9일 만에 210만 뷰, 3D 사건다큐 무료로 만드는 법 (클로드 코드 + 블렌더) — Transcript
Full transcript
- 0:00This channel reached 2.1 million views
- 0:01and 20,000 subscribers in just 9 days
- 0:03with its first video.
- 0:05It’s a true crime and cold case
- 0:07documentary channel where small dolls
- 0:08recreate crime scenes.
- 0:10A few days later, they uploaded a second
- 0:1219-minute video.
- 0:13That one also surpassed 90,000 views in
- 0:15just three days. After seeing this, I
- 0:17thought, "Should I try something like
- 0:18that too?" I've always liked mystery
- 0:20cases, so I decided to start a new
- 0:22channel.
- 0:23The channel is called "Cold Case Model
- 0:25Room." I have currently uploaded episode
- 0:271 and episode 2.
- 0:29All 3D scenes were created using a free
- 0:31tool called Blender.
- 0:32There is no AI used in the videos at all
- 0:34. So, there are no additional costs
- 0:37other than using Claude. However, as I
- 0:39made it for the first time, there was
- 0:40one thing I found lacking.
- 0:42The dolls wouldn't move their limbs.
- 0:45The first video of this Korean channel
- 0:46was the same way.
- 0:49But looking at their second video, this
- 0:51time a person is walking around. They
- 0:53had fixed the same thing I noticed. I
- 0:55had the same dilemma. So I did some
- 0:57research, applied some free motion
- 0:59capture data, and now they can sit and
- 1:01answer the phone like this.
- 1:04So, can you all try copying this channel
- 1:05right now?
- 1:07Today, I will explain why this channel
- 1:09blew up, how I made episodes 1 and 2,
- 1:11and even give you a kit so you can make
- 1:13the same thing on your own computer.
- 1:17And the fourth episode I'm currently
- 1:18working on actually looks much better
- 1:20than the Korean channel I referenced.
- 1:22I will reveal the secret to that as well
- 1:24.
- 1:25But this Korean channel isn't the one
- 1:26that invented this format. The original
- 1:28creator is an overseas channel called "
- 1:30Fern." It means "fern" in English. They
- 1:33have over 5.6 million subscribers. One
- 1:35of their signature videos has surpassed
- 1:3720 million views.
- 1:38It is a 23-minute video about the true
- 1:41story of teenage hackers who hacked the
- 1:43CIA.
- 1:44If you look, most of the screen is dark
- 1:46, and light falls like a puddle only
- 1:47where the doll is standing.
- 1:49It feels like looking down at a
- 1:51miniature set from above. The Korean
- 1:53channel brought the "Fern" style over to
- 1:54the Korean language. The dark stage, the
- 1:56faceless dolls, and even the low, calm
- 1:58voice.
- 2:00I asked Claude to analyze five Fern
- 2:02episodes and the Korean channel's videos
- 2:04in their entirety. I had it look at how
- 2:06many times the scene changes, how fast
- 2:08the speech is, and even how the sound is
- 2:09layered.
- 2:11First, the subject matter is true cold
- 2:13cases. "Who was the culprit?" That
- 2:15single question keeps viewers hooked
- 2:17until the end of the video.
- 2:18Even when it's over 10 minutes long,
- 2:20people don't leave. The fun part is the
- 2:21comments. The most-liked comments aren't
- 2:24even about the culprit.
- 2:25"Why did that person go out at dawn
- 2:27without their glasses?"
- 2:28Thousands of people start deducing
- 2:30theories over one small question like
- 2:31this. Even after the video ends, the
- 2:33investigative meeting continues in the
- 2:34comment section.
- 2:36And this comment also got over 1,000
- 2:38likes.
- 2:39This is their first video ever? People
- 2:41were truly surprised. Second, the
- 2:43faceless dolls. You don't have to draw
- 2:46or find photos of real people, so it's a
- 2:48huge relief.
- 2:49You don't have to use victims 'faces
- 2:51carelessly, and it even makes the
- 2:53visuals creepier.
- 2:55The third one was the most surprising.
- 2:57It's the sound. Pun keeps the background
- 2:59music much quieter than the voiceover.
- 3:01Almost to the point where you can't hear
- 3:02it.
- 3:03But at crucial moments, a door slamming
- 3:05or footsteps pop out as loudly as the
- 3:07voice.
- 3:09Since it's quiet most of the time, that
- 3:10one moment really hits you.
- 3:13My first video was the opposite.
- 3:15I layered background music and added
- 3:17tons of sound effects, but they were all
- 3:18the same volume, so nothing could be
- 3:19heard.
- 3:21Fourth, the thumbnail we saw earlier. It
- 3:24doesn't explain with text, it just asks
- 3:25: "What happened to this person?" It
- 3:27throws just one question at the viewer.
- 3:29And fifth, this is something I didn't
- 3:31know until I broke it down. I thought
- 3:33these videos were 3D from start to
- 3:35finish.
- 3:36But when I counted the scenes one by one
- 3:38, 3D made up just over half of the 117
- 3:40scenes.
- 3:42The remaining 40%or so were text cards
- 3:44for maps, people, relationship charts,
- 3:46and timelines.
- 3:48Their second video went a step further
- 3:50and even mixed in actual footage from
- 3:52the scene. The screen is also very dark
- 3:54.
- 3:55They show the events in 3D and use cards
- 3:57to clarify confusing parts.
- 4:00So even if it's a case you're seeing for
- 4:01the first time, you don't get lost.
- 4:04This also means you don't have to create
- 4:06100%3D content.
- 4:08I even measured their speech. About five
- 4:10syllables per second. It's quite slow.
- 4:13And there are 15 pauses that last longer
- 4:15than a second.
- 4:17They take a beat right before the scary
- 4:19parts.
- 4:20To summarize: one question, faceless
- 4:23dolls, quiet sounds with a sudden "thud
- 4:25," text-free thumbnails, and half 3D,
- 4:28half cards.
- 4:30So I decided to use these five as my
- 4:32framework.
- 4:33But I'll change the outer appearance to
- 4:34suit my style.
- 4:35However, the truly difficult part was
- 4:37creating the 3D visuals that followed.
- 4:41So, should you try copying this? Looking
- 4:44at the pros, the growth rate is
- 4:45incredible.
- 4:47I mean, 2.1 million in nine days.
- 4:49Once you build one set, you can keep
- 4:51filming different cases, so it's easy to
- 4:53upload weekly.
- 4:55There are only a handful of channels
- 4:56like this in Korea.
- 4:58There are many true crime channels, but
- 5:00most just use photos and voiceovers.
- 5:02Places that recreate things with 3D
- 5:03dolls are rare.
- 5:05Also, because each episode features a
- 5:07different case, there’s a low risk of
- 5:08YouTube flagging it as repetitive, "
- 5:10cookie-cutter" content.
- 5:12But there is one thing that holds me
- 5:13back.
- 5:14Honestly, it’s quite difficult to make.
- 5:16I mean, it's 3D. Making puppets,
- 5:19building stages, setting up cameras and
- 5:21lighting by hand takes hours for every
- 5:24single scene.
- 5:26This would be a great channel if only it
- 5:28were easier to make, but that’s the
- 5:29hardest part. So, I decided to entrust
- 5:32this entire difficult process to Claude
- 5:34.
- 5:36I’ve created two episodes.
- 5:38I’ll show you a preview here, and for
- 5:40those who want to see the full ending,
- 5:42I’ve left a link in the description.
- 5:45One episode is about a 1986 incident in
- 5:48Los Angeles where the culprit was caught
- 5:50after 23 years.
- 5:52The other is the 1968 300 million yen
- 5:54robbery in Japan, where a cash transport
- 5:56vehicle was stolen entirely.
- 5:58The production process goes like this.
- 6:00Claude writes the script, and then I add
- 6:02the voiceover using PC audio.
- 6:04Claude creates a table for every scene,
- 6:06detailing which stage to use, who is
- 6:08where, and where the camera should shoot
- 6:09.
- 6:10A free 3D program called Blender reads
- 6:12that table to build the stage and take
- 6:14the shots.
- 6:16Finally, a tool called Remotion adds the
- 6:18subtitles and text cards.
- 6:20From start to finish, I only had to give
- 6:22instructions in Korean.
- 6:24However, when I actually tried making it
- 6:25, there were more than just one or two
- 6:27problems.
- 6:28The puppets' fingernails kept flickering
- 6:30on screen, the visuals were flat, and
- 6:32the colors were almost grey.
- 6:34When I placed it side-by-side with Fun,
- 6:36the difference was immediately obvious.
- 6:38So, Claude and I identified the causes
- 6:40one by one and fixed them all.
- 6:44The flickering stopped, I changed the
- 6:46camera to orbit the puppets, which
- 6:47tripled the movement on screen, and
- 6:49adjusted the colors to match Fun.
- 6:52I’ve applied all these fixes and
- 6:54included them in the installation kit.
- 6:57You won’t have to go through the trial
- 6:59and error that took me days to resolve.
- 7:02If you use the kit, it comes with
- 7:04everything fixed from the start.
- 7:05You can get the kit from the Brop Youth
- 7:07group chat that I’m running.
- 7:10I plan to keep updating it as we find
- 7:12more issues, so please subscribe and
- 7:14join the chat.
- 7:16I also changed the audio based on what I
- 7:17learned from Fun earlier. I lowered the
- 7:19background music significantly below the
- 7:21voiceover and only boosted sound effects
- 7:23for moments like doors closing or
- 7:24cameras flashing.
- 7:26I made sure to include at least six such
- 7:27high-impact moments per episode.
- 7:29I also added a few of my own unique
- 7:31touches.
- 7:32An evidence board with red strings
- 7:34connecting suspects, and a map spread
- 7:36out on a table.
- 7:37And a transition where a period
- 7:38newspaper photo smoothly transforms into
- 7:40a 3D scene with the same composition.
- 7:42For the 300 million yen case, since the
- 7:44statute of limitations date is important
- 7:46, I added a clock counting down to that
- 7:48day.
- 7:49I put the script through a custom
- 7:51evaluation system I built, and only
- 7:53recorded it after it scored 95 out of
- 7:55100.
- 7:56Because if the script is weak, people
- 7:58will leave no matter how good the
- 7:59visuals are.
- 8:01After these refinements, it’s become
- 8:03quite similar to Fun, the original
- 8:04overseas channel. But there is one thing
- 8:06I couldn't fix until the end.
- 8:09Punn's puppet walks, turns its head, and
- 8:12opens the door. My puppet is stuck in a
- 8:14pose and doesn't move at all.
- 8:17To animate this by hand, you have to
- 8:19adjust elbows, knees, and heads frame by
- 8:21frame, which is the most time-consuming
- 8:23task in 3D.
- 8:25That's when I found Motion Bricks,
- 8:27released by NVIDIA this year.
- 8:29They announced it at the world's largest
- 8:31computer graphics conference and made it
- 8:33free for anyone to use. The principle is
- 8:35simple.
- 8:36If you provide the current pose and the
- 8:38final pose, the AI fills in the natural
- 8:41human movements in between. It’s like
- 8:43telling a person standing up to go sit
- 8:45in that chair, just giving it a
- 8:46destination.
- 8:48It creates short segments of around one
- 8:49second at a time.
- 8:51It uses the end pose of that segment as
- 8:53the start of the next and stitches them
- 8:54together.
- 8:55So, long movements like walking from a
- 8:58door to a chair come out seamlessly.
- 9:00The computer does in seconds what would
- 9:02take a person half a day to do by hand.
- 9:05And there is one more source of motion
- 9:07data.
- 9:09It’s a real human motion library called
- 9:11Boneseed, containing 70,000 real
- 9:13movements of people waiting, calling,
- 9:15and looking around—140,000 if you count
- 9:17the mirrored versions.
- 9:20When I transferred this to my puppet,
- 9:21this is what happened.
- 9:22It stands, waits, looks around, and
- 9:24crosses its arms.
- 9:25That same mannequin from before.
- 9:27But this didn't happen all at once
- 9:29either.
- 9:31Feet floated off the floor, arms
- 9:33disappeared into the chest when crossed
- 9:35, the phone hid behind the head, and it
- 9:37froze like ice upon arrival.
- 9:40I fixed each and every one of these
- 9:42issues and put all those methods into
- 9:44the kit.
- 9:46You just need to specify it as a scene
- 9:48of sitting and answering the phone.
- 9:52And I’d like to brag a little here. The
- 9:54Samaneun case I’m working on now looks
- 9:56better than the original Korean channel
- 9:58or the original foreign channel Punn. It
- 10:01turned out visually excellent.
- 10:03Since Samaneun is a case with many real
- 10:06victims, I created the scenes using only
- 10:08sets and lights. There are five secrets
- 10:10to it. First, I built the entire case as
- 10:13a village model on a black pedestal.
- 10:16When the pin light clicks on, the
- 10:17nameplate is revealed, and the
- 10:19criminal's path is shown only by a
- 10:21single red thread.
- 10:22Instead of people, the thread and window
- 10:25lights tell the story.
- 10:27Second, I built a separate, complete
- 10:29stage for every location. From the
- 10:31police sub-station and post office
- 10:32exchange to the investigation board and
- 10:34the bridge.
- 10:35I even set different lighting and colors
- 10:37for each location.
- 10:38I also created each prop one by one.
- 10:41The switchboard, the magnet-style phone
- 10:43, and even the police jeep.
- 10:45The shapes were generated with a free AI
- 10:47that turns a single image into 3D, and
- 10:48Claude refined them by applying
- 10:50materials in Blender.
- 10:52Even to my eyes, the prop detail is
- 10:54definitely better. Third, it's the
- 10:56realistic human movement you saw earlier
- 10:57.
- 10:58By adding a police hat and a
- 11:00mustard-colored uniform, it's
- 11:01immediately clear who the officer is.
- 11:04Fourth, there are no cuts; it moves
- 11:06continuously. The camera keeps rotating
- 11:08.
- 11:09The lights dim according to the story,
- 11:11and a clock ticks away at the bottom of
- 11:12the screen.
- 11:14If you measure the on-screen movement
- 11:16against that original Korean channel's
- 11:18first video, it moves about 1.5 times
- 11:19more.
- 11:20Fifth, I paired actual footage with 3D
- 11:23elements. I show a brief clip of the
- 11:25news from that time and transition
- 11:27directly into a model with the same
- 11:28composition.
- 11:30The screen you just saw was created
- 11:32without using a single paid AI video
- 11:34tool. I will update these methods into
- 11:36the kit as soon as they are organized.
- 11:38Once the video is up, please compare the
- 11:40original Korean channel with mine.
- 11:43So, what do you need to install on your
- 11:45computer to make videos like this?
- 11:47You only need three things.
- 11:50The first is Blender. It’s a 3D program
- 11:52used even by film studios, and it is
- 11:54completely free.
- 11:56Just go to the official Blender website
- 11:58, click the download button, run the
- 11:59file, and keep hitting next.
- 12:02I used version 4.2. Next, you install an
- 12:05add-on that creates human figures.
- 12:08It’s called MPFB; you just type the name
- 12:10into the extension menu inside Blender
- 12:12and click install.
- 12:14Using this to build a human figure and
- 12:16applying a clay-like texture gives you
- 12:17the faceless doll you saw earlier.
- 12:20But you don't need to install these one
- 12:23by one yourself.
- 12:26If you put the kit I mentioned into
- 12:27Claude Code or CodeX and type "install,
- 12:29" it checks if Blender is there and
- 12:31installs what is needed automatically.
- 12:34When you run it for the first time, it
- 12:36asks a few questions. What your channel
- 12:38name will be, which PC audio voice to
- 12:40use, and what the screen colors should
- 12:41be.
- 12:42Once you answer these, your videos will
- 12:44be generated with your channel's
- 12:46settings. I've set the default colors to
- 12:49the teal and amber I picked earlier. I
- 12:51recommend changing them to your own
- 12:53channel's unique colors. After all,
- 12:54color is what makes people recognize
- 12:56your channel just by looking at a
- 12:58thumbnail.
- 12:59It also checks your graphics card at
- 13:01this time, and if you don't have Nvidia
- 13:03, it guides you to skip the next steps.
- 13:06For the second part, Motion Bricks, you
- 13:08definitely need an Nvidia graphics card
- 13:10.
- 13:10It runs on Nvidia-exclusive acceleration
- 13:12technology.
- 13:14Nvidia stated it was tested on an RTX
- 13:164090, but my computer, which is a
- 13:19lower-spec RTX 4070 12GB model, ran it
- 13:21just fine.
- 13:23Those without that graphics card don't
- 13:24need to be disappointed.
- 13:26The waiting and looking-around motions
- 13:27we saw earlier were made using only the
- 13:29motion data I'll show you next, without
- 13:31Motion Bricks.
- 13:32Those users can simply move directly to
- 13:34step three.
- 13:37So, what’s the difference between using
- 13:38just the motion data and using it with
- 13:40Motion Bricks? Let’s compare the same
- 13:42scene.
- 13:44This is a scene where a detective enters
- 13:45through the door and sits in a desk
- 13:46chair.
- 13:48The motion data is a recording of an
- 13:50actual person acting it out.
- 13:52So, the number of steps and the stopping
- 13:54point are exactly as recorded.
- 13:56It doesn't know where the chair is
- 13:58located on our stage.
- 14:00If you just stitch the walking motion
- 14:01and the sitting motion together, the
- 14:03character will suddenly pop to the new
- 14:04position.
- 14:06So, when using only motion data, you
- 14:08have to cut from the walking scene to
- 14:10the sitting scene.
- 14:13If you make a cut like this, it still
- 14:15looks natural enough.
- 14:18Most stationary gestures can be done
- 14:20with just this.
- 14:21If you use Motion Bricks, you just
- 14:23provide the destination, like in front
- 14:25of that chair.
- 14:28Then, it automatically adjusts the steps
- 14:30to stop right in front of the chair and
- 14:31seamlessly continues into the sitting
- 14:33motion. The character moves within a
- 14:35single scene without needing a cut.
- 14:39To summarize, stationary gestures work
- 14:41with just motion data, but you need
- 14:43Motion Bricks to make them walk around
- 14:44the stage.
- 14:46You need two tools for installation. One
- 14:48is a tool for downloading large files,
- 14:50and the other is for setting up Python
- 14:51packages at once. The kit checks for
- 14:53these two, and installs them if they are
- 14:55missing.
- 14:56Then, when you download Motion Bricks
- 14:58and run it for the first time, it takes
- 14:59a few minutes to prepare for your
- 15:01graphics card.
- 15:02Do not turn it off thinking it has
- 15:04frozen. From the second time on, it will
- 15:06turn on quickly. Once it's running, the
- 15:07person will walk to the destination and
- 15:09stop like this.
- 15:10The kit also handles transferring this
- 15:11to the Blender character. There is one
- 15:13thing you must keep in mind.
- 15:15You can use the Motion Bricks code and
- 15:16AI model commercially, but the example
- 15:18motion files included inside are only
- 15:20allowed for testing.
- 15:22You should not use them as-is in your
- 15:23videos. So, for public videos, please
- 15:26use the motion data I will show you next
- 15:28; the third one is the motion data,
- 15:30Boneseed.
- 15:31It is available on an AI data site
- 15:33called Hugging Face. You cannot download
- 15:36it immediately; you must create an
- 15:37account and apply for access. When you
- 15:39apply, terms and conditions will appear
- 15:40, so make sure to read them yourself.
- 15:42If you earn revenue from YouTube, it is
- 15:44safer to treat yourself as a business
- 15:45and check the revenue criteria.
- 15:47Small businesses with annual revenue
- 15:49under $ 1 million can use it, but if you
- 15:51exceed that, you must pay a separate fee
- 15:53.
- 15:53And if you download this entire dataset
- 15:55, it is about 45GB.
- 15:57You don't need to download all of it. I
- 15:59have created and included a tool in the
- 16:01kit to download only the necessary
- 16:02motions.
- 16:03If you say something like "sitting and
- 16:04talking on the phone," it will find and
- 16:06download only that motion.
- 16:08Also, when you upload a video made with
- 16:10this data, you must write "Motion data
- 16:12by Bones Studio" in the description and
- 16:13include the link to the Bones Studio
- 16:15website.
- 16:17That is all for the installation and
- 16:19usage.
- 16:20Shall we see if a detective actually
- 16:22walks in if we give a one-line command?
- 16:25I typed just this one line into Claude
- 16:27Code.
- 16:29Make a scene where a detective walks
- 16:30into an office, sits in a chair, and
- 16:32answers the phone.
- 16:34Since it's my first time using it, it
- 16:35asks for my channel name first.
- 16:37Then, Claude builds the office set,
- 16:39places the detective figure at the door
- 16:41, and selects and stitches together the
- 16:44walking, sitting, and phone-answering
- 16:46motions.
- 16:47This is the result.
- 16:49He walks through the door, stops at the
- 16:51desk, sits in the chair, and answers the
- 16:53phone.
- 16:54Here, I'll add just one more instruction
- 16:56like a director.
- 16:57Have the camera slowly orbit the figure
- 16:59and keep the light focused only on the
- 17:01desk. But this time, it got too dark,
- 17:04and the detective wasn't very visible.
- 17:07Well, you can just fix that with words
- 17:09too. It's too dark. Add another light on
- 17:11the detective's face.
- 17:13Hold the phone with the left hand.
- 17:15If you say this, it just reshoots that
- 17:17scene.
- 17:18And just like that, the face is visible
- 17:20, and he answers the phone with his left
- 17:22hand.
- 17:23I didn't touch a single line of code
- 17:25myself.
- 17:28The reason this works is because Claude
- 17:30manages scenes in a table format.
- 17:32Each scene has fields for stage,
- 17:34character, motion, camera, and light, so
- 17:37when you edit with words, only those
- 17:39cells change while the rest remain the
- 17:41same.
- 17:44At first, my figure just stood there in
- 17:45a pose, but now he walks in, sits down,
- 17:47and even answers the phone.
- 17:50Because the character moves, it finally
- 17:52starts to look like a story is unfolding
- 17:53.
- 17:54There are drawbacks, too.
- 17:56First, Motion Bricks only runs on Nvidia
- 17:58graphics cards.
- 18:00If you don't have one, you have to stick
- 18:02to motion data, and walking sequences
- 18:03must be handled with cuts.
- 18:06Second, small movements like those of
- 18:07fingers or mouths are still awkward.
- 18:10I tried adding a smoking motion, but
- 18:12since the figure lacks fingers and a
- 18:13mouth, it turned out clunky like this.
- 18:16Third, 3D requires running the computer
- 18:18for a long time. You have to keep the
- 18:20computer on for quite a while to finish
- 18:22one episode.
- 18:23And this is more of a line to respect
- 18:25than a drawback. Since you're dealing
- 18:27with true stories, you shouldn't portray
- 18:29victims 'real names, faces, or death
- 18:31scenes sensationally.
- 18:33I don't make death scenes at all and I
- 18:35don't use real faces on screen.
- 18:37That way, viewers can watch comfortably
- 18:39, and the channel can last longer.
- 18:42There was one thing that held me back at
- 18:43first. It was the difficulty of making
- 18:44it.
- 18:45By leaving everything from installation
- 18:47to filming to Claude using the kit, that
- 18:48one issue was largely resolved.
- 18:51So, here is my answer.
- 18:53If you enjoy true crime stories, it's
- 18:55worth starting now.
- 18:56You can start even if you don't have an
- 18:58Nvidia graphics card. If you do have one
- 19:00, you can even make the figures walk
- 19:01around.
- 19:02Channels like this have just started to
- 19:04emerge in Korea.
- 19:07And for those curious about the endings
- 19:08of the two previews you saw today, check
- 19:10out the Cold Case Model Room via the
- 19:11link in the description.
- 19:13I've uploaded the versions where the
- 19:15figures come to life.
- 19:17If you just copy it as is, you'll only
- 19:19end up as a knock-off of someone else's
- 19:20channel. People subscribe not for the
- 19:22technique, but for that channel's unique
- 19:24fun and personality. Choose your stage
- 19:26colors, character designs, and the
- 19:28stories you want to cover.
- 19:29Try changing the way you develop those
- 19:31stories as well.
- 19:32The framework can be similar, but your
- 19:34channel needs its own unique identity.
- 19:36To be honest, you can get perfectly
- 19:39decent results using only free tools.
- 19:42However, some scenes that really need to
- 19:43grab the viewer's attention felt a bit
- 19:45lacking. So, I’m testing ways to elevate
- 19:48those specific scenes, and I'll show you
- 19:50in the next video.
- 19:52You know, in the past, while watching
- 19:54YouTube, you’d think, "Oh, this is fun.
- 19:55"
- 19:56Even if you wanted to try it yourself,
- 19:57you often couldn't because you lacked
- 19:59the technical skills.
- 20:01But now, if you do a little research
- 20:02with AI, we’ve entered an era where
- 20:04almost anything is possible.
- 20:07I research these things for you first to
- 20:09help you do it with minimal time and
- 20:11cost.
- 20:14You can get the kit I showed today and
- 20:16more information by joining my group
- 20:17chat.
- 20:19Please enter through the link in the
- 20:20pinned comment.
- 20:22Oh, and
- 20:24The video you are watching right now was
- 20:26also edited entirely with AI from start
- 20:28to finish.
- 20:29I didn't even open an editing program to
- 20:31create this video.
- 20:33If you want to learn how to use AI like
- 20:35this, please subscribe to the Brock
- 20:37Youth channel.
About this transcript
This page contains the full transcript of 9일 만에 210만 뷰, 3D 사건다큐 무료로 만드는 법 (클로드 코드 + 블렌더) by 불혹청년, generated from the public captions YouTube serves with the video. The transcript has 3,862 words across 654 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.