YouTube2Text

How to Create High Quality AI Videos With Claude — Transcript

by Parker Prompts · 2,011 words · 290 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Making AI videos just got a whole lot
  2. 0:01simpler because you can now run the
  3. 0:03entire thing from a single Claude chat.
  4. 0:05You describe what you want and Claude
  5. 0:07generates it in the background using
  6. 0:08Higgs Field.
  7. 0:28I used it to build this full cinematic
  8. 0:30underwater film and I'm going to walk
  9. 0:32you through the whole thing. If you've
  10. 0:33ever made AI videos, you know the
  11. 0:35struggle. You write prompts in one tab,
  12. 0:37generate in another, download,
  13. 0:39re-upload, and half your session is just
  14. 0:41moving files between tools. Higgs Field
  15. 0:43fixed that with something called MCP,
  16. 0:45which basically lets Claude plug
  17. 0:46straight into Higgs Field so Claude can
  18. 0:48generate images and videos itself right
  19. 0:50in the chat. MCP stands for Model
  20. 0:52Context Protocol and in plain terms,
  21. 0:54it's a standard way for AI tools to talk
  22. 0:55to each other. So once the connection is
  23. 0:57established, Claude can see Higgs
  24. 0:58Field's tools and run them as if they
  25. 1:00were its own. And the setup takes about
  26. 1:022 minutes, so let's get it done. If you
  27. 1:03want to follow along, the Higgs Field
  28. 1:05link is in the description. The page
  29. 1:06shows you the setup steps along with the
  30. 1:08exact link to copy. [music] Now over in
  31. 1:10Claude, go to settings, then connectors,
  32. 1:12and click the little plus icon to add a
  33. 1:14custom connector. Name it Higgs Field,
  34. 1:16paste that link in, and hit add. Once
  35. 1:18it's connected, a permissions panel pops
  36. 1:20up where you pick which tools Claude is
  37. 1:21allowed to use. And at minimum, you want
  38. 1:23generate image and generate video turned
  39. 1:25on. And the full list goes way past
  40. 1:27those two. [music] There's background
  41. 1:28removal, reframing, presets, character
  42. 1:30tools, nearly everything the platform
  43. 1:32can do, all callable from the chat.
  44. 1:34[music] I'd keep that list tight, too,
  45. 1:35because every tool you enable is
  46. 1:37something Claude can run on its own. So
  47. 1:39only switch on what you actually plan to
  48. 1:40use. Now let's test it and I'll paste in
  49. 1:43this prompt. And when it runs, you
  50. 1:44actually see the Higgs Field job appear
  51. 1:46right inside the chat. It processes
  52. 1:48there and the finished image drops into
  53. 1:49the conversation like any other reply.
  54. 1:51We get a golden retriever with full
  55. 1:53eagle wings flying above the clouds with
  56. 1:54that cinematic blue tint I asked for in
  57. 1:564K. And notice I never opened Higgs
  58. 1:58Field because the prompt itself
  59. 2:00specifies which model to use. And if you
  60. 2:02break it down, that one sentence gave
  61. 2:03six separate instructions. The subject,
  62. 2:05the color grade, the shot style, the
  63. 2:07model, the quality, [music] and the
  64. 2:08aspect ratio. I made the test 4x3 on
  65. 2:10purpose, too, because ratio control is
  66. 2:12the easiest thing to check at a glance.
  67. 2:14And the film we're making next is 16x9,
  68. 2:16so switching between them is nothing
  69. 2:18more than changing the words in the
  70. 2:19sentence. [music] And that's the core of
  71. 2:20this whole workflow. Your settings go
  72. 2:22inside the sentence, so instead of
  73. 2:23clicking through menus, you just say
  74. 2:24what you want, the model, the quality,
  75. 2:26the resolution, the ratio, and Claude
  76. 2:28handles the rest. But a test image is
  77. 2:30the small [music] stuff, because the
  78. 2:31real reason we connected these two is to
  79. 2:33make an actual film. By the way, if you
  80. 2:35haven't opened Higgs Field yet, I'd
  81. 2:36really recommend opening it from the
  82. 2:38link in the description and following
  83. 2:39these steps along with me, because it
  84. 2:41makes the whole thing much easier to do.
  85. 2:43And you'll walk away from this video
  86. 2:44with a finished film of your own. We're
  87. 2:45making a 36-second underwater short film
  88. 2:48in first-person view with three videos
  89. 2:50of 12 seconds each. And the story runs
  90. 2:52from calm exploration to a chase to a
  91. 2:53discovery at the end. And first person
  92. 2:55is a deliberate choice because the
  93. 2:57diver's hands are in every single shot,
  94. 2:59and that's exactly what makes them easy
  95. 3:00to keep consistent. And the whole thing,
  96. 3:02from the concept to the final footage,
  97. 3:04never leaves the conversation. And
  98. 3:05instead of prompting each piece one at a
  99. 3:07time, I'm giving Claude the entire
  100. 3:08production plan in one message, so I'll
  101. 3:10paste this in. It's a long one, but the
  102. 3:12structure is simple. It splits the job
  103. 3:13into three stages: the reference frame,
  104. 3:15the sequence breakdown, and the
  105. 3:17generation. And it tells Claude to stop
  106. 3:19and wait for my approval after every
  107. 3:20stage. Those approval stops matter
  108. 3:22because without them, Claude would run
  109. 3:24all three generations in one go. And if
  110. 3:26the first video is off, everything after
  111. 3:27it is off, too. And you can reuse this
  112. 3:29exact structure for any film you want to
  113. 3:31make. Keep the three stages and the
  114. 3:32approval stops, and just swap out the
  115. 3:34story in the middle. Stage one is the
  116. 3:36reference frame, and this is the film's
  117. 3:37consistency trick. Before any video gets
  118. 3:39made, Claude generates one image of the
  119. 3:41diver's hands and gloves underwater, and
  120. 3:43that exact image becomes the anchor for
  121. 3:45all three videos. So the gloves, the
  122. 3:47equipment, and the water tone stay
  123. 3:48identical from start to to And the
  124. 3:50prompt specifically says this is not a
  125. 3:52scene, just the hands and gloves against
  126. 3:54open water." Because an anchor works
  127. 3:55best when there's nothing in it except
  128. 3:57the things that need to stay consistent.
  129. 3:59The first version comes back honestly
  130. 4:0090% there. The gloves, the water tone,
  131. 4:03and the light shafts all look right. But
  132. 4:05there's a dive watch on the left wrist
  133. 4:06that I don't want. So I'll tell it to
  134. 4:08remove the watch and put something else
  135. 4:10on his left hand. And the second version
  136. 4:11comes back with the torch in his left
  137. 4:13hand and the beam cutting through the
  138. 4:14water. This frame is exactly what an
  139. 4:16anchor should be. The charcoal gloves,
  140. 4:18the torch with its beam visible, deep
  141. 4:20teal water with light shafts coming down
  142. 4:22from above. Simple enough to match in
  143. 4:23every shot and specific enough that you
  144. 4:25notice if it drifted. So I'll approve it
  145. 4:27and tell Claude to move to stage two.
  146. 4:29Now Claude writes the full prompt
  147. 4:30[music] for each video and shows me all
  148. 4:32three together before generating
  149. 4:33anything. So I get to approve the shape
  150. 4:35of the whole film at once instead of
  151. 4:37finding problems one generation at a
  152. 4:39time. And these prompts are worth a
  153. 4:40closer look because every single one
  154. 4:42starts with the exact same description
  155. 4:44of the hands, the charcoal gloves, the
  156. 4:45wet suit sleeves, the torch with the
  157. 4:47yellow grip ring. And that repeated
  158. 4:48block is what keeps the diver identical
  159. 4:50across all three videos. Then each video
  160. 4:52is broken into five numbered shots with
  161. 4:54a time on each one. Two seconds here,
  162. 4:56three seconds there, so every 12-second
  163. 4:57video is really a five-shot sequence
  164. 4:59with its own edit already built in. Take
  165. 5:01the first one. Shot one pushes through a
  166. 5:03rock arch, shot two tilts up into a
  167. 5:05spiraling school of silverfish, shot
  168. 5:07three grazes a sea fan coral, shot four
  169. 5:09glides through a field of jellyfish, and
  170. 5:11shot five brings in the sea turtle and
  171. 5:12that dark shape at the end. All time
  172. 5:14down to the second. And one more detail
  173. 5:16in the master prompt is doing a lot of
  174. 5:17work here. Each video has to end on a
  175. 5:19frame that cuts into the next one and
  176. 5:20that's [music] what makes three separate
  177. 5:22generations feel like one continuous
  178. 5:23film instead of three pieces stuck
  179. 5:25together. And Claude closes every one of
  180. 5:27them with the same color grade line,
  181. 5:28deep teal blue, cinematic, which is the
  182. 5:30third consistency layer after the hands
  183. 5:32and the settings. [music] All three run
  184. 5:34on SeaDance 2.0, 12 seconds each, 720p,
  185. 5:3716 by 9 with ambient sound and Foley
  186. 5:39only, no music. So the only sound in the
  187. 5:41raw footage is the regulator breathing,
  188. 5:42the bubbles, [music] and the water
  189. 5:43itself because the music gets added in
  190. 5:45the edit where we control it scene by
  191. 5:47scene. And the 720p is on purpose, too,
  192. 5:49because the last step of this workflow
  193. 5:51upscales everything to 4K. So, there's
  194. 5:53no reason to spend extra credits on
  195. 5:55resolution now. The prompts all look
  196. 5:56good, so I'll approve them, and Claude
  197. 5:58starts generating. Video one first, then
  198. 6:00it shows me the result and waits, then
  199. 6:01video two, then video three. That pause
  200. 6:03between generations runs on the same
  201. 6:05logic as the approval stops. Every video
  202. 6:07costs credits, so you check the one that
  203. 6:08just finished before the next one spends
  204. 6:10anything. And the three come back
  205. 6:11exactly on story. Calm exploration with
  206. 6:13a sea turtle and a dark shape drifting
  207. 6:15past at the end. Then the chase through
  208. 6:17the coral tunnel, then the chest on the
  209. 6:18seafloor that does not stay shut. The
  210. 6:20gloves, the torch, and the water tone
  211. 6:22stay the same through all three, which
  212. 6:23is exactly what that reference frame was
  213. 6:25for. And the quality is honestly right
  214. 6:27up there with generating on the platform
  215. 6:29directly, which makes sense, because
  216. 6:30Claude isn't generating anything itself.
  217. 6:32It's running the same C-DANs 2.0 you'd
  218. 6:34use in Higgs Field. Now, there's one
  219. 6:36part of this that Claude can't do for
  220. 6:37you, and that's the final edit, because
  221. 6:39stitching the clips into a single film
  222. 6:41is still a manual job. So, download the
  223. 6:43three videos, drop them [music] into
  224. 6:44CapCut, editor you use, and line them up
  225. 6:47in story order. Then trim the edges of
  226. 6:49each video, because AI footage always
  227. 6:51has a rough second or two around the
  228. 6:52cuts, and trimming those edges is what
  229. 6:54makes the transitions feel smooth. After
  230. 6:56that, layer your music to match each
  231. 6:58part. Something calm for the
  232. 6:59exploration, something driving for the
  233. 7:01chase, something uneasy for the
  234. 7:02discovery. And one thing to know here,
  235. 7:05ending up shorter than you generated is
  236. 7:06completely normal. A 36-second
  237. 7:08generation usually trims down to around
  238. 7:1030, so always generate more footage than
  239. 7:12your target, because that extra is your
  240. 7:14room to edit. And export the final cut
  241. 7:16as a 720p video. But a 720p file is not
  242. 7:19what you want to publish, and fixing
  243. 7:20that is the one step people skip. C-DANs
  244. 7:22generates at 720p, which looks fine in
  245. 7:24the edit, but soft on a big screen. So,
  246. 7:26the last move is taking the whole film
  247. 7:27up to 4K. And this happens right back
  248. 7:29inside Claude, so I'll upload the
  249. 7:31finished MP4 into the same chat and
  250. 7:33paste this in. The settings go inside
  251. 7:34the sentence again. The source is our
  252. 7:36720 export, the provider is Topaz Video
  253. 7:38AI, the model is Starlight Precise 2.5,
  254. 7:41the target is 4K, and the aspect ratio
  255. 7:43stays on auto, so nothing gets
  256. 7:44stretched. And Claude does the entire
  257. 7:46thing on its own. It uploads the file to
  258. 7:47Higgsfield, checks that the upload went
  259. 7:49through, confirms the job, and runs the
  260. 7:51upscale through Topaz Video AI with the
  261. 7:53Starlight Precise 2.5 model, taking our
  262. 7:55720p film to full 4K with the aspect
  263. 7:57ratio preserved.
  264. 8:28And the difference is obvious. The water
  265. 8:30texture, the particles, the torch beam,
  266. 8:32everything holds up on a big screen now.
  267. 8:34And that order, generate at 720p, edit,
  268. 8:36then upscale once at the end, is the
  269. 8:38cheap way to do this, because editing
  270. 8:39low-quality files is faster on any
  271. 8:41machine, and you only pay for 4K a
  272. 8:43single time on the finished cut. So, the
  273. 8:45whole film took one Claude conversation
  274. 8:46and one editing session. Claude
  275. 8:48generated the anchor image, wrote the
  276. 8:50prompts, made all three videos, and
  277. 8:51finished it in 4K, and the only thing I
  278. 8:53did by hand was the cut. And the same
  279. 8:55setup is how I run my image jobs, video
  280. 8:57jobs, and upscales now. One chat, plain
  281. 8:59English, approve the steps, collect the
  282. 9:01result. And the part that sold me on
  283. 9:03this setup is how little I ever left the
  284. 9:04chat, because from the first test image
  285. 9:06to the 4K export, Claude ran all of it
  286. 9:08through one connection. So, when you're
  287. 9:10ready to build your own, the Higgsfield
  288. 9:12link is in the description below. Thank
  289. 9:13you for watching, and I'll see you in
  290. 9:15the next one.

About this transcript

This page contains the full transcript of How to Create High Quality AI Videos With Claude by Parker Prompts, generated from the public captions YouTube serves with the video. The transcript has 2,011 words across 290 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.