AI is making the screenwriter dream more reachable than ever.
You no longer need a Hollywood connection to get your story onto the page and in front of people.
@second_bassoon@DecimaVault It is. The thing I struggled the most with was understanding that every generation flow needs an output node at the end, and that every single thing I need to do needs a node for it, and that not all nodes receive multiple images but they need an image batch node ๐ฎโ๐จ
Welcome to Comfy for Developers.
This is where weโll share:
โ Use cases from the community
โ Product improvements for developers
โ Developer info for every launch
โ Posts from our engineering blog
โ Hackathons and dev events
โ Freebies and swag giveaways
For company news, follow @ComfyUI. Subscribe to our dev-focused blog!
The first 50 to quote tweet us and fill out the form below will get $50 in Comfy credits (you need a Comfy account to get this!)
sora is still the only video model that can generate indistinguishable real video
other ones feel kinda like gpt 4o, the first time you see an output you think "wow it's identical to human text" then 10 outputs in you're like "yeah this is mode-collapsed slop"
Don't start fighting agains your model thinking that your prompt is bad.
Sometimes the model is just dumb, and you need to use a higher model to not lose more tokens.
I needeSame prompt, 2 different results:
2. Keep the prompt simple. Tell it specifically what you need, not what you donโt want.
I know this could've been simpler in Photoshop, but I really wanted to learn what was happening.
I wasted hundreds of tokens trying to get the tennis player to stand to the right of the vertical white line, but I learned two valuable lessons:
1. If your current model canโt make a simple edit in one or two attempts, switch to a more capable model.