Imagine other people might like to know: if you use an API key to access OpenAI models in the Codex Desktop application, you can't access the faster inference.
I have not heard of FableCut, but looks like an interesting project!
It's similar in a sense that we're both exposing a mcp server to let claude runs editing tools, and also a traditional video editor interface.
It's different in terms of how you run it. FableCut requires you to install node, ffmpeg, and run a local server and connect to it via Chromium-based browser, whereas Palmier Pro is a native Mac App.
I don't know enough about the feature parity to comment on it. How is your experience with FableCut?
On the same machine where they are currently storing the sqlite file maybe? It's not that hard to run. A sqlite backup cron is also not that much more work than a pg_dump cron.
The post itself was technically useful, I found, and they posted their entire project to GitHub https://github.com/hershalb/music-video-arena which I think makes it worthy of a HN post and discussion since it is technical and at least IMHO a very interesting post.
We are still a bit of a ways away from being able to compile a script into a show. Also Veo 3.1 is old AF, it can't even do audio references! the cool kids are currently focused on Seedance 2.
This is the most recent thing I put together (sry it's on LinkedIn, haven't posted the series elsewhere yet)
I'm also unsure if screenplay to video is the desired... workflow we will eventually want.
I am biased, but am a bit kinder to AI video slop than most. Like I bet the first ever movies were pretty shit too. Give it a couple years while we figure out this new technology, and it's currently as bad as it ever will be, it can only get better.
I finally took the thirty seconds to find that famous video of the moon getting hit with a bullet, lmk if you are entertained to watch all the way through https://youtu.be/ZNAHcMMOHE8?si=GndI2PShs_vtGUT8
TBH as vapid as it is I think the overwhelming majority of people would call the AI video more entertaining. I mean good lord nothing happens in the first 3 minutes of Lumiere's movie except a bunch of academics waving their hands around.
At first usage I'm not impressed. I've probably spent a couple grand on Seedance 2 to date, and I can't find anything google omni flash does better than Seedance from running a handful of samples through the system. You can find some of the videos I've made in my HN bio link.
> allowing anyone to make whatever content they want.
One could draw a Mickey Mouse three-disc logo in inkscape, but nobody would sue inkscape because the application did not know what was being created. Likewise, asking for e.g. "a black disc, with two smaller discs tangent to it at a 105 degree angle" would not be infringing if the model never saw the Mickey Mouse logo to begin with.
I have exactly the same thought. Anyone who had used seedance 2.0 a bit can tell Gemini is a bit behind, and seedance 2.1 is on the horizontal already.
Higgsfield have multiple models available, people use Kling usually 2.5 & 3. There are a few good examples posted right now you'll notice the subtle differences.
I have tried to generate things myself and it's extremely hard to have more than 7-8 clips that are consistent, eventually you'll accept a compromise. I think it's why there isn't any long form content being done yet. Getting good results is sometimes just "chance" regardless of how many reference data you have.
When I was starting Traceway I was heavily inspired by skylightio from the Ruby ecosystem. I loved their SLOs/ranking perf issues, but I also wanted the features that Sentry offered in one place.
Meet Jerry- it's literally a guy named Jerry answering your questions.