YouTube2Text

How to Make Seamless Transitions with AI — Transcript

by Thomas Creates · 2,463 words · 357 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Have you ever tried editing your AI
  2. 0:01videos only for the transitions between
  3. 0:03your shots to ruin the whole thing?
  4. 0:05Well, I just found the best workflow
  5. 0:06where you can make every cut completely
  6. 0:08seamless and people won't even be able
  7. 0:09to tell that it was done with AI. So, in
  8. 0:11this video, I'll walk you through the
  9. 0:13process step-by-step from the very first
  10. 0:15transition to a finished project. This
  11. 0:17whole workflow runs inside Higgsfield,
  12. 0:19which is an all-in-one tool that gives
  13. 0:20me access to image and video generation
  14. 0:22and it also allows me to add smooth
  15. 0:24transitions between them. If you want to
  16. 0:26follow along, I've left a link for
  17. 0:27Higgsfield in the description below.
  18. 0:29Once you log in, you'll notice that
  19. 0:30everything is organized at the top
  20. 0:32navigation bar. The first transition I
  21. 0:34want to create is a brand logo that
  22. 0:35morphs into the product and to do this,
  23. 0:37I need the images for both. So, click on
  24. 0:39image and that will take you to the
  25. 0:41image generation workspace. I'll select
  26. 0:43GPT image 2 as my model because it's
  27. 0:45currently one of the best models for
  28. 0:46photorealism and adding details. I'll
  29. 0:48paste in this prompt and generate my
  30. 0:50logo. This looks pretty good. It looks
  31. 0:52like a logo an actual brand would use
  32. 0:54for their product, which was exactly the
  33. 0:55point and the text rendering is
  34. 0:57accurate. Then for the product shot,
  35. 0:58I'll tag that logo as a reference so
  36. 1:00that it looks exactly the same and we
  37. 1:02don't lose consistency. With the same
  38. 1:03settings, I'll now paste in this and
  39. 1:05generate. We get this sneaker with the
  40. 1:07logo perfectly integrated on it. Both
  41. 1:09images match the brand, which will help
  42. 1:11make the transition feel a lot more
  43. 1:12smooth. Then back in the main page,
  44. 1:14click on apps at the top. There you'll
  45. 1:16see a collection of all the different
  46. 1:17apps Higgsfield has to offer. There are
  47. 1:19honestly a lot here, but the one we want
  48. 1:21to focus on is called transitions. So,
  49. 1:23click on that. Then upload the logo
  50. 1:24image as the starting frame and the
  51. 1:26sneaker as the end frame. There is a
  52. 1:28huge library of different transitions to
  53. 1:30choose from. For this one, I'll select
  54. 1:32logo transform and generate. The logo
  55. 1:34morphs straight into the sneaker.
  56. 1:36Honestly, this looks like an ad an
  57. 1:37actual company could use to showcase its
  58. 1:39brand new sneakers. There are so many
  59. 1:41things you can do with the transition
  60. 1:42app and this is still just using images
  61. 1:44as the start and end frames. So, the
  62. 1:46next thing I want to test is if it can
  63. 1:47maintain perfect consistency on human
  64. 1:49characters as well. So, back in the
  65. 1:51image generation workspace with GPT
  66. 1:53image 2 selected, I'll paste in this
  67. 1:55prompt to make a young woman wearing a
  68. 1:56cream outfit. She looks exactly like I
  69. 1:58asked for. Now, I want to create a
  70. 2:00transition that focuses on the color
  71. 2:02change. So, for the next image, I'll
  72. 2:03reference the one we just made to keep
  73. 2:05her face looking exactly the same, but
  74. 2:07in my prompt, I'll ask to put her in a
  75. 2:08different outfit and in another
  76. 2:10environment. It's still the same woman,
  77. 2:11but now she's wearing an all black fit
  78. 2:13and she's walking down a street. The two
  79. 2:15images have totally different vibes,
  80. 2:17which is exactly why I want to connect
  81. 2:18them. So, back in the transitions tab,
  82. 2:20I'll upload both of them as a start and
  83. 2:22end frame. And this time, I'll select
  84. 2:24Raven transition from the transition
  85. 2:26library. The whole thing feels very
  86. 2:27smooth. It's still the same woman, but
  87. 2:29now she transforms from one frame to the
  88. 2:31next with the Raven effect that I added.
  89. 2:33And this fits well because of her black
  90. 2:34outfit. It looks like the Ravens
  91. 2:36transform her with a totally new style.
  92. 2:38But, creating transitions from images is
  93. 2:40only the beginning. Images only give you
  94. 2:42so much control over what your final
  95. 2:44result will look like. So, to push this
  96. 2:46tool to its limits, we need to see how
  97. 2:47it can hold up against video
  98. 2:48transitions, which is what this tab was
  99. 2:50built for. And the first [music] problem
  100. 2:52you run into with video is that two
  101. 2:53separate generations almost never give
  102. 2:55you the same person twice, which is a
  103. 2:57real problem for a transition because it
  104. 2:59only looks smooth if it's obviously the
  105. 3:00same person on both sides of the cut.
  106. 3:02So, before I generate any video, I'll go
  107. 3:04back into the image workspace with GPT
  108. 3:07image 2 and build something called a
  109. 3:08character sheet. A character sheet is
  110. 3:10just one image of the same person from
  111. 3:12different angles, which will help the
  112. 3:13video model keep them looking exactly
  113. 3:15the same when we animate them. I'll
  114. 3:17paste in this prompt, which asks for a
  115. 3:19woman from three different angles. And
  116. 3:20here she is from the front, back,
  117. 3:22[music] and up close in one image. Now,
  118. 3:24I'll switch over to the video tab, which
  119. 3:26takes you to the video generation
  120. 3:27workspace, and I'll pick Sea Dance 2.0
  121. 3:29as the model because right now, it's one
  122. 3:31of the best for realistic movement and
  123. 3:33for actually following the prompt. This
  124. 3:35first video is the start of my
  125. 3:36transition, so I'll use the character
  126. 3:38sheet as the reference for maximum
  127. 3:39consistency. Paste in my prompt, set the
  128. 3:41length to 4 seconds, and generate.
  129. 3:44>> This is how beautiful Japan is.
  130. 3:47>> In this one, she's standing in a plain
  131. 3:49studio. The camera slowly pushes in on
  132. 3:51her, and she looks into the lens and
  133. 3:52talks about how beautiful Japan is. The
  134. 3:54part that matters most is how it ends,
  135. 3:56on a calm steady shot of her just
  136. 3:58standing there, because the transition
  137. 3:59needs a still moment like that to flow
  138. 4:01smoothly into whatever comes next. Then
  139. 4:03for the second video, I'll keep that
  140. 4:04same character sheet uploaded, keep the
  141. 4:06same settings, and generate using this
  142. 4:08prompt.
  143. 4:13This time she's in a traditional
  144. 4:14Japanese kimono, and then she walks out
  145. 4:16of frame while the camera pushes
  146. 4:17forward. You can almost imagine how
  147. 4:19these two videos will connect even
  148. 4:20though I haven't done anything yet. And
  149. 4:22the key here is that this video opens on
  150. 4:24the exact same standing position the
  151. 4:25first one ended on. So the two will
  152. 4:27match perfectly. Now, I'll go back to
  153. 4:29the transitions tab and upload the first
  154. 4:31video as the start frame and the second
  155. 4:32[music] as the end frame like we did
  156. 4:34before. I'll run these two videos across
  157. 4:36two different effects from the library
  158. 4:37because I really want to show you the
  159. 4:39variety it has. First, I'll select the
  160. 4:41hand transition and generate. This
  161. 4:42already looks way better than the
  162. 4:44images. The way she extends her hand,
  163. 4:46the camera pushes into her palm, and
  164. 4:47when it zooms out, she is in a
  165. 4:48completely different environment is
  166. 4:50something that would normally require a
  167. 4:52lot of editing to do. Then I'll run them
  168. 4:54through the smoke transition. This one
  169. 4:55works, too. But side-by-side, [music]
  170. 4:57the hand transition felt smoother and
  171. 4:58more natural for these shots. So when
  172. 5:00you have your two videos, run them
  173. 5:02through a couple of effects first and
  174. 5:03keep whichever one actually matches the
  175. 5:05movement. Now, both of these videos were
  176. 5:075 seconds long because the transition
  177. 5:08tab doesn't work for videos longer than
  178. 5:10that. But I found a way to bypass this
  179. 5:12problem. And to show you how I do it,
  180. 5:14I'll use a video of a full property tour
  181. 5:16that starts outside the house and then
  182. 5:17takes you inside. This time instead of
  183. 5:19generating a person, I want to use
  184. 5:20myself as a character in those videos.
  185. 5:23So just like before, I'll go into the
  186. 5:24image workspace with GPT image 2 and
  187. 5:27build a character sheet, except this
  188. 5:28time it's one of myself. I'll upload a
  189. 5:30photo of me as the reference and paste
  190. 5:32in a prompt that turns it into that same
  191. 5:34three angle sheet, this time dressed in
  192. 5:36a navy blazer like a real estate agent
  193. 5:38would wear. And there I am from the
  194. 5:39front, back, and up close, which gives
  195. 5:41Seedance everything it needs to keep me
  196. 5:42looking exactly the same across both
  197. 5:44videos. Now, I'll head into the video
  198. 5:46tab with Seedance
  199. 5:47selected. And for the first video, I'll
  200. 5:49use my character sheet, set it to 5
  201. 5:51seconds, and paste in my prompt for the
  202. 5:52outside shot.
  203. 5:54>> Welcome. Let me give you a tour of this
  204. 5:56place.
  205. 5:58>> In this one, I'm standing in front of a
  206. 6:00modern glass house introducing the
  207. 6:02place. The camera moves around me, and
  208. 6:03it ends right on the big glass wall of
  209. 6:05the house. That ending is completely on
  210. 6:07purpose because the glass wall is the
  211. 6:09thing both videos are going to connect
  212. 6:10from. Then for the inside tour, I'll
  213. 6:12keep my character sheet in, keep the
  214. 6:13same settings, and generate a longer
  215. 6:15one, 15 seconds this time, walking
  216. 6:17through the house. Now, there's one
  217. 6:18really important trick built into this
  218. 6:20prompt. SeaDance [music] 2 is built for
  219. 6:22multi-shot generation, which means that
  220. 6:23you can create a lot of different shots
  221. 6:25using just one prompt. And in every
  222. 6:27single shot, I give myself a line
  223. 6:29delivery.
  224. 6:30>> Step inside. This is the main living
  225. 6:31area, open plan and full of light. Down
  226. 6:34here, you've got three bedrooms, each
  227. 6:36with its own great natural light. And
  228. 6:38this is the kitchen, fully renovated
  229. 6:41with all new finishes. Come take a look.
  230. 6:43I think you're going to love it.
  231. 6:45>> is honestly impressive. It's a full
  232. 6:4615-second walk through the house where
  233. 6:48the lighting stays consistent room to
  234. 6:50room, and it opens on that same glass
  235. 6:51wall from the inside. So, you can
  236. 6:53already see how these two are going to
  237. 6:55meet. There's also a funny moment where
  238. 6:56the generation briefly made a second me
  239. 6:58with a clean switch between them. I
  240. 7:00actually kept it in because it looks
  241. 7:02like a deliberate high-end VFX shot that
  242. 7:04a real agency would pay a lot of money
  243. 7:05to pull off. Now, for the actual fix to
  244. 7:07the length problem, because the second
  245. 7:09one is too long for the transitions tab,
  246. 7:11I can't just drop them in. So, instead,
  247. 7:12I'll export only two frames, the [music]
  248. 7:14very last frame of the outside video and
  249. 7:16the very first frame of the inside
  250. 7:17video. In theory, you could upload the
  251. 7:19outside video whole since it's just 5
  252. 7:21seconds, but to show you the full
  253. 7:22process, I'll export frames from both of
  254. 7:24them. Both of those frames are the same
  255. 7:26glass wall, one from the outside and one
  256. 7:28from the inside. So, I'll bring just
  257. 7:30those two into the transitions tab as my
  258. 7:32start and end frame, pick the seamless
  259. 7:33transition, and generate. We get this
  260. 7:35smooth transition that connects the two
  261. 7:37[music] shots by turning the outside
  262. 7:38view into the inside one. So, now we
  263. 7:40have the exterior video, the transition
  264. 7:42shot, and the interiors video. There's
  265. 7:44one last thing to sort out, and that's
  266. 7:46the voice. Because the two videos were
  267. 7:47generated separately, I come out
  268. 7:49sounding like two slightly different
  269. 7:50people. So I'll go into the audio tab
  270. 7:52inside Higgsfield, upload a few minutes
  271. 7:54of my real voice, and train a voice
  272. 7:55model on it. Then I'll apply that
  273. 7:57trained voice across the whole thing,
  274. 7:58and now it's my actual voice narrating
  275. 8:00the entire tour from start [music] to
  276. 8:02finish. Welcome. Let me give you a tour
  277. 8:04of this place.
  278. 8:13Step inside. This is the main living
  279. 8:15area, open-plan and full of light. Down
  280. 8:18here you've got three bedrooms, each
  281. 8:19with its own great natural light, and
  282. 8:22this is the kitchen, fully renovated
  283. 8:24with all new finishes. Come take a look.
  284. 8:27I think you're going to love it. And now
  285. 8:28the whole property tour plays as one
  286. 8:29continuous shot with my face and voice
  287. 8:31remaining consistent through all of it.
  288. 8:33However, there are a few awkward
  289. 8:34moments, and a lot of the movement still
  290. 8:36feels a little flat. The motion through
  291. 8:38it is just too stiff, and fixing exactly
  292. 8:40that is what finally makes a transition
  293. 8:42look truly seamless. The transition
  294. 8:44straight out of Higgsfield has two
  295. 8:45issues. It plays at one flat even speed
  296. 8:48the whole way through, and it comes with
  297. 8:49no sound at all. And for that, I'll
  298. 8:51bring the video into DaVinci Resolve.
  299. 8:53I'll turn on retime controls, then add a
  300. 8:55speed point at the two spots where the
  301. 8:56transition happens, and pull those
  302. 8:58points together, and add an ease curve
  303. 8:59so it starts slow, speeds up fast
  304. 9:01through the middle of the transition,
  305. 9:02and slows back down on the way out. That
  306. 9:04change in speed is what gives the
  307. 9:06transition momentum. So instead of one
  308. 9:08even mechanical blend, it actually feels
  309. 9:10intentional. While I'm in here, I'll
  310. 9:11also drop an adjustment layer on top
  311. 9:13with deep glow and a bit of exposure,
  312. 9:15which adds a soft bloom to the
  313. 9:16highlights, which makes it look like an
  314. 9:18actual real estate video. Then I'll move
  315. 9:20on to the second fix, the sound. A big
  316. 9:22reason the raw transition feels flat is
  317. 9:24that it's completely silent, so adding
  318. 9:25audio is what actually sells the
  319. 9:27movement and covers the seam. [music]
  320. 9:28I'll add some music that fits the tour,
  321. 9:30and then layer a few sound effects right
  322. 9:32around the transition point. Finally,
  323. 9:34I'll turn on optical flow, which smooths
  324. 9:35the motion out even more, so nothing
  325. 9:37falls flat. Let's look at our final
  326. 9:39result. Welcome. Let me give you a tour
  327. 9:41of this place.
  328. 9:44>> [music]
  329. 9:45>> Step inside. This is the main living
  330. 9:47area, open plan and full of light. Down
  331. 9:49here, you've got three bedrooms, each
  332. 9:51with its own great natural [music]
  333. 9:53light, and this is the kitchen. Fully
  334. 9:55renovated with all new finishes. Come
  335. 9:58take a look. I think you're going to
  336. 9:59love it. These small edits give this
  337. 10:00video a completely different feel. Even
  338. 10:02though the details are small, if you
  339. 10:04compare it to the previous version, you
  340. 10:06can immediately notice how professional
  341. 10:07this one is. And honestly, almost none
  342. 10:09of that was even on me, because
  343. 10:11Higgsfield planned the camera moves,
  344. 10:12generated the entire location, and did
  345. 10:14the hard part of blending the join. So,
  346. 10:16all I added at the end was the speed
  347. 10:18ramp and the sound, which you'd be doing
  348. 10:19in any normal edit anyway. With this
  349. 10:21workflow, you can create seamless
  350. 10:23transitions in a fraction of the time it
  351. 10:24would normally take you, and it all
  352. 10:26happens under one platform. So, if you
  353. 10:28want to start creating your own seamless
  354. 10:29transitions for your videos, use the
  355. 10:31link in [music] the description to sign
  356. 10:33up to Higgsfield. Thanks for watching,
  357. 10:35and I'll see you in the next one.

About this transcript

This page contains the full transcript of How to Make Seamless Transitions with AI by Thomas Creates, generated from the public captions YouTube serves with the video. The transcript has 2,463 words across 357 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.