
Oneleg
580 posts


Alright, here's where I'm at on V12 of creating seo videos with remotion and claude.
I have a set up on a VPS that I SSH into to interact with Claude
Then I have ran SEO reports, that search a keyword multiple times.
Up to 10 times, gathers all the pages in the results for that keyword.
Crawls all the pages, and looks over all the factors your competitors have to then figure out the delta, or difference, you page needs to rank higher than them
I have around 200 of these CSV or excel sheet reports to feed claude
Claude takes them and all of our current projects / work to vector embed them into a weaviate context index
It can then use this index, or I like to consider it a knowledge base / graph, to use as information to say in the video
It also helps claude with the story telling portion to get the information across.
Since the original goal was to have this heavily automated this idea keeps coming into play when I'm trying to create more and more videos
I'll ask it to templaitze and get it into a more mass production state, but every time I do this I think It goes combinitorics in complexity and ruins everything
So I have to go back to the orignial base to maintain qulaity.
Curious to hear your thoughts on AI generated videos via Remotion + Claude!
English

Descript, auto transcribes and is easy to edit.
It's a little different, in that you don't edit a time line you edit from the transcript.
It's a text based editor, so it can take a minute to learn.
But if you're just recording then hitting send, it's super easy.
You can login to your Youtube account and publish from the application.
I would still recommend going on youtube to add more stuff like time line or etc..
English

Anybody have a good alternative for @loom
I'm tired of having it glitch out and re-record videos often.
English

这不是电影截图
这是 AI 生成的
RX-78-2 元祖高达,完整视觉系统:
战场仰角主视觉 / 胸甲战损特写 / Earth Federation 徽章
正面全身站姿 / 侧角展台陈列 / 武器配件平铺 / 光剑战斗大片 / 暗角剪影
8 张图,从模型展示到电影级战场,全覆盖
最炸的是第一张 👆
仰视 + 战场废墟 + 火星四溅 + 体积光穿云
这种"初代主角登场"的史诗感
AI 稍微差一点就会变成玩具宣传图
踩到那个临界点的感觉,懂的人懂
高达 45 周年了
它值得被这样对待
完整 Prompt 框架评论区扣「RX-78」🤖
#Gundam #高达 #RX78 #AIGC #PromptEngineering

Vigo Zhao@VigoCreativeAI
奢侈品大片,AI 拍的 一套 Louis Vuitton Neverfull 完整视觉系统: 主视觉悬浮图 / Monogram 纹理特写 / LV logo 氛围 展台陈列正面 / 侧角低光 / 开箱平铺全家桶 / 极简仰角 7 张图,覆盖从种草到转化的完整视觉链路 LV 官方拍一套这样的图 要飞巴黎,租场地,请摄影师,做后期 时间以周计,费用以十万计 我花了两个小时 最难的是 Monogram 纹理那张 👆 皮革压纹 + 金属扣件反光 + 暗调氛围光 这种高奢质感,AI 稍微差一点就会变成"高仿货的感觉" 踩到正确关键词的那一刻,差距立刻出来了 做奢侈品电商、买手店、二手包平台的 这套视觉方案值得认真看一下 完整 Prompt 框架评论区扣「LV」👜 #LouisVuitton #奢侈品摄影 #AI产品图 #电商视觉 #PromptEngineering
中文

@TexasOncologist Fat Tony respects the fat tails, be more like Tony.
English

@rootforcegrind Neither am I lol, I want to create it for those starting their businesses so they have a foot in the door with seo
It's meant to be more actionable that viral.
I also like getting feedback from multiple persepctives cause then like a tree branch i can test everything.
English

IMHO, this is great. The animations are nice and your points are easily digested. Some things to consider...
1. Speed. It should continue to be readable, but I suspect you can get to you points faster. Besides, snappier animations are going to be a little more eye catching.
2. Stop the scroll. Add Initial title screen. Immediately show something like - "Top 3 reasons no one finds your laundromat on Google". Then continue to deliver details.
3. Call to Action. Might as well include the CTA ("Get your free gap audit") right out the gate rather than making someone wait. Always give them a means to take action.
4. Vid dimension. Is the video dimension optimal for the platform you're going to show this on?
Push back on any of it. I'm no expert man.
English

@OneLegchris Happy to. Heading out the door rn, but will check it out when i get back.
English

So for the past five days I've been going through test iterations of how to get Claude to use Remotion.
To then create infographics or static images that are dense with information that I fed from 200+ Excel reports on SEO.
I've also been brain dumping my own knowledge and scraping YouTube transcripts and vector embedding them.
I've been putting them into a Weaviate context index for Claude to then ping to get an understanding and information base to create these graphics.
(There's 7 in total, X will only allow me to upload 4)




English

@AlexLathery Honestly, it does diminish my performance.
More of a distraction, but I'll create the groups so that I can minimize them to come back later.
English

@floriandarroman I've been testing with this too,
About 4 days into this project. It's been a learning journey
English

Ok, Remotion is INSANE.
I installed it on my OpenClaw this morning.
I gave my logo, my landing page and the 14 steps I needed for that video.
After 5 iterations, I got this 👇
Well... almost.
I wasn't happy about the voiceover.
So I sent the script that Marc (my OpenClaw agent) created to ElevenLabs (free plan).
Then used Capcut to add the music + the voiceover from ElevenLabs.
And I got this video.
I am sure it's just a question of time until it becomes perfect (or a lack of skills on my side).
But this will go straight to my landing page and I'll save $500 on a video editor.
Remotion@Remotion
What's Remotion? A way to compose your video with code! • Use React, a powerful frontend technology, to build sophisticated videos with code. • Put together building blocks - like
English

@rootforcegrind I sent you my Grok conversation having it explain it to me.
In dm's
English

@OneLegchris Man, I'm going to have to look this up. MCP's, whatever that is, instead of tv tonight.
English

@pcshipp @Sathi_thiyagar Is it coded in a Javascript framework like react?
That might actually be your problem
English

@rootforcegrind This is hilarious, I've created you and now you've created all these problems for me.
Now go and fix all these issues I created
English





