. 7:53Every chapter has a checklist, homework,
191. 7:56and a real sample video built from
192. 7:58scratch so you can follow every step.
193. 8:01already making videos. This will 10
194. 8:03times your speed and quality. Brand new?
195. 8:06You'll have your first video done by the
196. 8:08end. [music] Link in the description.
197. 8:10The last step is putting everything
198. 8:12together. I'm using Cap Cut here, but
199. 8:15you can use any editor you're
200. 8:17comfortable with. I'm starting with an
201. 8:19empty timeline, so we can build the
202. 8:21final video from scratch. First voice
203. 8:24over. I go back to Notebook LM and copy
204. 8:27the part of the narration that covers
205. 8:28scenes 31 through 37. Then open Google
206. 8:32AI Studio. Go to generate speech and
207. 8:35paste the script. For this video, I'm
208. 8:38using the Enzo voice. I specifically
209. 8:41chose a lower pitch voice because of the
210. 8:42niche we selected. This geography and
211. 8:45history documentary style feels more
212. 8:47serious and grounded, so a lower voice
213. 8:50fits the tone better. Generate the
214. 8:52narration and download the audio.
215. 8:55Next, music. Open the YouTube Studio
216. 8:58audio library and find a background
217. 9:00track that supports the documentary
218. 9:02without competing with the voice over.
219. 9:05Download the track, then return to Cap
220. 9:07Cut. The voice over goes on the timeline
221. 9:10first. After that, import the flow clips
222. 9:13and the music. Now, I use the visual
223. 9:15plan as my assembly guide. If I need to
224. 9:18check what belongs to a particular line,
225. 9:20I can jump back to the scene table, find
226. 9:23the matching scene, then place that clip
227. 9:25over the correct part of the narration.
228. 9:27Because we already decided the visuals
229. 9:29in step two, I'm not inventing the video
230. 9:32again during editing. I'm mostly
231. 9:34assembling what the engine already
232. 9:36planned. Arranging these seven scenes
233. 9:38took me about 5 minutes. So, at the same
234. 9:41pace, the full video could be edited in
235. 9:43roughly 30 minutes. Then I add the music
236. 9:46underneath, lower the volume so it sits
237. 9:48behind the narration and make any timing
238. 9:50adjustments. Once everything is synced,
239. 9:53I preview the sequence and make the
240. 9:55final adjustments.
241. 9:57Finally, I export. Let's see the result.
242. 10:01For 3 months, the French chief minister,
243. 10:03Cardinal Maserin, and the Spanish
244. 10:05diplomat Luis Mendesa Otto met in
245. 10:08temporary wooden pavilions built on the
246. 10:10island across [music] 24 distinct
247. 10:12diplomatic conferences.
248. 10:15>> [music]
249. 10:16>> Out of those intense negotiations came
250. 10:19the Treaty of the Pyrees signed on
251. 10:21November 7th, 1659. [music]
252. 10:24To cement the peace, the treaty arranged
253. 10:26a highstakes [music] royal marriage
254. 10:28between King Louis the 14th of France
255. 10:30and Maria Teresa of Spain, the daughter
256. 10:33of King Philip IV. On [music] this very
257. 10:35island, the Spanish princess bade
258. 10:38farewell to her father in court before
259. 10:40crossing into France to become its
260. 10:41queen.
261. 10:44If you want me to build this video
262. 10:46engine for another documentary format,
263. 10:48comment continue and tell me the style
264. 10:51or niche. And if you want to see another
265. 10:53complete faceless workflow, click the
266. 10:56video on your screen now. I will see you
267. 10:58there.