YT YouTube2Text

YouTube2Text → Transcripts → I Turned Claude Into a $880/Day Motion Graphics Skill (No After Effects)

I Turned Claude Into a $880/Day Motion Graphics Skill (No After Effects) — Transcript

by Sanji Nai-Chien · 3,343 words · 489 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00People are making over $500 a day
0:03selling motion design like this. And
0:05their clients have no idea that it was
0:07made with Claude without opening After
0:09Effects and without a single key frame.
0:12Every animation that you're watching is
0:14actually code that Claude wrote. And it
0:16takes only one day to make, so you can
0:18take on way more clients than any motion
0:20designer out there ever could. [music]
0:22At the end of this video, I'll show you
0:24exactly how much profit I kept from this
0:26single project. But ultimately, is it
0:29actually as good as a human pro? To find
0:32out, I hired a real motion designer and
0:34gave them the exact same brief, files,
0:37and budget to see just how big the gap
0:40between us really is. Now, he asked for
0:425 days and Claude got one. And only one
0:45of us actually made the deadline. So
0:47[snorts] in this video, I'm going to
0:49show you everything. How I found a
0:50client, the exact message that got me
0:53hired, how Claude built the video, and
0:55ultimately how I got paid. So without
0:57further ado, let's get into it.
1:03So step one is to find a client. Now,
1:05there's a lot of motion design styles
1:07out there and not every client is
1:09willing to pay well. So it's important
1:11that if you want to earn more, you need
1:13to do the style that pays the most. Now,
1:16today that style is SaaS or tech launch
1:19motion videos. For just 30 seconds of
1:21this kind of motion design, clients pay
1:23anywhere from 1,500 to $10,000.
1:27And finding clients here is a lot easier
1:29than my niche scene. You can do it with
1:31no experience and an empty portfolio. So
1:34here's exactly how to get it done. I go
1:37to Upwork and type SaaS launch video
1:40into the search. And as you can see,
1:42there's a ton of jobs here at all kinds
1:44of budgets. And here's the first tip.
1:46Make sure to use different searches to
1:48get more jobs to choose from. You always
1:50want to try things like product launch
1:52video, UI animation, app promo video,
1:56and SaaS explainer. But, the most
1:58important thing is picking the right
2:00jobs out of all of these. Now, most
2:02people would scroll and literally apply
2:05to everything. That's exactly how you
2:07burn your connects and ultimately get
2:09zero replies, which means zero money.
2:12So, here's three things that I always
2:14check before I apply. First, you want
2:17the client to say payment verified, 10K
2:20plus spent, and a higher rate that's
2:22above 50%. That means that ultimately
2:25they actually hire and they actually
2:27pay. So, you're not wasting connects on
2:29somebody who's going to ghost you.
2:30Second, freshness. Now, you want to sort
2:33by newest and always keep refreshing. I
2:35go for posts that are less than 30
2:37minutes old, which means that the client
2:39just posted the job, so they're probably
2:41still online watching proposals come in.
2:43>> [music]
2:44>> You want to apply in that window and
2:45you'll end up being one of the first
2:47people that they see. Third, the
2:49competition. You want to scroll down to
2:51activity on this job. There you can
2:54actually see how many people already
2:55applied and how many of the client is
2:57already interviewing. Fewer than 15
3:00proposals and nobody interviewed yet
3:02means that they haven't picked anybody
3:04and your message is actually going to
3:06get read. Fourth, the money. You want to
3:09open their website and check whether
3:11they actually just raised a round. A
3:13freshly funded startup about to launch
3:15has both the budget and the deadline, so
3:18you know they're going to be paying well
3:19and fast. Now, if all four of these
3:21things check out, [music] then apply. If
3:23not, skip it and move on. Now, another
3:27insider tip, and yes, it does break my
3:29first rule, you're going to see jobs for
3:31$10 or $5 an hour from clients with zero
3:35hires. Most freelancers are going to
3:37skip them instantly, but a lot of the
3:39time that's just a brand new client who
3:41doesn't know what to put in the budget
3:43field yet. So, make sure to check out
3:45their website. If the company is real,
3:47then maybe it's worth applying anyways
3:49and quoting your real price. There's
3:51almost no competition and it's a client
3:54who's actively looking. And here's the
3:56most hardcore tip of all. Always add
3:59cover letter and use the exact keywords
4:01from their post. Now, if they wrote
4:03product launch video and Figma, then
4:06those words need to be right in your
4:08first lines. That way the client
4:10instantly sees you actually read the
4:12post. My structure here is always the
4:14same. Their task, my solution, and a
4:17clear deadline. So, that way the client
4:19knows exactly what they're getting and
4:21when they're getting [music] it. I'm
4:22going to show you the exact template in
4:24just a second. Now, quick pause. If this
4:27is useful, please make sure to
4:29subscribe. This channel is all about
4:31making real money with AI tools. The
4:33workflows, the prompts, and the
4:35experiments just like this one. And for
4:38everything first, follow me on
4:40Instagram. That's where the newest stuff
4:42drops before YouTube. Okay, let's get
4:45back to it. Now, you found the job, so
4:47let's get hired.
4:52Okay, I've picked five jobs that look
4:55promising. Now, I'm going to send each
4:57of them a message just like this. Feel
4:59free to pause the video and read it.
5:01This message is written so that the
5:03client picks you even if you don't have
5:05a portfolio. Now, why like this? Well,
5:07first, it's because the opening line is
5:09actually about them. On Upwork, the
5:12client only sees the first two lines in
5:14the preview. And almost every single
5:17proposal out there starts with something
5:18like, "Hi, I'm a motion designer with X
5:21years of experience." And honestly,
5:24that's boring. I'm pretty sure that none
5:25of you would click on a message like
5:27that. Now, the next reason is that the
5:29concept replaces the portfolio. A
5:32portfolio shows what you did for
5:34somebody else. A ready-made [music]
5:36concept shows what you'll do for that.
5:38The client can already picture their
5:40video and you're the only one who showed
5:42it to them. And most importantly, the
5:45small first milestone. Because of it,
5:48the client takes almost no risk. With a
5:50message like this, clients will start
5:52choosing you over your competitors.
5:55Because literally a couple hours later,
5:57three out of the five clients replied,
6:00already ready to start working with me.
6:02But this project looks the most
6:04interesting to me. It's a startup
6:05building an AI-powered video recording
6:08tool,
6:08>> [music]
6:08>> and they need a launch video before they
6:11go live. Now, they already have a rough
6:13vision that matches mine, and their UI
6:16is already done in Figma. So, I quoted
6:18one day and a thousand dollars for a
6:2030-second motion video. Now, that's
6:22actually not that much, since you
6:24remembered how much videos like this go
6:26for on the market. We messaged back and
6:28forth a bit more, and you can actually
6:30pause the video and read it, and that's
6:32when they ultimately agreed. Now, we're
6:34ready to start generating.
6:40But before we generate anything, [music]
6:42remember the challenge from the start of
6:44the video. So, we need to find a motion
6:46designer. Because I want to know whether
6:49AI can actually replace a pro. So, a
6:52real motion designer is going to get the
6:54exact same task, same brief, same Figma
6:57files. Now, I'm going to go to Fiverr
6:59and search for a SaaS motion designer.
7:02And I'll skip everyone who doesn't have
7:04UI work in their portfolio. Character
7:07animation and UI motion are actually
7:09completely different skills. I found
7:12this guy Alex. He has 47 reviews and
7:15clean UI work, and he quoted a thousand
7:18dollars for 30 seconds in five days. So,
7:21I sent him the brief. And a few minutes
7:23later, he replies. [music]
7:25Okay, so the deal is done. The pro gets
7:28five days and a thousand dollars. Claude
7:30gets one day and the exact same brief.
7:33Now, let's see what Claude can actually
7:36do with this.
7:40Now, if you think you can just drop a
7:41prompt into Claude and get a masterpiece
7:44out of it right away, then you're dead
7:46wrong. It never works on the first try,
7:48but once you've set up the process, you
7:50can actually turn it into a custom
7:52skill. Which is to say, it'll become a
7:54permanent instruction set that Claude
7:56will follow every single time for this
7:58kind of work. After that, then you can
8:01generate everything you need from a
8:02single prompt. And by the way, I've
8:04attached my entire motion design skill
8:07in the blog post link below so that you
8:09can download it and use it on all of
8:11your own projects. But what's actually
8:13inside of it? Basically, it's everything
8:16that a senior motion designer carries
8:18around in their head.
8:20First, the story. Every launch video
8:23that I make follows the same exact path.
8:26Grab attention, show the problem, show
8:28how the product fixes it, and finish
8:31with the payoff. Now, if a scene isn't
8:33serving one of those four needs, then
8:35it's out. Second, timing. Anything
8:38important stays on screen for about a
8:41second and a half. It's long enough to
8:43read, but short enough that nobody's
8:45getting bored. Third, the motion rules.
8:49And this is actually where most of the
8:50magic happens. Nothing ever moves
8:53linearly. Things need to come in over
8:55about half a second with a slight
8:57overshoot, and they leave faster than
8:59they came in. Every click gets a tiny
9:01press, and the cursor never moves in a
9:04straight line. It moves like an arc as a
9:07real hand would. But the golden rule of
9:10smooth animation is this. Nothing should
9:14ever fully stop. In the skill, I tell
9:17Claude to work the way that a motion
9:18designer works in After Effects. Group
9:21related elements under a parent,
9:23basically a null object, and move the
9:25whole group instead of every piece
9:27separately. Then, you want to offset the
9:29keyframes, so each element starts a
9:31couple of frames after the one before
9:34it. And start the next move before the
9:36last one has ever fully settled. When
9:39the keys overlap like that, then every
9:42movement flows into the next one.
9:44Nothing freezes, nothing snaps, and it
9:46all feels like one continuous, coherent
9:49motion. That's the difference between
9:51looks like a website and looks like a
9:54real launch video. And Claude does all
9:56of this in code, so you never have to
9:59set a single keyframe yourself.
10:01>> [music]
10:01>> And fourth, the tech specs. 1920 pixels
10:04wide, 1080 pixels high, 60 frames per
10:08second. Plus safe zones, so when you
10:10adapt the video to vertical, everything
10:12important stays inside the frame and
10:14nothing gets cut off. Now, right now,
10:17I'm just dropping it into the chat, but
10:19before we hit generate, there's two more
10:21strict rules that you always need to
10:23keep in mind. Rule number one,
10:26storyboard first, then animation.
10:29Without a good storyboard, your output
10:31is going to be absolute garbage. Now, in
10:33a bit, I'll show you the same exact
10:35video generated with and without a
10:37storyboard, so you can see the
10:39difference in quality yourself. Rule
10:41number two, if it's motion, you have to
10:44teach it. You have to feed your UI
10:46components, define your easing curves,
10:49and give it a solid baseline. And this
10:52next part is important. Before this
10:54project, I spent more than 10 hours
10:56testing different setups, and this is
10:59actually the one that worked for me, so
11:01it'll work for you, too. So, here's
11:02everything that I gave Claude. The brief
11:04from the client, the storyboard scene by
11:06scene, breaking out what's on screen,
11:09how long it stays, and where the camera
11:10goes. The brand colors and fonts, the
11:13background images, and the UI straight
11:15from Figma, exported as SVG, not
11:18screenshots. Now, the part that matters,
11:21when the camera zooms into a button, an
11:23SVG stays razor sharp, and a screenshot
11:26turns into a blurry mess. Then, one
11:29prompt. Using the motion skill, build
11:31this storyboard as a 30-second launch
11:33video scene by scene. Claude starts
11:37writing. Every scene becomes its own
11:39block of code. The tasks pile up, the
11:41copy-paste grind, the agents kicking in,
11:44the payoff, and the logo. Then, the
11:47whole thing renders frame by frame into
11:49a video. So, I ended up dropping it all
11:52into Claude Opus 5.5, and right away, it
11:55broke. Just look at this. The text is
11:58overlapping, the camera pan is way too
12:00fast. Also, the whole layout just
12:03explodes off the canvas, the font at
12:05some points fell back to Times New
12:07Roman, the task counter jumped to 200 in
12:10a single frame, and all of the agents'
12:12cards pop in at once attached on top of
12:15each other. It's total chaos. Now, this
12:18is where most people quit and say, "AI
12:20motion doesn't work." But, almost every
12:23single one of these bugs has a simple
12:24cause. And once you know that cause,
12:27then the fix is just a single second.
12:29The layout exploded because the Figma
12:31screens were designed 1440 pixels wide,
12:34and the video we asked for is 1920. So,
12:36I need to tell Claude, "Lock the canvas
12:38to 1920 by 1080, and position everything
12:42relative to the center, not the edges."
12:45The font, on the other hand, broke
12:47because it wasn't loaded before the
12:49first frame. I just need to say, "Load
12:51the font before rendering anything." The
12:53counter, "Make the count up last about a
12:56second. Long enough to read, but quick
12:58enough that it doesn't drag."
13:00>> [music]
13:00>> The agent cards, bring them in one by
13:03one a few frames apart, and let each
13:05card land before its check mark appears.
13:09The camera, slow down the pan and give
13:11it an easy ease, so it starts and stops
13:14softly. I spent the next hour going back
13:17and forth just like this, debugging it
13:20line by line. The first few versions
13:22were all about fixing what's broken.
13:24After that, it became about taste. The
13:27task counter needed a soft red glow as
13:30it climbs. The cursor needed to hesitate
13:33between tabs for a second as if you're
13:35losing track of what you were copying.
13:37Each of those is one sentence to Claude
13:40and just a few seconds of waiting. And
13:42every one of those fixes is now actually
13:44baked into the skill in the blog post
13:47below. So, when you use it, you skip all
13:50of this pain of the first hour.
13:52And then it clicked. Just look at that.
13:55It's a buttery smooth 60 FPS animation
13:58rendered straight from code. No timeline
14:01stretching, no missing plugins, and I
14:03didn't set a single keyframe by hand.
14:06Now, do you remember our first rule?
14:09Here's what happens without a
14:11storyboard. Same assets, same skill,
14:14same prompt. [music]
14:15Simply no storyboard. Claude will start
14:18guessing.
14:19>> [music]
14:19>> Scenes come in in a totally random
14:21order. Everything moves at the same time
14:24and there's just no coherent story. It
14:26feels like just a bunch of nice-looking
14:28UI flying around the screen. With a
14:31storyboard, it's problem, pain, fix,
14:35payoff. Now, the same tool but a
14:38completely different output. And here's
14:40why this whole thing takes just a single
14:43day. Say the client comes back and asks
14:45to change the task counter from 200 to
14:48999.
14:49In After Effects, that means opening the
14:52project, finding the layer, fixing the
14:54keyframes, and waiting for a re-render.
14:57Here, that's just one line of code. I
15:00patch it up in seconds and the new
15:02version is ready in just a couple of
15:04minutes. So, the total time is under a
15:07day. Total cost is a $20 Claude
15:10subscription. Meanwhile, Alex was still
15:13working on his style frames.
15:18>> [music]
15:19>> Before we watch the full generation, I
15:22checked in with the motion designer and
15:23they said that they'd make the deadline.
15:26So, I'll be back in a couple of days to
15:27show you both results. Now, in the
15:30meantime, I sent what we've made to the
15:32client and here's what they replied.
15:35The same day I sent the final files,
15:37they approved them and the payment got
15:39released.
15:43A full 7 days have passed and we're
15:45back. The main reason for the delay is
15:47that the motion designer missed their
15:49deadline. On the day when they were
15:51supposed to send the final video, they
15:53sent me this message. And the next day,
15:56this one. But, today they finally
15:59delivered all of the videos, so we're
16:01ready to watch both versions.
16:07All right, [music] let's look at Alex's
16:09video first.
16:29Honestly, it's solid. It hits the brief,
16:32the UI looks clean, and the animation is
16:35quite smooth. If I were the client, then
16:37I'd be totally fine with this. But,
16:39there's two catches. We waited seven
16:42full days for it because they kept
16:43pushing the deadline and it cost a
16:45thousand dollars. Now, here's our
16:47version built entirely in Cloudopus 5.5
16:50using code.
17:13Look at the camera moves, the photo real
17:15background behind the UI, and how every
17:18click has weight. So, which one do you
17:20think turned out better? Ours or Alex's?
17:24Let me know in the comments below.
17:30Okay, let's talk money because that's
17:32probably why you're still watching. Now,
17:34the client paid me $1,000. Claude cost
17:37me 20 bucks for the subscription and I
17:40spent less than a day on it. And most of
17:42that day actually went into the
17:44storyboard and not the actual
17:46generation. After Upwork's fees and the
17:49subscription, I kept $880
17:51for just a single day of work. And
17:53here's a little hack that I didn't use
17:55this time, but you should be aware. If
17:57the client is solid and you explain it
17:59the right way, then you can ask them to
18:01actually cover the Upwork fee. Just be
18:03up front with them. My price is $1,000
18:06and Upwork's fee goes on top. A client
18:09who's happy with your work will usually
18:11say yes without thinking twice. That's
18:14another hundred dollars straight into
18:16your pocket. Now, flip it. Imagine that
18:19you're the client. Your launch is in 2
18:21weeks and you've got two options. Option
18:24one is Alex. Nothing against this guy.
18:27His video is obviously good, but if
18:29you're looking at a full week and you're
18:31waiting on your prize, going back and
18:33forth on revisions, and hoping that it
18:35lands before your launch date. Option
18:38two is somebody like [music] me. It's
18:40the same level of quality but delivered
18:43the very next day. And the revisions
18:45take 1 hour, not days. Who are you
18:48hiring?
18:49Exactly. And here's the part that most
18:51people miss. A SaaS company doesn't need
18:54just one video. They need a launch
18:56video, then a feature update, then an
18:58ad, and then the following launch. So,
19:00if I'm the guy who can deliver fast,
19:02whose content looks good, and I never
19:04miss a deadline, then I'm not just
19:05getting one $1,000 job, I'm becoming
19:09their go-to motion guy. [music] And one
19:11client like that can quickly become an
19:14anchor source of income. And remember,
19:17don't treat this as a magic button. The
19:19storyboard and the skill working
19:21hand-in-hand are what make this work,
19:23and they're both in the blog post linked
19:25below. So, here's the bottom line.
19:28Clients aren't buying this because they
19:30care about code or they care about
19:32keyframes. They buy it because they get
19:35a launch-ready video by the very next
19:37day. Everything that I used for this,
19:40the motion skill, the storyboard layout,
19:42and the exact pitch messaging that
19:44landed this job is in the blog post
19:46linked below. Go grab it, run a quick
19:49test on your own, and let me know how it
19:51goes. As always, I'll see you guys in
19:54the next one.

About this transcript

This page contains the full transcript of I Turned Claude Into a $880/Day Motion Graphics Skill (No After Effects) by Sanji Nai-Chien, generated from the public captions YouTube serves with the video. The transcript has 3,343 words across 489 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 translate the transcript into 100+ languages.

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, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.