Resize, Reframe, Cut and Caption videos from Horizontal to Vertical/Square for TikTok, Instagram, Snapchat, YouTubeShorts and more using AI Automation. Make high quality video GIFs. All in your browser.
No reviews yetBe the first to leave a review for Kamua
Your product is the best of the best! I've been obsessed with it all day! I can't stop exploring it, and thinking about it. This is going to be huge! I'd love to get involved somehow.
You must be very busy with your launch, but whenever you have the time, I'd love to chat. I have some ideas.
I'm a Project Manager who codes (with a marketing background).
@eek Nice that's an interesting pipeline! Going to copy it ;)
Jk lol. One question that I ask every ML startup is, where do you host your models? And how do you optimize for costs? I'm probably going to launch something soon and I'm concerned about GPU server costs and paying 200 USD/mo just for one server before I get any significant traction.
@cbsudu haha :D Yeah, costs are always problematic. The cheapest hosting service I knew that offered a GPU server was Hetzner (offering a 1080 with $100 / month in EU), unfortunately they no longer offer that (because of crypto and miners). I guess VPS with GPUs might be a bit too expensive for you at the moment (they start from $450 in the Cloud), although you could go and apply for some startup credits (Google Cloud / AWS / Azure / IBM / Oracle) they all have some programs for startups that offer one year
(or two) worth of credits, anywhere from $2.5k to $100k.
Other solutions would be things like AWS Sagemaker (https://aws.amazon.com/sagemaker/) that use elastic inference services, with which you can easily scale up with (it starts from $86 / month I think, and it depends on how much processing power and VRAM you need) - https://aws.amazon.com/machine-l... other cloud might have similar things as well, see what bests suits you :D
What kind of model are you trying to do? If it's something more generic like predictions or object detection / labeling / etc you might be better to start with a cloud API and build your MVP on that until you get traction.
@eek Hetzner looks interesting, unfortunate that they stopped it. Applying for credits is actually a good idea, completely forgot about that. (I might have some lying around too lol).
I'll have a look at SageMaker, Elastic GPUs would be great. Can save me so much money from having a fixed instance.
My project will involve some type of Segmentation (background removal), super resolution and maybe object detection if needed. Cloud APIs won't have these unfortunately since I'll be customizing and creating a pipeline.
Where do you guys host your models btw? :)
@juliusdesign Thank you (e posso dire grazie anche!)! Your title says it all! Can you share a little bit about the kinds of video challenges you encounter in your work?
@prcary hi! Yes, these are the most common video challenges:
1. Resize video vertically intelligently.
2. Crop the video to specific size
3. Add subtitles or keywords that are mentioned in the video creatively. Could the AI interpret the text semantically and offer you the option to highlight certain words in the video?
4. Add objects / stickers anchored to people.
Report
Easy-peasy video editing, what more could you ask for. Awesome job guys!
TinyVid
TinyVid
Community List
TinyVid
Community List
TinyVid
Community List
TinyVid
TinyVid
Social Creative Awards
TinyVid
Social Creative Awards
TinyVid
TinyVid
TinyVid